Umeå universitets logga

umu.sePublikationer
Ändra sökning
Avgränsa sökresultatet
1 - 49 av 49
RefereraExporteraLänk till träfflistan
Permanent länk
Referera
Referensformat
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Annat format
Fler format
Språk
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Annat språk
Fler språk
Utmatningsformat
  • html
  • text
  • asciidoc
  • rtf
Träffar per sida
  • 5
  • 10
  • 20
  • 50
  • 100
  • 250
Sortering
  • Standard (Relevans)
  • Författare A-Ö
  • Författare Ö-A
  • Titel A-Ö
  • Titel Ö-A
  • Publikationstyp A-Ö
  • Publikationstyp Ö-A
  • Äldst först
  • Nyast först
  • Skapad (Äldst först)
  • Skapad (Nyast först)
  • Senast uppdaterad (Äldst först)
  • Senast uppdaterad (Nyast först)
  • Disputationsdatum (tidigaste först)
  • Disputationsdatum (senaste först)
  • Standard (Relevans)
  • Författare A-Ö
  • Författare Ö-A
  • Titel A-Ö
  • Titel Ö-A
  • Publikationstyp A-Ö
  • Publikationstyp Ö-A
  • Äldst först
  • Nyast först
  • Skapad (Äldst först)
  • Skapad (Nyast först)
  • Senast uppdaterad (Äldst först)
  • Senast uppdaterad (Nyast först)
  • Disputationsdatum (tidigaste först)
  • Disputationsdatum (senaste först)
Markera
Maxantalet träffar du kan exportera från sökgränssnittet är 250. Vid större uttag använd dig av utsökningar.
  • 1.
    Brynolfsson, Patrik
    et al.
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Löfstedt, Tommy
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Asklund, Thomas
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Onkologi.
    Nyholm, Tufve
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Garpebring, Anders
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Gray-level invariant Haralick texture features2018Ingår i: Radiotherapy and Oncology, ISSN 0167-8140, E-ISSN 1879-0887, Vol. 127, s. S279-S280Artikel i tidskrift (Övrigt vetenskapligt)
  • 2.
    Bylund, Mikael
    et al.
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Jonsson, Joakim
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Lundman, Josef
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Brynolfsson, Patrik
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Garpebring, Anders
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Nyholm, Tufve
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Löfstedt, Tommy
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Using deep learning to generate synthetic CTs for radiotherapy treatment planning2018Ingår i: Radiotherapy and Oncology, ISSN 0167-8140, E-ISSN 1879-0887, Vol. 127, s. S283-S283Artikel i tidskrift (Övrigt vetenskapligt)
  • 3.
    Chandra, Abel
    et al.
    Umeå universitet, Teknisk-naturvetenskapliga fakulteten, Institutionen för datavetenskap.
    Tünnermann, Laura
    Umeå Plant Science Centre (UPSC), Department of Forest Genetics and Plant Physiology, Swedish University of Agricultural Sciences, Umeå, Sweden.
    Löfstedt, Tommy
    Umeå universitet, Teknisk-naturvetenskapliga fakulteten, Institutionen för datavetenskap.
    Gratz, Regina
    Umeå Plant Science Centre (UPSC), Department of Forest Genetics and Plant Physiology, Swedish University of Agricultural Sciences, Umeå, Sweden; Department of Forest Ecology and Management, Swedish University of Agricultural Sciences, Umeå, Sweden.
    Transformer-based deep learning for predicting protein properties in the life sciences2023Ingår i: eLIFE, E-ISSN 2050-084X, Vol. 12, artikel-id e82819Artikel i tidskrift (Refereegranskat)
    Abstract [en]

    Recent developments in deep learning, coupled with an increasing number of sequenced proteins, have led to a breakthrough in life science applications, in particular in protein property prediction. There is hope that deep learning can close the gap between the number of sequenced proteins and proteins with known properties based on lab experiments. Language models from the field of natural language processing have gained popularity for protein property predictions and have led to a new computational revolution in biology, where old prediction results are being improved regularly. Such models can learn useful multipurpose representations of proteins from large open repositories of protein sequences and can be used, for instance, to predict protein properties. The field of natural language processing is growing quickly because of developments in a class of models based on a particular model - the Transformer model. We review recent developments and the use of large-scale Transformer models in applications for predicting protein characteristics and how such models can be used to predict, for example, post-translational modifications. We review shortcomings of other deep learning models and explain how the Transformer models have quickly proven to be a very promising way to unravel information hidden in the sequences of amino acids.

    Ladda ner fulltext (pdf)
    fulltext
  • 4.
    de Pierrefeu, Amicie
    et al.
    NeuroSpin, CEA, Paris-Saclay, Gif-sur-Yvette, France.
    Fovet, Thomas
    Univ. Lille, CNRS UMR 9193, Laboratoire de Sciences Cognitives et Sciences Affectives (SCALab), PsyCHIC team, Lille F- 59000, France; CHU Lille, Pôle de Psychiatrie, Unité CURE, Lille F-59000, France.
    Hadj-Selem, Fouad
    Energy Transition Institute: VeDeCoM, France.
    Löfstedt, Tommy
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Ciuciu, Philippe
    NeuroSpin, CEA, Paris-Saclay, Gif-sur-Yvette, France; INRIA, CEA, Parietal team, Univ. Paris-Saclay, France.
    Lefebvre, Stephanie
    Univ. Lille, CNRS UMR 9193, Laboratoire de Sciences Cognitives et Sciences Affectives (SCALab), PsyCHIC team, Lille F-59000, France; CHU Lille, Pôle de Psychiatrie, Unité CURE, Lille F-59000, France.
    Thomas, Pierre
    Univ. Lille, CNRS UMR 9193, Laboratoire de Sciences Cognitives et Sciences Affectives (SCALab), PsyCHIC team, Lille F-59000, France; CHU Lille, Pôle de Psychiatrie, Unité CURE, Lille F-59000, France.
    Lopes, Renaud
    Imaging Dpt., Neuroradiology unit, CHU Lille, Lille F-59000, France; U1171 - Degenerative and Vascular Cognitive Disorders, Univ. Lille, INSERM, CHU Lille, Lille F-59000, France.
    Jardri, Renaud
    Univ. Lille, CNRS UMR 9193, Laboratoire de Sciences Cognitives et Sciences Affectives (SCALab), PsyCHIC team, Lille F-59000, France; CHU Lille, Pôle de Psychiatrie, Unité CURE, Lille F-59000, France.
    Duchesnay, Edouard
    NeuroSpin, CEA, Paris-Saclay, Gif-sur-Yvette, France.
    Prediction of activation patterns preceding hallucinations in patients with schizophrenia using machine learning with structured sparsity2018Ingår i: Human Brain Mapping, ISSN 1065-9471, E-ISSN 1097-0193, Vol. 39, nr 4, s. 1777-1788Artikel i tidskrift (Refereegranskat)
    Abstract [en]

    Despite significant progress in the field, the detection of fMRI signal changes during hallucinatory events remains difficult and time-consuming. This article first proposes a machine-learning algorithm to automatically identify resting-state fMRI periods that precede hallucinations versus periods that do not. When applied to whole-brain fMRI data, state-of-the-art classification methods, such as support vector machines (SVM), yield dense solutions that are difficult to interpret. We proposed to extend the existing sparse classification methods by taking the spatial structure of brain images into account with structured sparsity using the total variation penalty. Based on this approach, we obtained reliable classifying performances associated with interpretable predictive patterns, composed of two clearly identifiable clusters in speech-related brain regions. The variation in transition-to-hallucination functional patterns not only from one patient to another but also from one occurrence to the next (e.g., also depending on the sensory modalities involved) appeared to be the major difficulty when developing effective classifiers. Consequently, second, this article aimed to characterize the variability within the prehallucination patterns using an extension of principal component analysis with spatial constraints. The principal components (PCs) and the associated basis patterns shed light on the intrinsic structures of the variability present in the dataset. Such results are promising in the scope of innovative fMRI-guided therapy for drug-resistant hallucinations, such as fMRI-based neurofeedback.

  • 5.
    de Pierrefeu, Amicie
    et al.
    NeuroSpin, CEA, Université Paris-Saclay, 91191 Gif-sur-Yvette, France.
    Löfstedt, Tommy
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Hadj-Selem, Fouad
    Energy Transition Institute: VeDeCoM, 78 000 Versailles, France.
    Dubois, Mathieu
    NeuroSpin, CEA, Université Paris-Saclay, 91191 Gif-sur-Yvette, France.
    Jardri, Renaud
    Univ. Lille, CNRS UMR 9193, SCALab, CHU Lille, Pôle de Psychiatrie (unit CURE), 59000 Lille, France.
    Fovet, Thomas
    Univ. Lille, CNRS UMR 9193, SCALab, CHU Lille, Pôle de Psychiatrie (unit CURE), 59000 Lille, France.
    Ciuciu, Philippe
    NeuroSpin, CEA, Université Paris-Saclay, 91191 Gif-sur-Yvette, France.
    Frouin, Vincent
    NeuroSpin, CEA, Université Paris-Saclay, 91191 Gif-sur-Yvette, France.
    Duchesnay, Edouard
    NeuroSpin, CEA, Université Paris-Saclay, 91191 Gif-sur-Yvette, France.
    Structured Sparse Principal Components Analysis With the TV-Elastic Net Penalty2018Ingår i: IEEE Transactions on Medical Imaging, ISSN 0278-0062, E-ISSN 1558-254X, Vol. 37, nr 2, s. 396-407Artikel i tidskrift (Refereegranskat)
    Abstract [en]

    Principal component analysis (PCA) is an exploratory tool widely used in data analysis to uncover the dominant patterns of variability within a population. Despite its ability to represent a data set in a low-dimensional space, PCA’s interpretability remains limited. Indeed, the components produced by PCA are often noisy or exhibit no visually meaningful patterns. Furthermore, the fact that the components are usually non-sparse may also impede interpretation, unless arbitrary thresholding is applied. However, in neuroimaging, it is essential to uncover clinically interpretable phenotypic markers that would account for the main variability in the brain images of a population. Recently, some alternatives to the standard PCA approach, such as sparse PCA (SPCA), have been proposed, their aim being to limit the density of the components. Nonetheless, sparsity alone does not entirely solve the interpretability problem in neuroimaging, since it may yield scattered andunstable components. We hypothesized that the incorporation of prior information regarding the structure of the data may lead to improved relevance and interpretability of brain patterns. We therefore present a simple extension of the popular PCA framework that adds structured sparsity penalties on the loading vectors in order to identify the few stable regions in the brain images that capture most of the variability. Such structured sparsity can be obtained by combining, e.g., l1 and total variation (TV) penalties, where the TV regularization encodes information on the underlying structure of the data. This paper presents the structured SPCA (denoted SPCA-TV) optimization framework and its resolution. We demonstrate SPCA-TV’s effectiveness and versatility on three different data sets. It can be applied to any kind of structured data, such as, e.g., N-dimensional array images or meshes of cortical surfaces. The gains of SPCA-TV over unstructured approaches (such as SPCA and ElasticNet PCA) or structured approach (such as GraphNet PCA) are significant, since SPCA-TV reveals the variability within a data set in the form of intelligible brain patterns that are easier to interpret and more stable across different samples.

  • 6.
    de Pierrefeu, Amicie
    et al.
    NeuroSpin, CEA, Gif-sur-Yvette, France.
    Löfstedt, Tommy
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Laidi, C.
    NeuroSpin, CEA, Gif-sur-Yvette, France; Institut National de la Santé et de la Recherche Médicale (INSERM), U955, Institut Mondor de Recherche Biomédicale, Psychiatrie Translationnelle, Créteil, France; Fondation Fondamental, Créteil, France; Pôle de Psychiatrie, Assistance Publique–Hôpitaux de Paris (AP-HP), Faculté, de Médecine de Créteil, DHU PePsy, Hôpitaux Universitaires Mondor, Créteil, France.
    Hadj-Selem, Fouad
    Energy Transition Institute: VeDeCoM, Versailles, France.
    Bourgin, Julie
    Department of Psychiatry, Louis-Mourier Hospital, AP-HP, Colombes, France; INSERM U894, Centre for Psychiatry and Neurosciences, Paris, France.
    Hajek, Tomas
    Department of Psychiatry, Dalhousie University, Halifax, NS, Canada; National Institute of Mental Health, Klecany, Czech Republic.
    Spaniel, Filip
    National Institute of Mental Health, Klecany, Czech Republic.
    Kolenic, Marian
    National Institute of Mental Health, Klecany, Czech Republic.
    Ciuciu, Philippe
    NeuroSpin, CEA, Gif-sur-Yvette, France; INRIA, CEA, Parietal team, University of Paris-Saclay, France.
    Hamdani, Nora
    Institut National de la Santé et de la Recherche Médicale (INSERM), U955, Institut Mondor de Recherche Biomédicale, Psychiatrie Translationnelle, Créteil, France; Fondation Fondamental, Créteil, France; Pôle de Psychiatrie, Assistance Publique–Hôpitaux de Paris (AP-HP), Faculté, de Médecine de Créteil, DHU PePsy, Hôpitaux Universitaires Mondor, Créteil, France.
    Leboyer, Marion
    Institut National de la Santé et de la Recherche Médicale (INSERM), U955, Institut Mondor de Recherche Biomédicale, Psychiatrie Translationnelle, Créteil, France; Fondation Fondamental, Créteil, France; Pôle de Psychiatrie, Assistance Publique–Hôpitaux de Paris (AP-HP), Faculté, de Médecine de Créteil, DHU PePsy, Hôpitaux Universitaires Mondor, Créteil, France.
    Fovet, Thomas
    Laboratoire de Sciences Cognitives et Sciences Affectives (SCALab-PsyCHIC), CNRS UMR 9193, University of Lille; Pôle de Psychiatrie, Unité CURE, CHU Lille, Lille, France.
    Jardri, Renaud
    INRIA, CEA, Parietal team, University of Paris-Saclay, France; Laboratoire de Sciences Cognitives et Sciences Affectives (SCALab-PsyCHIC), CNRS UMR 9193, University of Lille; Pôle de Psychiatrie, Unité CURE, CHU Lille, Lille, France.
    Houenou, Josselin
    NeuroSpin, CEA, Gif-sur-Yvette, France; Institut National de la Santé et de la Recherche Médicale (INSERM), U955, Institut Mondor de Recherche Biomédicale, Psychiatrie Translationnelle, Créteil, France; Fondation Fondamental, Créteil, France; Pôle de Psychiatrie, Assistance Publique–Hôpitaux de Paris (AP-HP), Faculté, de Médecine de Créteil, DHU PePsy, Hôpitaux Universitaires Mondor, Créteil, France.
    Duchesnay, Edouard
    NeuroSpin, CEA, Gif-sur-Yvette, France.
    Identifying a neuroanatomical signature of schizophrenia, reproducible across sites and stages, using machine-learning with structured sparsity2018Ingår i: Acta Psychiatrica Scandinavica, ISSN 0001-690X, E-ISSN 1600-0447, Vol. 138, s. 571-580Artikel i tidskrift (Refereegranskat)
    Abstract [en]

    ObjectiveStructural MRI (sMRI) increasingly offers insight into abnormalities inherent to schizophrenia. Previous machine learning applications suggest that individual classification is feasible and reliable and, however, is focused on the predictive performance of the clinical status in cross‐sectional designs, which has limited biological perspectives. Moreover, most studies depend on relatively small cohorts or single recruiting site. Finally, no study controlled for disease stage or medication's effect. These elements cast doubt on previous findings’ reproducibility.

    MethodWe propose a machine learning algorithm that provides an interpretable brain signature. Using large datasets collected from 4 sites (276 schizophrenia patients, 330 controls), we assessed cross‐site prediction reproducibility and associated predictive signature. For the first time, we evaluated the predictive signature regarding medication and illness duration using an independent dataset of first‐episode patients.

    ResultsMachine learning classifiers based on neuroanatomical features yield significant intersite prediction accuracies (72%) together with an excellent predictive signature stability. This signature provides a neural score significantly correlated with symptom severity and the extent of cognitive impairments. Moreover, this signature demonstrates its efficiency on first‐episode psychosis patients (73% accuracy).

    ConclusionThese results highlight the existence of a common neuroanatomical signature for schizophrenia, shared by a majority of patients even from an early stage of the disorder.

  • 7.
    De Pierrefeu, Amicie
    et al.
    NeuroSpin, CEA, Paris-Saclay, France.
    Löfstedt, Tommy
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper.
    Laidi, Charles
    NeuroSpin, CEA, Paris-Saclay, France; Institut National de la Santé et de la Recherche Médicale (INSERM), Institut Mondor de Recherche Biomédicale, Psychiatrie Translationnelle, Créteil, France; Fondation Fondamental, Cŕeteil, France; Pole de Psychiatrie, Assistance Publique Hopitaux de Paris (AP-HP), Faculté de Médecine de Cŕeteil, Cŕeteil, France.
    Hadj-Selem, Fouad
    Energy Transition Institute: VeDeCoM, France.
    Leboyer, Marion
    Institut National de la Santé et de la Recherche Médicale (INSERM), Institut Mondor de Recherche Biomédicale, Psychiatrie Translationnelle, Créteil, France; Fondation Fondamental, Cŕeteil, France; Pole de Psychiatrie, Assistance Publique Hopitaux de Paris (AP-HP), Faculté de Médecine de Cŕeteil, Cŕeteil, France.
    Ciuciu, Philippe
    NeuroSpin, CEA, Paris-Saclay, France.
    Houenou, Josselin
    NeuroSpin, CEA, Paris-Saclay, France; Institut National de la Santé et de la Recherche Médicale (INSERM), Institut Mondor de Recherche Biomédicale, Psychiatrie Translationnelle, Créteil, France; Fondation Fondamental, Cŕeteil, France; Pole de Psychiatrie, Assistance Publique Hopitaux de Paris (AP-HP), Faculté de Médecine de Cŕeteil, Cŕeteil, France.
    Duchesnay, Edouard
    NeuroSpin, CEA, Paris-Saclay, France.
    Interpretable and stable prediction of schizophrenia on a large multisite dataset using machine learning with structured sparsity2018Ingår i: 2018 International Workshop on Pattern Recognition in Neuroimaging (PRNI), IEEE, 2018, artikel-id 8423946Konferensbidrag (Refereegranskat)
    Abstract [en]

    The use of machine-learning (ML) in neuroimaging offers new perspectives in early diagnosis and prognosis of brain diseases. Indeed, ML algorithms can jointly examine all brain features to capture complex relationships in the data in order to make inferences at a single-subject level. To deal with such high dimensional input and the associated risk of overfitting on the training data, a proper regularization (or feature selection) is required. Standard ℓ2-regularized predictors, such as Support Vector Machine, provide dense patterns of predictors. However, in the context of predictive disease signature discovery, it is now essential to understand the brain pattern that underpins the prediction. Despite ℓ1-regularized (sparse) has often been advocated as leading to more interpretable models, they generally lead to scattered and unstable patterns. We hypothesize that the integration of prior knowledge regarding the structure of the input images should improve the relevance and the stability of the predictive signature. Such structured sparsity can be obtained by combining together ℓ1 (possibly ℓ2) and Total variation (TV) penalties. We demonstrated the relevance of using ML with structured sparsity on a large multisite dataset of schizophrenia patients and controls. Using 3D maps of grey matter density, we obtained promising inter-site prediction performances. More importantly, we have uncovered a predictive signature of schizophrenia that is clinically interpretable and stable across resampling. This suggests that structured sparsity provides a major breakthrough over 'off-The-shelf' algorithms to perform a robust selection of important brain regions in the context of biomarkers discovery.

  • 8.
    Dubois, Mathieu
    et al.
    NeuroSpin, I2BM, CEA.
    Hadj-Selem, Fouad
    NeuroSpin, I2BM, CEA.
    Löfstedt, Tommy
    NeuroSpin, I2BM, CEA.
    Perrot, Matthieu
    Centre d'Acquisition et de Traitement des Images (CATI).
    Fischer, Clara
    Centre d'Acquisition et de Traitement des Images (CATI).
    Frouin, Vincent
    NeuroSpin, I2BM, CEA.
    Duchesnay, Edouard
    NeuroSpin, I2BM, CEA.
    Predictive support recovery with TV-Elastic Net penalty and logistic regression: An application to structural MRI2014Konferensbidrag (Refereegranskat)
    Abstract [en]

    The use of machine-learning in neuroimaging offers new perspectives in early diagnosis and prognosis of brain diseases. Although such multivariate methods can capture complex relationships in the data, traditional approaches provide irregular (l12 penalty) or scattered (l1 penalty) predictive pattern with a very limited relevance. A penalty like Total Variation (TV) that exploits the natural 3D structure of the images can increase the spatial coherence of the weight map. However, TV penalization leads to non-smooth optimization problems that are hard to minimize. We propose an optimization framework that minimizes any combination of l1, l2, and TV penalties while preserving the exact l1 penalty. This algorithm uses Nesterov's smoothing technique to approximate the TV penalty with a smooth function such that the loss and the penalties are minimized with an exact accelerated proximal gradient algorithm. We propose an original continuation algorithm that uses successively smaller values of the smoothing parameter to reach a prescribed precision while achieving the best possible convergence rate. This algorithm can be used with other losses or penalties. The algorithm is applied on a classification problem on the ADNI dataset. We observe that the TV penalty does not necessarily improve the prediction but provides a major breakthrough in terms of support recovery of the predictive brain regions.

  • 9.
    Fetty, Lukas
    et al.
    Department of Radiation Oncology, Medical University of Vienna, Vienna, Austria.
    Bylund, Mikael
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Kuess, Peter
    Department of Radiation Oncology, Medical University of Vienna, Vienna, Austria.
    Heilemann, Gerd
    Department of Radiation Oncology, Medical University of Vienna, Vienna, Austria.
    Nyholm, Tufve
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Georg, Dietmar
    Department of Radiation Oncology, Medical University of Vienna, Vienna, Austria.
    Löfstedt, Tommy
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Latent space manipulation for high-resolution medical image synthesis via the StyleGAN2020Ingår i: Zeitschrift für Medizinische Physik, ISSN 0939-3889, E-ISSN 1876-4436, Vol. 30, nr 4, s. 305-314Artikel i tidskrift (Refereegranskat)
    Abstract [en]

    Introduction: This paper explores the potential of the StyleGAN model as an high-resolution image generator for synthetic medical images. The possibility to generate sample patient images of different modalities can be helpful for training deep learning algorithms as e.g. a data augmentation technique.

    Methods: The StyleGAN model was trained on Computed Tomography (CT) and T2- weighted Magnetic Resonance (MR) images from 100 patients with pelvic malignancies. The resulting model was investigated with regards to three features: Image Modality, Sex, and Longitudinal Slice Position. Further, the style transfer feature of the StyleGAN was used to move images between the modalities. The root-mean-squard error (RMSE) and the Mean Absolute Error (MAE) were used to quantify errors for MR and CT, respectively.

    Results: We demonstrate how these features can be transformed by manipulating the latent style vectors, and attempt to quantify how the errors change as we move through the latent style space. The best results were achieved by using the style transfer feature of the StyleGAN (58.7 HU MAE for MR to CT and 0.339 RMSE for CT to MR). Slices below and above an initial central slice can be predicted with an error below 75 HU MAE and 0.3 RMSE within 4 cm for CT and MR, respectively.

    Discussion: The StyleGAN is a promising model to use for generating synthetic medical images for MR and CT modalities as well as for 3D volumes.

    Ladda ner fulltext (pdf)
    fulltext
  • 10. Fetty, Lukas
    et al.
    Löfstedt, Tommy
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Heilemann, Gerd
    Furtado, Hugo
    Nesvacil, Nicole
    Nyholm, Tufve
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Georg, Dietmar
    Kuess, Peter
    Investigating conditional GAN performance with different generator architectures, an ensemble model, and different MR scanners for MR-sCT conversion2020Ingår i: Physics in Medicine and Biology, ISSN 0031-9155, E-ISSN 1361-6560, Vol. 65, nr 10, artikel-id 105004Artikel i tidskrift (Refereegranskat)
    Abstract [en]

    Recent developments in magnetic resonance (MR) to synthetic computed tomography (sCT) conversion have shown that treatment planning is possible without an initial planning CT. Promising conversion results have been demonstrated recently using conditional generative adversarial networks (cGANs). However, the performance is generally only tested on images from one MR scanner, which neglects the potential of neural networks to find general high-level abstract features. In this study, we explored the generalizability of the generator models, trained on a single field strength scanner, to data acquired with higher field strengths. T2-weighted 0.35T MRIs and CTs from 51 patients treated for prostate (40) and cervical cancer (11) were included. 25 of them were used to train four different generators (SE-ResNet, DenseNet, U-Net, and Embedded Net). Further, an ensemble model was created from the four network outputs. The models were validated on 16 patients from a 0.35T MR scanner. Further, the trained models were tested on the Gold Atlas dataset, containing T2-weighted MR scans of different field strengths; 1.5T(7) and 3T(12), and 10 patients from the 0.35T scanner. The sCTs were dosimetrically compared using clinical VMAT plans for all test patients. For the same scanner (0.35T), the results from the different models were comparable on the test set, with only minor differences in the mean absolute error (MAE) (35-51HU body). Similar results were obtained for conversions of 3T GE Signa and the 3T GE Discovery images (40-62HU MAE) for three of the models. However, larger differences were observed for the 1.5T images (48-65HU MAE). The overall best model was found to be the ensemble model. All dose differences were below 1%. This study shows that it is possible to generalize models trained on images of one scanner to other scanners and different field strengths. The best metric results were achieved by the combination of all networks.

    Ladda ner fulltext (pdf)
    fulltext
  • 11.
    Garpebring, Anders
    et al.
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Brynolfsson, Patrik
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Kuess, Peter
    Department of Radiotherapy, Medical University of Vienna, Vienna, Austria; Christian Doppler Laboratory for Medical Radiation Research for Radiation Oncology, Vienna, Austria.
    Georg, Dietmar
    Department of Radiotherapy, Medical University of Vienna, Vienna, Austria; Christian Doppler Laboratory for Medical Radiation Research for Radiation Oncology, Vienna, Austria.
    Helbich, Thomas H.
    Department of Biomedical Imaging and Image-Guided Therapy, Medical University of Vienna, Vienna, Austria; Christian Doppler Laboratory for Medical Radiation Research for Radiation Oncology, Vienna, Austria.
    Nyholm, Tufve
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Löfstedt, Tommy
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Density Estimation of Grey-Level Co-Occurrence Matrices for Image Texture Analysis2018Ingår i: Physics in Medicine and Biology, ISSN 0031-9155, E-ISSN 1361-6560, Vol. 63, nr 19, s. 9-15, artikel-id 195017Artikel i tidskrift (Refereegranskat)
    Abstract [en]

    The Haralick texture features are common in the image analysis literature, partly because of their simplicity and because their values can be interpreted. It was recently observed that the Haralick texture features are very sensitive to the size of the GLCM that was used to compute them, which led to a new formulation that is invariant to the GLCM size. However, these new features still depend on the sample size used to compute the GLCM, i.e. the size of the input image region-of-interest (ROI).

    The purpose of this work was to investigate the performance of density estimation methods for approximating the GLCM and subsequently the corresponding invariant features.

    Three density estimation methods were evaluated, namely a piece-wise constant distribution, the Parzen-windows method, and the Gaussian mixture model. The methods were evaluated on 29 different image textures and 20 invariant Haralick texture features as well as a wide range of different ROI sizes.

    The results indicate that there are two types of features: those that have a clear minimum error for a particular GLCM size for each ROI size, and those whose error decreases monotonically with increased GLCM size. For the first type of features, the Gaussian mixture model gave the smallest errors, and in particular for small ROI sizes (less than about 20×20).

    In conclusion, the Gaussian mixture model is the preferred method for the first type of features (in particular for small ROIs). For the second type of features, simply using a large GLCM size is preferred.

    Ladda ner fulltext (pdf)
    fulltext
  • 12.
    Garpebring, Anders
    et al.
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Tommy, Löfstedt
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Parameter estimation using weighted total least squares in the two-compartment exchange model2018Ingår i: Magnetic Resonance in Medicine, ISSN 0740-3194, E-ISSN 1522-2594, Vol. 79, nr 1, s. 561-567Artikel i tidskrift (Refereegranskat)
    Abstract [en]

    Purpose

    The linear least squares (LLS) estimator provides a fast approach to parameter estimation in the linearized two-compartment exchange model. However, the LLS method may introduce a bias through correlated noise in the system matrix of the model. The purpose of this work is to present a new estimator for the linearized two-compartment exchange model that takes this noise into account.

    Method

    To account for the noise in the system matrix, we developed an estimator based on the weighted total least squares (WTLS) method. Using simulations, the proposed WTLS estimator was compared, in terms of accuracy and precision, to an LLS estimator and a nonlinear least squares (NLLS) estimator.

    Results

    The WTLS method improved the accuracy compared to the LLS method to levels comparable to the NLLS method. This improvement was at the expense of increased computational time; however, the WTLS was still faster than the NLLS method. At high signal-to-noise ratio all methods provided similar precisions while inconclusive results were observed at low signal-to-noise ratio.

    Conclusion

    The proposed method provides improvements in accuracy compared to the LLS method, however, at an increased computational cost. Magn Reson Med, 2017. © 2017 International Society for Magnetic Resonance in Medicine.

  • 13. Guigui, N.
    et al.
    Philippe, C.
    Gloaguen, A.
    Karkar, S.
    Guillemot, V.
    Löfstedt, Tommy
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Frouin, V.
    Network Regularization in Imaging Genetics Improves Prediction Performances and Model Interpretability on Alzheimer’s Disease2019Ingår i: 2019 IEEE 16th International Symposium on Biomedical Imaging (ISBI 2019), IEEE, 2019, s. 1403-1406Konferensbidrag (Refereegranskat)
    Abstract [en]

    Imaging genetics is a growing popular research avenue which aims to find genetic variants associated with quantitative phenotypes that characterize a disease. In this work, we combine structural MRI with genetic data structured by prior knowledge of interactions in a Canonical Correlation Analysis (CCA) model with graph regularization. This results in improved prediction performance and yields a more interpretable model.

  • 14.
    Guillemot, Vincent
    et al.
    Bioinformatics and Biostatistics Hub, Institut Pasteur, Paris, France.
    Beaton, Derek
    The Rotman Research Institute, Institution at Baycrest, Toronto, Canada.
    Gloaguen, Arnaud
    L2S, UMR CNRS 8506, CNRS–Centrale Supélec–Université Paris-Sud, Université Paris-Saclay, 3 rue Joliot-Curie, 91192 Gif-sur-Yvette, France.
    Löfstedt, Tommy
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Levine, Brian
    The Rotman Research Institute, Institution at Baycrest, Toronto, Canada.
    Raymond, Nicolas
    IRMAR, UMR 6625, Université de Rennes, Rennes, France.
    Tenenhaus, Arthur
    L2S, UMR CNRS 8506, CNRS–Centrale Supélec–Université Paris-Sud, Université Paris-Saclay, 3 rue Joliot-Curie, Gif-sur-Yvette, France.
    Abdi, Hervé
    School of Behavioral and Brain Sciences, The University of Texas at Dallas, Richardson, TX, United States of America.
    A constrained singular value decomposition method that integrates sparsity and orthogonality2019Ingår i: PLOS ONE, E-ISSN 1932-6203, Vol. 14, nr 3, artikel-id e0211463Artikel i tidskrift (Refereegranskat)
    Abstract [en]

    We propose a new sparsification method for the singular value decomposition—called the constrained singular value decomposition (CSVD)—that can incorporate multiple constraints such as sparsification and orthogonality for the left and right singular vectors. The CSVD can combine different constraints because it implements each constraint as a projection onto a convex set, and because it integrates these constraints as projections onto the intersection of multiple convex sets. We show that, with appropriate sparsification constants, the algorithm is guaranteed to converge to a stable point. We also propose and analyze the convergence of an efficient algorithm for the specific case of the projection onto the balls defined by the norms L1 and L2. We illustrate the CSVD and compare it to the standard singular value decomposition and to a non-orthogonal related sparsification method with: 1) a simulated example, 2) a small set of face images (corresponding to a configuration with a number of variables much larger than the number of observations), and 3) a psychometric application with a large number of observations and a small number of variables. The companion R-package, csvd, that implements the algorithms described in this paper, along with reproducible examples, are available for download from https://github.com/vguillemot/csvd.

    Ladda ner fulltext (pdf)
    fulltext
  • 15.
    Hadj-Selem, Fouad
    et al.
    Energy Transition Institute VeDeCoM, Versailles, France.
    Löfstedt, Tommy
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Dohmatob, Elvis
    PARIETAL Team, INRIA/CEA, Université Paris-Saclay, Gif-sur-Yvette, France.
    Frouin, Vincent
    NeuroSpin, CEA, Université Paris-Saclay, Gif-sur-Yvette, France.
    Dubois, Mathieu
    NeuroSpin, CEA, Université Paris-Saclay, Gif-sur-Yvette, France.
    Guillemot, Vincent
    NeuroSpin, CEA, Université Paris-Saclay, Gif-sur-Yvette, France.
    Duchesnay, Edouard
    NeuroSpin, CEA, Université Paris-Saclay, Gif-sur-Yvette, France.
    Continuation of Nesterov's Smoothing for Regression With Structured Sparsity in High-Dimensional Neuroimaging2018Ingår i: IEEE Transactions on Medical Imaging, ISSN 0278-0062, E-ISSN 1558-254X, Vol. 37, nr 11, s. 2403-2413Artikel i tidskrift (Refereegranskat)
    Abstract [en]

    Predictive models can be used on high-dimensional brain images to decode cognitive states or diagnosis/prognosis of a clinical condition/evolution. Spatial regularization through structured sparsity offers new perspectives in this context and reduces the risk of overfitting the model while providing interpretable neuroimaging signatures by forcing the solution to adhere to domain-specific constraints. Total variation (TV) is a promising candidate for structured penalization: it enforces spatial smoothness of the solution while segmenting predictive regions from the background. We consider the problem of minimizing the sum of a smooth convex loss, a non-smooth convex penalty (whose proximal operator is known) and a wide range of possible complex, non-smooth convex structured penalties such as TV or overlapping group Lasso. Existing solvers are either limited in the functions they can minimize or in their practical capacity to scale to high-dimensional imaging data. Nesterov’s smoothing technique can be used to minimize a large number of non-smooth convex structured penalties. However, reasonable precision requires a small smoothing parameter, which slows down the convergence speed to unacceptable levels. To benefit from the versatility of Nesterov’s smoothing technique, we propose a first order continuation algorithm, CONESTA, which automatically generates a sequence of decreasing smoothing parameters. The generated sequence maintains the optimal convergence speed toward any globally desired precision. Our main contributions are: gap to probe the current distance to the global optimum in order to adapt the smoothing parameter and the To propose an expression of the duality convergence speed. This expression is applicable to many penalties and can be used with other solvers than CONESTA. We also propose an expression for the particular smoothing parameter that minimizes the number of iterations required to reach a given precision. Furthermore, we provide a convergence proof and its rate, which is an improvement over classical proximal gradient smoothing methods. We demonstrate on both simulated and high-dimensional structural neuroimaging data that CONESTA significantly outperforms many state-of-the-art solvers in regard to convergence speed and precision.

  • 16.
    Hellström, Max
    et al.
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper.
    Löfstedt, Tommy
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper. Umeå universitet, Teknisk-naturvetenskapliga fakulteten, Institutionen för datavetenskap.
    Garpebring, Anders
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper.
    Denoising and uncertainty estimation in parameter mapping with approximate Bayesian deep image priors2023Ingår i: Magnetic Resonance in Medicine, ISSN 0740-3194, E-ISSN 1522-2594, Vol. 90, nr 6, s. 2557-2571Artikel i tidskrift (Refereegranskat)
    Abstract [en]

    Purpose: To mitigate the problem of noisy parameter maps with high uncertainties by casting parameter mapping as a denoising task based on Deep Image Priors.

    Methods: We extend the concept of denoising with Deep Image Prior (DIP) into parameter mapping by treating the output of an image-generating network as a parametrization of tissue parameter maps. The method implicitly denoises the parameter mapping process by filtering low-level image features with an untrained convolutional neural network (CNN). Our implementation includes uncertainty estimation from Bernoulli approximate variational inference, implemented with MC dropout, which provides model uncertainty in each voxel of the denoised parameter maps. The method is modular, so the specifics of different applications (e.g., T1 mapping) separate into application-specific signal equation blocks. We evaluate the method on variable flip angle T1 mapping, multi-echo T2 mapping, and apparent diffusion coefficient mapping.

    Results: We found that deep image prior adapts successfully to several applications in parameter mapping. In all evaluations, the method produces noise-reduced parameter maps with decreased uncertainty compared to conventional methods. The downsides of the proposed method are the long computational time and the introduction of some bias from the denoising prior.

    Conclusion: DIP successfully denoise the parameter mapping process and applies to several applications with limited hyperparameter tuning. Further, it is easy to implement since DIP methods do not use network training data. Although time-consuming, uncertainty information from MC dropout makes the method more robust and provides useful information when properly calibrated.

    Ladda ner fulltext (pdf)
    fulltext
  • 17.
    Hägerlind, E.
    et al.
    Department of Medical and Health Sciences, Division of Community Medicine, Primary Care, Linköping University, Linköping, Sweden.
    Falk, M.
    Department of Medical and Health Sciences, Division of Community Medicine, Primary Care, Linköping University, Linköping, Sweden.
    Löfstedt, Tommy
    Computational Solutions, Umeå, Sweden.
    Lindholm-Sethson, Britta
    Umeå universitet, Teknisk-naturvetenskapliga fakulteten, Kemiska institutionen.
    Bodén, Ida
    Ivida AB, Umeå, Sweden.
    Near infrared and skin impedance spectroscopy: a possible support in the diagnostic process of skin tumours in primary health care2015Ingår i: Skin research and technology, ISSN 0909-752X, E-ISSN 1600-0846, Vol. 21, nr 4, s. 493-499Artikel i tidskrift (Refereegranskat)
    Abstract [en]

    Background/purpose: The global incidence of skin cancer has increased drastically in recent decades, especially in Australia and Northern Europe. Early detection is crucial for good prognosis and high survival rates. In general, primary care physicians have considerably lower sensitivity and specificity rates for detection of skin cancer, compared to dermatologists. A probable main reason for this is that current diagnostic tools are subjective in nature, and therefore diagnostic skills highly depend on experience. Illustratively, in Sweden, approximately 155500 benign skin lesions are excised unnecessarily every year. An objective instrument, added to the clinical examination, might improve the diagnostic accuracy, and thus promote earlier detection of malignant skin tumours, as well as reduce medical costs associated with unnecessary biopsies and excisions. The general aim of this study was to investigate the usefulness of the combination of near infrared (NIR) and skin impedance spectroscopy as a supportive tool in the diagnosis and evaluation of skin tumours in primary health care.

    Methods: Near infrared and skin impedance data were collected by performing measurements on suspect malignant, premalignant and benign tumours in the skin of patients seeking primary health care for skin tumour evaluation. The obtained data were analysed using multivariate analysis and compared with the diagnosis received by the conventional diagnostic process.

    Results: The observed sensitivity and specificity rates were both 100%, when discriminating malignant and premalignant skin tumours from benign skin tumours, and the observed sensitivity and specificity for separating malignant skin tumours from premalignant and benign skin tumours were also 100%, respectively.

    Conclusion: The results of this study indicate that the NIR and skin impedance spectroscopy may be a useful supportive tool for the general practitioner in the diagnosis and evaluation of skin tumours in primary health care, as a complement to the visual assessment.

  • 18.
    Kia, Seyed Mostafa
    et al.
    Radboud University Medical Center, Nijmegen, Netherlands.
    Marquand, Andre
    Radboud University, Nijmegen, Netherlands.
    Duchesnay, Edouard
    NeuroSpin, CEA, Saclay, Paris, France.
    Löfstedt, Tommy
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper.
    MLCN 2018 preface2018Ingår i: Understanding and interpreting machine learning in medical image computing applications: First International Workshops MLCN 2018, DLF 2018, and iMIMIC 2018, Held in Conjunction with MICCAI 2018, Granada, Spain, September 16-20, 2018, Proceedings / [ed] Danail Stoyanov, Zeike Taylor, Seyed Mostafa Kia, Ipek Oguz, Mauricio Reyes et al., Springer, 2018Konferensbidrag (Övrigt vetenskapligt)
  • 19.
    Löfstedt, Tommy
    Umeå universitet, Teknisk-naturvetenskapliga fakulteten, Kemiska institutionen.
    OnPLS: Orthogonal projections to latent structures in multiblock and path model data analysis2012Doktorsavhandling, sammanläggning (Övrigt vetenskapligt)
    Abstract [en]

    The amounts of data collected from each sample of e.g. chemical or biological materials have increased by orders of magnitude since the beginning of the 20th century. Furthermore, the number of ways to collect data from observations is also increasing. Such configurations with several massive data sets increase the demands on the methods used to analyse them. Methods that handle such data are called multiblock methods and they are the topic of this thesis.

    Data collected from advanced analytical instruments often contain variation from diverse mutually independent sources, which may confound observed patterns and hinder interpretation of latent variable models. For this reason, new methods have been developed that decompose the data matrices, placing variation from different sources of variation into separate parts. Such procedures are no longer merely pre-processing filters, as they initially were, but have become integral elements of model building and interpretation. One strain of such methods, called OPLS, has been particularly successful since it is easy to use, understand and interpret.

    This thesis describes the development of a new multiblock data analysis method called OnPLS, which extends the OPLS framework to the analysis of multiblock and path models with very general relationships between blocks in both rows and columns. OnPLS utilises OPLS to decompose sets of matrices, dividing each matrix into a globally joint part (a part shared with all the matrices it is connected to), several locally joint parts (parts shared with some, but not all, of the connected matrices) and a unique part that no other matrix shares.

    The OnPLS method was applied to several synthetic data sets and data sets of “real” measurements. For the synthetic data sets, where the results could be compared to known, true parameters, the method generated global multiblock (and path) models that were more similar to the true underlying structures compared to models without such decompositions. I.e. the globally joint, locally joint and unique models more closely resembled the corresponding true data. When applied to the real data sets, the OnPLS models revealed chemically or biologically relevant information in all kinds of variation, effectively increasing the interpretability since different kinds of variation are distinguished and separately analysed.

    OnPLS thus improves the quality of the models and facilitates better understanding of the data since it separates and separately analyses different kinds of variation. Each kind of variation is purer and less tainted by other kinds. OnPLS is therefore highly recommended to anyone engaged in multiblock or path model data analysis.

    Ladda ner fulltext (pdf)
    OnPLS
  • 20.
    Löfstedt, Tommy
    et al.
    Umeå universitet, Teknisk-naturvetenskapliga fakulteten, Kemiska institutionen.
    Ahnlund, Olof
    Peolsson, Michael
    Trygg, Johan
    Umeå universitet, Teknisk-naturvetenskapliga fakulteten, Kemiska institutionen.
    Dynamic ultrasound imaging: a multivariate approach for the analysis and comparison of time-dependent musculoskeletal movements2012Ingår i: BMC Medical Imaging, ISSN 1471-2342, E-ISSN 1471-2342, Vol. 12, nr 1, s. 29-Artikel i tidskrift (Refereegranskat)
    Abstract [en]

    BACKGROUND: Muscle functions are generally assumed to affect a wide variety of conditions and activities, including pain, ischemic and neurological disorders, exercise and injury. It is therefore very desirable to obtain more information on musculoskeletal contributions to and activity during clinical processes such as the treatment of muscle injuries, post-surgery evaluations, and the monitoring of progressive degeneration in neuromuscular disorders.The spatial image resolution achievable with ultrasound systems has improved tremendously in the last few years and it is nowadays possible to study skeletal muscles in real-time during activity. However, ultrasound imaging has an inherent problem that makes it difficult to compare different measurement series or image sequences from two or more subjects. Due to physiological differences between different subjects, the ultrasound sequences will be visually different -- partly because of variation in probe placement and partly because of the difficulty of perfectly reproducing any given movement. METHODS: Ultrasound images of the biceps and calf of a single subject were transformed to achieve congruence and then efficiently compressed and stacked to facilitate analysis using a multivariate method known as O2PLS. O2PLS identifies related and unrelated variation in and between two sets of data such that different phases of the studied movements can be analysed. The methodology was used to study the dynamics of the Achilles tendon and the calf and also the Biceps brachii and upper arm. The movements of these parts of the body are both of interest in clinical orthopaedic research. RESULTS: This study extends the novel method of multivariate analysis of congruent images (MACI) to facilitate comparisons between two series of ultrasound images. This increases its potential range of medical applications and its utility for detecting, visualising and quantifying the dynamics and functions of skeletal muscle. CONCLUSIONS: The most important results of this study are that MACI with O2PLS is able to consistently extract meaningful variability from pairs of ultrasound sequences. The MACI method with O2PLS is a powerful tool with great potential for visualising and comparing dynamics between movements. It has many potential clinical applications in the study of muscle injuries, post-surgery evaluations and evaluations of rehabilitation, and the assessment of athletic training interventions.

    Ladda ner fulltext (pdf)
    Dynamic ultrasound imaging
  • 21.
    Löfstedt, Tommy
    et al.
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Brynolfsson, Patrik
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Nyholm, Tufve
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Garpebring, Anders
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Gray-level invariant Haralick texture features2019Ingår i: PLOS ONE, E-ISSN 1932-6203, Vol. 14, nr 2, artikel-id e0212110Artikel i tidskrift (Refereegranskat)
    Abstract [en]

    Haralick texture features are common texture descriptors in image analysis. To compute the Haralick features, the image gray-levels are reduced, a process called quantization. The resulting features depend heavily on the quantization step, so Haralick features are not reproducible unless the same quantization is performed. The aim of this work was to develop Haralick features that are invariant to the number of quantization gray-levels. By redefining the gray-level co-occurrence matrix (GLCM) as a discretized probability density function, it becomes asymptotically invariant to the quantization. The invariant and original features were compared using logistic regression classification to separate two classes based on the texture features. Classifiers trained on the invariant features showed higher accuracies, and had similar performance when training and test images had very different quantizations. In conclusion, using the invariant Haralick features, an image pattern will give the same texture feature values independent of image quantization.

    Ladda ner fulltext (pdf)
    fulltext
  • 22.
    Löfstedt, Tommy
    et al.
    Umeå universitet, Teknisk-naturvetenskapliga fakulteten, Kemiska institutionen.
    Eriksson, Lennart
    Gunilla Wormbs, Gunilla
    Trygg, Johan
    Umeå universitet, Teknisk-naturvetenskapliga fakulteten, Kemiska institutionen.
    Bi-modal OnPLS2012Ingår i: Journal of Chemometrics, ISSN 0886-9383, E-ISSN 1099-128X, Vol. 26, nr 6, s. 236-245Artikel i tidskrift (Refereegranskat)
    Abstract [en]

    This paper presents an extension to the recently published OnPLS data analysis method. Bi-modal OnPLS allows for arbitrary block relationships in both columns and rows and is able to extract orthogonal variation in both columns and rows without bias towards any particular direction or matrix: the method is fully symmetric with regard to both rows and columns.

    Bi-modal OnPLS extracts a minimal number of globally predictive score vectors that exhibit maximal covariance and correlation in the column space and a corresponding set of predictive loading vectors that exhibit maximal correlation in the row space. The method also extracts orthogonal variation (i.e. variation that is not related to all other matrices) in both columns and rows. The method was applied to two synthetic datasets and one real data set regarding sensory information and consumer likings of dairy products. It was shown that Bi-modal OnPLS greatly improves the intercorrelations between both loadings and scores while still finding the correct variation. This facilitates interpretation of the predictive components and makes it possible to study the orthogonal variation in the data.

  • 23.
    Löfstedt, Tommy
    et al.
    Brainomics Team, Neurospin, CEA Saclay, France.
    Guillemot, Vincent
    Brainomics Team, Neurospin, CEA Saclay, France.
    Frouin, Vincent
    Brainomics Team, Neurospin, CEA Saclay, France.
    Duchesnay, Edouard
    Brainomics Team, Neurospin, CEA Saclay, France.
    Hadj-Selem, Fouad
    Brainomics Team, Neurospin, CEA Saclay, France.
    Simulated Data for Linear Regression with Structured and Sparse Penalties: Introducing pylearn-simulate2018Ingår i: Journal of Statistical Software, E-ISSN 1548-7660, Vol. 87, nr 3Artikel i tidskrift (Refereegranskat)
    Abstract [en]

    A currently very active field of research is how to incorporate structure and prior knowledge in machine learning methods. It has lead to numerous developments in the field of non-smooth convex minimization. With recently developed methods it is possible to perform an analysis in which the computed model can be linked to a given structure of the data and simultaneously do variable selection to find a few important features in the data. However, there is still no way to unambiguously simulate data to test proposed algorithms, since the exact solutions to such problems are unknown.

    The main aim of this paper is to present a theoretical framework for generating simulated data. These simulated data are appropriate when comparing optimization algorithms in the context of linear regression problems with sparse and structured penalties. Additionally, this approach allows the user to control the signal-to-noise ratio, the correlation structure of the data and the optimization problem to which they are the solution.

    The traditional approach is to simulate random data without taking into account the actual model that will be fit to the data. But when using such an approach it is not possible to know the exact solution of the underlying optimization problem. With our contribution, it is possible to know the exact theoretical solution of a penalized linear regression problem, and it is thus possible to compare algorithms without the need to use, e.g., cross-validation.

    We also present our implementation, the Python package pylearn-simulate, available at https://github.com/neurospin/pylearn-simulate and released under the BSD 3clause license. We describe the package and give examples at the end of the paper.

  • 24.
    Löfstedt, Tommy
    et al.
    Umeå universitet, Teknisk-naturvetenskapliga fakulteten, Kemiska institutionen.
    Hadj-Selem, Fouad
    Guillemot, Vincent
    Philippe, Cathy
    Duchesnay, Edouard
    Frouin, Vincent
    Tenenhaus, Arthur
    Structured Variable Selection for Regularized Generalized Canonical Correlation Analysis2016Ingår i: MULTIPLE FACETS OF PARTIAL LEAST SQUARES AND RELATED METHODS / [ed] Abdi, H Vinzi, VE Russolillo, G Saporta, G Trinchera, L, SPRINGER INT PUBLISHING AG , 2016, Vol. 173, s. 129-139Konferensbidrag (Refereegranskat)
    Abstract [en]

    Regularized Generalized Canonical Correlation Analysis (RGCCA) extends regularized canonical correlation analysis to more than two sets of variables. Sparse GCCA(SGCCA) was recently proposed to address the issue of variable selection. However, the variable selection scheme offered by SGCCA is limited to the covariance (tau = 1) link between blocks. In this paper we go beyond the covariance link by proposing an extension of SGCCA for the full RGCCA model. (tau epsilon [0; 1]). In addition, we also propose an extension of SGCCA that exploits pre-given structural relationships between variables within blocks. Specifically, we propose an algorithm that allows structured and sparsity-inducing penalties to be included in the RGCCA optimization problem.

  • 25.
    Löfstedt, Tommy
    et al.
    Umeå universitet, Teknisk-naturvetenskapliga fakulteten, Kemiska institutionen.
    Hanafi, Mohamed
    Unité de Recherches "Sensometrics and Chemometrics", ONIRIS, Site de la Géraudière, BP 82 225 Nantes 44322 Cedex 03, France.
    Mazerolles, Gérard
    INRA-UMR 1083 SPO, INRA, 2 Place Viala, 34060 Montpellier, France.
    Trygg, Johan
    Umeå universitet, Teknisk-naturvetenskapliga fakulteten, Kemiska institutionen.
    OnPLS path modelling2012Ingår i: Chemometrics and Intelligent Laboratory Systems, ISSN 0169-7439, E-ISSN 1873-3239, Vol. 118, s. 139-149Artikel i tidskrift (Refereegranskat)
    Abstract [en]

    OnPLS was recently presented as a general extension of O2PLS to the multiblock case. OnPLS is equivalent to O2PLS in the case of two matrices, but generalises symmetrically to cases with more than two matrices, i.e. without giving preference to any one of the matrices.

    This article presents a straight-forward extension to this method and thereby also introduces the OPLS framework to the field of PLS path modelling. Path modelling links a number of data blocks to each other, thereby establishing a set of paths along which information is considered to flow between blocks, representing for instance a known time sequence, an assumed causality order, or some other chosen organising principle. Compared to existing methods for path analysis, OnPLS path modelling extracts a minimum number of predictive components that are maximally covarying with maximised correlation. This is a significant contribution to path modelling, because other methods may yield score vectors with variation that obstructs the interpretation. The method achieves this by extracting a set of "orthogonal" components that capture local phenomena orthogonal to the variation shared with all the connected blocks.

    Two applications will be used to illustrate the method. The first is based on a simulated dataset that show how the interpretation is improved by removing orthogonal variation and the second on a real data process for monitoring of protein structure changes during cheese ripening by analysing infrared data.

  • 26.
    Löfstedt, Tommy
    et al.
    Umeå universitet, Teknisk-naturvetenskapliga fakulteten, Kemiska institutionen.
    Hanafi, Mohamed
    Trygg, Johan
    Umeå universitet, Teknisk-naturvetenskapliga fakulteten, Kemiska institutionen.
    Multiblock and Path Modeling with OnPLS2013Ingår i: New Perspectives in Partial Least Squares and Related Methods (Part IV) / [ed] Herve Abdi, Wynne W. Chin, Vincenzo Esposito Vinzi, Giorgio Russolillo, Laura Trinchera, Springer Science+Business Media B.V., 2013, 56, , s. 209-220s. 209-220Konferensbidrag (Refereegranskat)
    Abstract [en]

    OnPLS was recently proposed as a general extension of O2PLS for applications in multiblock and path model analysis. OnPLS is very similar to O2PLS in the case with two matrices, but generalizes symmetrically to cases with more than two matrices without giving preference to any matrix.

    OnPLS extracts a minimal number of globally joint components that exhibit maximal covariance and correlation. A number of locally joint components are also extracted. These are shared between some matrices, but not between all. These components are also maximally covarying with maximal correlation. The variation that remains after the joint and locally joint variation has been extracted is unique to a particular matrix. This unique variation is orthogonal to all other matrices and captures phenomena specific in its matrix.

    The method's utility has been demonstrated by its application to synthetic datasets with very good results in terms of its ability to decompose the matrices. It has been shown that OnPLS affords a reduced number of globally joint components and increased intercorrelations of scores, and that it greatly facilitates interpretation of the models. Preliminary results in the application on real data has also given positive results. The results are similar to previous results using other multiblock and path model methods, but afford an increased interpretability because of the locally joint and unique components.

  • 27.
    Löfstedt, Tommy
    et al.
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik. Umeå universitet, Teknisk-naturvetenskapliga fakulteten, Institutionen för datavetenskap.
    Hellström, Max
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Bylund, Mikael
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Garpebring, Anders
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Bayesian non-linear regression with spatial priors for noise reduction and error estimation in quantitative MRI with an application in T1 estimation2020Ingår i: Physics in Medicine and Biology, ISSN 0031-9155, E-ISSN 1361-6560, Vol. 65, nr 22, artikel-id 225036Artikel i tidskrift (Refereegranskat)
    Abstract [en]

    Purpose. To develop a method that can reduce and estimate uncertainty in quantitative MR parameter maps without the need for hand-tuning of any hyperparameters.

    Methods. We present an estimation method where uncertainties are reduced by incorporating information on spatial correlations between neighbouring voxels. The method is based on a Bayesian hierarchical non-linear regression model, where the parameters of interest are sampled, using Markov chain Monte Carlo (MCMC), from a high-dimensional posterior distribution with a spatial prior. The degree to which the prior affects the model is determined by an automatic hyperparameter search using an information criterion and is, therefore, free from manual user-dependent tuning. The samples obtained further provide a convenient means to obtain uncertainties in both voxels and regions. The developed method was evaluated on T1 estimations based on the variable flip angle method.

    Results. The proposed method delivers noise-reduced T1 parameter maps with associated error estimates by combining MCMC sampling, the widely applicable information criterion, and total variation-based denoising. The proposed method results in an overall decrease in estimation error when compared to conventional voxel-wise maximum likelihood estimation. However, this comes with an increased bias in some regions, predominately at tissue interfaces, as well as an increase in computational time.

    Conclusions. This study provides a method that generates more precise estimates compared to the conventional method, without incorporating user subjectivity, and with the added benefit of uncertainty estimation.

    Ladda ner fulltext (pdf)
    fulltext
  • 28.
    Löfstedt, Tommy
    et al.
    Umeå universitet, Teknisk-naturvetenskapliga fakulteten, Kemiska institutionen.
    Hoffman, Daniel
    Umeå universitet, Teknisk-naturvetenskapliga fakulteten, Institutionen för molekylärbiologi (Teknisk-naturvetenskaplig fakultet).
    Trygg, Johan
    Umeå universitet, Teknisk-naturvetenskapliga fakulteten, Kemiska institutionen.
    Global, local and unique decompositions in OnPLS for multiblock data analysis2013Ingår i: Analytica Chimica Acta, ISSN 0003-2670, E-ISSN 1873-4324, Vol. 791, s. 13-24Artikel i tidskrift (Övrigt vetenskapligt)
    Abstract [en]

    Background OnPLS is an extension of O2PLS that decomposes a set of matrices, in either multiblock or path model analysis, such that each matrix consists of two parts: a globally joint part containing variation shared with all other connected matrices, and another containing unique or locally joint variation, i.e. variation that is specific to a particular matrix or shared with some, but not all, other connected matrices.

    Results A further extension of OnPLS suggested here decomposes the non-globally joint parts into locally joint and unique parts, using the OnPLS method to first find and extract a globally joint model, and then applying OnPLS recursively to subsets of matrices containing the non-globally joint variation remaining after the globally joint variation has been extracted. This results in a set of locally joint models. The variation that is left after the globally joint and locally joint variation has been extracted is not related (by definition) to the other matrices and thus represents the strictly unique variation specific to each matrix. The method's utility is demonstrated by its application to both a simulated data set and a real data set acquired from metabolomic, proteomic and transcriptomic profiling of three genotypes of hybrid aspen.

    Conclusions The results show that OnPLS can successfully decompose each matrix into global, local and unique models, resulting in lower numbers of globally joint components and higher intercorrelations of scores. OnPLS also increases the interpretability of models of connected matrices, because of the locally joint and unique models it generates.

  • 29.
    Löfstedt, Tommy
    et al.
    Umeå universitet, Teknisk-naturvetenskapliga fakulteten, Kemiska institutionen.
    Trygg, Johan
    Umeå universitet, Teknisk-naturvetenskapliga fakulteten, Kemiska institutionen.
    OnPLS—a novel multiblock method for the modelling of predictive and orthogonal variation2011Ingår i: Journal of Chemometrics, ISSN 0886-9383, E-ISSN 1099-128X, Vol. 25, nr 8, s. 441-455Artikel i tidskrift (Refereegranskat)
    Abstract [en]

    This paper presents a new multiblock analysis method called OnPLS, a general extension of O2PLS to the multiblock case. The proposed method is equivalent to O2PLS in cases involving only two matrices, but generalises to cases involving more than two matrices without giving preference to any particular matrix: the method is fully symmetric. OnPLS extracts a minimal number of globally predictive components that exhibit maximal covariance and correlation. Furthermore, the method can be used to study orthogonal variation, i.e. local phenomena captured in the data that are specific to individual combinations of matrices or to individual matrices. The method's utility was demonstrated by its application to three synthetic data sets. It was shown that OnPLS affords a reduced number of globally predictive components and increased intercorrelations of scores, and that it greatly facilitates interpretation of the predictive model.

  • 30. Mehta, Raghav
    et al.
    Filos, Angelos
    Baid, Ujjwal
    Sako, Chiharu
    McKinley, Richard
    Rebsamen, Michael
    Dätwyler, Katrin
    Meier, Raphael
    Radojewski, Piotr
    Murugesan, Gowtham Krishnan
    Nalawade, Sahil
    Ganesh, Chandan
    Wagner, Ben
    Yu, Fang F.
    Fei, Baowei
    Madhuranthakam, Ananth J.
    Maldjian, Joseph A.
    Daza, Laura
    Gómez, Catalina
    Arbeláez, Pablo
    Dai, Chengliang
    Wang, Shuo
    Reynaud, Hadrien
    Mo, Yuanhan
    Angelini, Elsa
    Guo, Yike
    Bai, Wenjia
    Banerjee, Subhashis
    Pei, Linmin
    AK, Murat
    Rosas-González, Sarahi
    Zemmoura, Ilyess
    Tauber, Clovis
    Vu, Minh Hoang
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Nyholm, Tufve
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Löfstedt, Tommy
    Umeå universitet, Teknisk-naturvetenskapliga fakulteten, Institutionen för datavetenskap.
    Ballestar, Laura Mora
    Vilaplana, Veronica
    McHugh, Hugh
    Talou, Gonzalo Maso
    Wang, Alan
    Patel, Jay
    Chang, Ken
    Hoebel, Katharina
    Gidwani, Mishka
    Arun, Nishanth
    Gupta, Sharut
    Aggarwal, Mehak
    Singh, Praveer
    Gerstner, Elizabeth R.
    Kalpathy-Cramer, Jayashree
    Boutry, Nicolas
    Huard, Alexis
    Vidyaratne, Lasitha
    Rahman, Md Monibor
    Iftekharuddin, Khan M.
    Chazalon, Joseph
    Puybareau, Elodie
    Tochon, Guillaume
    Ma, Jun
    Cabezas, Mariano
    Llado, Xavier
    Oliver, Arnau
    Valencia, Liliana
    Valverde, Sergi
    Amian, Mehdi
    Soltaninejad, Mohammadreza
    Myronenko, Andriy
    Hatamizadeh, Ali
    Feng, Xue
    Dou, Quan
    Tustison, Nicholas
    Meyer, Craig
    Shah, Nisarg A.
    Talbar, Sanjay
    Weber, Marc-André
    Mahajan, Abhishek
    Jakab, Andras
    Wiest, Roland
    Fathallah-Shaykh, Hassan M.
    Nazeri, Arash
    Milchenko, Mikhail
    Marcus, Daniel
    Kotrotsou, Aikaterini
    Colen, Rivka
    Freymann, John
    Kirby, Justin
    Davatzikos, Christos
    Menze, Bjoern
    Bakas, Spyridon
    Gal, Yarin
    Arbel, Tal
    QU-BraTS: MICCAI BraTS 2020 Challenge on QuantifyingUncertainty in Brain Tumor Segmentation - Analysis of Ranking Scores and Benchmarking Results2022Ingår i: Journal of Machine Learning for Biomedical Imaging, ISSN 2766-905X, s. 1-54, artikel-id 026Artikel i tidskrift (Refereegranskat)
    Abstract [en]

    Deep learning (DL) models have provided the state-of-the-art performance in a wide variety of medical imaging benchmarking challenges, including the Brain Tumor Segmentation (BraTS) challenges. However, the task of focal pathology multi-compartment segmentation (e.g., tumor and lesion sub-regions) is particularly challenging, and potential errors hinder the translation of DL models into clinical workflows. Quantifying the reliability of DL model predictions in the form of uncertainties, could enable clinical review of the most uncertain regions, thereby building trust and paving the way towards clinical translation. Recently, a number of uncertainty estimation methods have been introduced for DL medical image segmentation tasks. Developing scores to evaluate and compare the performance of uncertainty measures will assist the end-user in making more informed decisions. In this study, we explore and evaluate a score developed during the BraTS 2019-2020 task on uncertainty quantification (QU-BraTS), and designed to assess and rank uncertainty estimates for brain tumor multi-compartment segmentation. This score (1) rewards uncertainty estimates that produce high confidence in correct assertions, and those that assign low confidence levels at incorrect assertions, and (2) penalizes uncertainty measures that lead to a higher percentages of under-confident correct assertions. We further benchmark the segmentation uncertainties generated by 14 independent participating teams of QU-BraTS 2020, all of which also participated in the main BraTS segmentation task. Overall, our findings confirm the importance and complementary value that uncertainty estimates provide to segmentation algorithms, and hence highlight the need for uncertainty quantification in medical image analyses. Our evaluation code is made publicly available at https://github.com/RagMeh11/QU-BraTS

    Ladda ner fulltext (pdf)
    fulltext
  • 31.
    Meyers, Charles
    et al.
    Umeå universitet, Teknisk-naturvetenskapliga fakulteten, Institutionen för datavetenskap.
    Löfstedt, Tommy
    Umeå universitet, Teknisk-naturvetenskapliga fakulteten, Institutionen för datavetenskap.
    Elmroth, Erik
    Umeå universitet, Teknisk-naturvetenskapliga fakulteten, Institutionen för datavetenskap.
    Safety-critical computer vision: an empirical survey of adversarial evasion attacks and defenses on computer vision systems2023Ingår i: Artificial Intelligence Review, ISSN 0269-2821, E-ISSN 1573-7462, Vol. 56, s. 217-251Artikel i tidskrift (Refereegranskat)
    Abstract [en]

    Considering the growing prominence of production-level AI and the threat of adversarial attacks that can poison a machine learning model against a certain label, evade classification, or reveal sensitive data about the model and training data to an attacker, adversaries pose fundamental problems to machine learning systems. Furthermore, much research has focused on the inverse relationship between robustness and accuracy, raising problems for real-time and safety-critical systems particularly since they are governed by legal constraints in which software changes must be explainable and every change must be thoroughly tested. While many defenses have been proposed, they are often computationally expensive and tend to reduce model accuracy. We have therefore conducted a large survey of attacks and defenses and present a simple and practical framework for analyzing any machine-learning system from a safety-critical perspective using adversarial noise to find the upper bound of the failure rate. Using this method, we conclude that all tested configurations of the ResNet architecture fail to meet any reasonable definition of ‘safety-critical’ when tested on even small-scale benchmark data. We examine state of the art defenses and attacks against computer vision systems with a focus on safety-critical applications in autonomous driving, industrial control, and healthcare. By testing a combination of attacks and defenses, their efficacy, and their run-time requirements, we provide substantial empirical evidence that modern neural networks consistently fail to meet established safety-critical standards by a wide margin.

    Ladda ner fulltext (pdf)
    fulltext
  • 32. Peolsson, A.
    et al.
    Peolsson, Michael
    Umeå universitet, Teknisk-naturvetenskapliga fakulteten, Kemiska institutionen.
    Jull, G.
    Löfstedt, Tommy
    Umeå universitet, Teknisk-naturvetenskapliga fakulteten, Kemiska institutionen.
    Trygg, Johan
    Umeå universitet, Teknisk-naturvetenskapliga fakulteten, Kemiska institutionen.
    O'Leary, S.
    Preliminary evaluation of dorsal muscle activity during resisted cervical extension in patients with longstanding pain and disability following anterior cervical decompression and fusion surgery2015Ingår i: Physiotherapy, ISSN 0031-9406, E-ISSN 1873-1465, Vol. 101, nr 1, s. 69-74Artikel i tidskrift (Refereegranskat)
    Abstract [en]

    Objectives To compare mechanical activity (deformation and deformation rate) of the dorsal neck muscles between individuals with longstanding symptoms after anterior cervical decompression and fusion (ACDF) surgery and healthy controls.

    Design Preliminary cross-sectional study.

    Setting Neurosurgery clinic.

    Participants Ten individuals {mean age 60 [standard deviation (SD) 7.111 who had undergone ACDF surgery 10 to 13 years previously and 10 healthy age- and sex-matched controls.

    Main outcomes Mechanical activity of the different layers of dorsal neck muscles, measured at the C4 segment using ultrasonography (speckle tracking analysis) during a standardised, resisted cervical extension task.

    Results A significant group x muscle interaction was found for muscle deformation (P<0.03) but not for deformation rate (P>0.79). The ACDF group showed significantly less deformation of the semispinalis capitis muscle during the extension task compared with the control group [mean 3.12 (SD 2.06) and 6.64 (SD 4.17), respectively; mean difference 3.34 (95% confidence interval 0.54 to 7.21)].

    Conclusions As the semispinalis capitis muscle is a powerful neck extensor, the finding of altered activation following ACDF surgery lends support to the inclusion of exercise to train neck muscle performance in the management of these patients.

  • 33. Peolsson, Anneli
    et al.
    Löfstedt, Tommy
    Umeå universitet, Teknisk-naturvetenskapliga fakulteten, Kemiska institutionen.
    Trygg, Johan
    Umeå universitet, Teknisk-naturvetenskapliga fakulteten, Kemiska institutionen.
    Peolsson, Michael
    Umeå universitet, Teknisk-naturvetenskapliga fakulteten, Kemiska institutionen.
    Ultrasound imaging with speckle tracking of cervical muscle deformation and deformation rate: isometric contraction of patients after anterior cervical decompression and fusion for cervical disc disease and controls2012Ingår i: Manual Therapy, ISSN 1356-689X, E-ISSN 1532-2769, Vol. 17, nr 6, s. 519-525Artikel i tidskrift (Refereegranskat)
    Abstract [en]

    There is currently a lack of information regarding neck muscle activity during specific exercises. The purpose of the present study was to investigate deformation and deformation rate in different layers of dorsal and ventral neck muscles during isometric neck muscle contraction in individuals after anterior cervical decompression and fusion and in healthy controls. This study included 10 individuals (mean age 60 years; SD 7.1) with a verified, long-standing neck disorder and 10 healthy, age- and sex-matched controls. Ultrasonography and post-process speckle tracking analysis was used to investigate the degree and the rate of neck muscles motions at the C4 segmental level during sub-maximal, isometric resistance of the head in a seated position. None of the analyses performed showed significant differences between groups (p > 0.05). In the dorsal muscles, both groups exhibited a higher deformation rate in the multifidus than in the trapezius, splenius, and semispinalis capitis (p ≤ 0.01). In the neck disorder group, the multifidus also showed a higher deformation rate compared to the semispinalis cervicis (p = 0.02). In the ventral muscles of patients with neck disorders, the longus colli had a higher deformation rate than the sternocleidomastoid (p = 0.02). Among the healthy controls, the multifidus showed a higher degree of deformation (p = 0.02) than the trapezius. In conclusion, our results showed no significant differences between the two groups in mechanical neck muscle activation. Larger studies with different exercises, preferably with a standardized measure of resistance, are needed to investigate whether patients and controls show differences in deformation and deformation rates in neck muscles.

  • 34. Peolsson, Michael
    et al.
    Löfstedt, Tommy
    Umeå universitet, Teknisk-naturvetenskapliga fakulteten, Kemiska institutionen.
    Vogt, Susanna
    Umeå universitet, Teknisk-naturvetenskapliga fakulteten, Kemiska institutionen.
    Stenlund, Hans
    Umeå universitet, Teknisk-naturvetenskapliga fakulteten, Kemiska institutionen.
    Arndt, Anton
    Trygg, Johan
    Umeå universitet, Teknisk-naturvetenskapliga fakulteten, Kemiska institutionen.
    Modelling human musculoskeletal functional movements using ultrasound imaging2010Ingår i: BMC Medical Imaging, ISSN 1471-2342, E-ISSN 1471-2342, Vol. 10, nr 9Artikel i tidskrift (Refereegranskat)
    Abstract [en]

    BACKGROUND: A widespread and fundamental assumption in the health sciences is that muscle functions are related to a wide variety of conditions, for example pain, ischemic and neurological disorder, exercise and injury. It is therefore highly desirable to study musculoskeletal contributions in clinical applications such as the treatment of muscle injuries, post-surgery evaluations, monitoring of progressive degeneration in neuromuscular disorders, and so on.The spatial image resolution in ultrasound systems has improved tremendously in the last few years and nowadays provides detailed information about tissue characteristics. It is now possible to study skeletal muscles in real-time during activity.

    METHODS: The ultrasound images are transformed to be congruent and are effectively compressed and stacked in order to be analysed with multivariate techniques. The method is applied to a relevant clinical orthopaedic research field, namely to describe the dynamics in the Achilles tendon and the calf during real-time movements.

    RESULTS: This study introduces a novel method to medical applications that can be used to examine ultrasound image sequences and to detect, visualise and quantify skeletal muscle dynamics and functions.

    CONCLUSIONS: This new objective method is a powerful tool to use when visualising tissue activity and dynamics of musculoskeletal ultrasound registrations.

    Ladda ner fulltext (pdf)
    fulltext
  • 35.
    Simkó, Attila
    et al.
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper.
    Bylund, Mikael
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper.
    Jönsson, Gustav
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper.
    Löfstedt, Tommy
    Umeå universitet, Teknisk-naturvetenskapliga fakulteten, Institutionen för datavetenskap.
    Garpebring, Anders
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper.
    Nyholm, Tufve
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper.
    Jonsson, Joakim
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper.
    Towards MR contrast independent synthetic CT generation2023Ingår i: Zeitschrift für Medizinische Physik, ISSN 0939-3889, E-ISSN 1876-4436Artikel i tidskrift (Refereegranskat)
    Abstract [en]

    The use of synthetic CT (sCT) in the radiotherapy workflow would reduce costs and scan time while removing the uncertainties around working with both MR and CT modalities. The performance of deep learning (DL) solutions for sCT generation is steadily increasing, however most proposed methods were trained and validated on private datasets of a single contrast from a single scanner. Such solutions might not perform equally well on other datasets, limiting their general usability and therefore value. Additionally, functional evaluations of sCTs such as dosimetric comparisons with CT-based dose calculations better show the impact of the methods, but the evaluations are more labor intensive than pixel-wise metrics.

    To improve the generalization of an sCT model, we propose to incorporate a pre-trained DL model to pre-process the input MR images by generating artificial proton density, T1 and T2 maps (i.e. contrast-independent quantitative maps), which are then used for sCT generation. Using a dataset of only T2w MR images, the robustness towards input MR contrasts of this approach is compared to a model that was trained using the MR images directly. We evaluate the generated sCTs using pixel-wise metrics and calculating mean radiological depths, as an approximation of the mean delivered dose. On T2w images acquired with the same settings as the training dataset, there was no significant difference between the performance of the models. However, when evaluated on T1w images, and a wide range of other contrasts and scanners from both public and private datasets, our approach outperforms the baseline model. Using a dataset of T2w MR images, our proposed model implements synthetic quantitative maps to generate sCT images, improving the generalization towards other contrasts. Our code and trained models are publicly available.

    Ladda ner fulltext (pdf)
    fulltext
  • 36.
    Simkó, Attila
    et al.
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper.
    Garpebring, Anders
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper.
    Jonsson, Joakim
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper.
    Nyholm, Tufve
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper.
    Löfstedt, Tommy
    Umeå universitet, Teknisk-naturvetenskapliga fakulteten, Institutionen för datavetenskap.
    Reproducibility of the methods in medical imaging with deep learning2023Konferensbidrag (Refereegranskat)
    Abstract [en]

    Concerns about the reproducibility of deep learning research are more prominent than ever, with no clear solution in sight. The Medical Imaging with Deep Learning (MIDL) conference has made advancements in employing empirical rigor with regards to reproducibility by advocating open access, and recently also recommending authors to make their code public---both aspects being adopted by the majority of the conference submissions. We have evaluated all accepted full paper submissions to MIDL between 2018 and 2022 using established, but adjusted guidelines addressing the reproducibility and quality of the public repositories. The evaluations show that publishing repositories and using public datasets are becoming more popular, which helps traceability, but the quality of the repositories shows room for improvement in every aspect. Merely 22% of all submissions contain a repository that was deemed repeatable using our evaluations. From the commonly encountered issues during the evaluations, we propose a set of guidelines for machine learning-related research for medical imaging applications, adjusted specifically for future submissions to MIDL. We presented our results to future MIDL authors who were eager to continue an open discussion on the topic of code reproducibility.

  • 37.
    Simkó, Attila
    et al.
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Löfstedt, Tommy
    Umeå universitet, Teknisk-naturvetenskapliga fakulteten, Institutionen för datavetenskap.
    Garpebring, Anders
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Bylund, Mikael
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Nyholm, Tufve
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Jonsson, Joakim
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Changing the Contrast of Magnetic Resonance Imaging Signals using Deep Learning2021Ingår i: Proceedings of the Fourth Conference on Medical Imaging with Deep Learning, PMLR / [ed] Mattias Heinrich; Qi Dou; Marleen de Bruijne; Jan Lellmann; Alexander Schläfer; Floris Ernst, Lübeck University; Hamburg University of Technology , 2021, Vol. 143, s. 713-727Konferensbidrag (Refereegranskat)
    Abstract [en]

     The contrast settings to select before acquiring magnetic resonance imaging (MRI) signal depend heavily on the subsequent tasks. As each contrast highlights different tissues, automated segmentation tools for example might be optimized for a certain contrast. While for radiotherapy, multiple scans of the same region with different contrasts can achieve a better accuracy for delineating tumours and organs at risk. Unfortunately, the optimal contrast for the subsequent automated methods might not be known during the time of signal acquisition, and performing multiple scans with different contrasts increases the total examination time and registering the sequences introduces extra work and potential errors. Building on the recent achievements of deep learning in medical applications, the presented work describes a novel approach for transferring any contrast to any other. The novel model architecture incorporates the signal equation for spin echo sequences, and hence the model inherently learns the unknown quantitative maps for proton density, 𝑇1 and 𝑇2 relaxation times (𝑃𝐷, 𝑇1 and 𝑇2, respectively). This grants the model the ability to retrospectively reconstruct spin echo sequences by changing the contrast settings Echo and Repetition Time (𝑇𝐸 and 𝑇𝑅, respectively). The model learns to identify the contrast of pelvic MR images, therefore no paired data of the same anatomy from different contrasts is required for training. This means that the experiments are easily reproducible with other contrasts or other patient anatomies. Despite the contrast of the input image, the model achieves accurate results for reconstructing signal with contrasts available for evaluation. For the same anatomy, the quantitative maps are consistent for a range of contrasts of input images. Realized in practice, the proposed method would greatly simplify the modern radiotherapy pipeline. The trained model is made public together with a tool for testing the model on example images. 

    Ladda ner fulltext (pdf)
    fulltext
  • 38.
    Simkó, Attila
    et al.
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Löfstedt, Tommy
    Umeå universitet, Teknisk-naturvetenskapliga fakulteten, Institutionen för datavetenskap.
    Garpebring, Anders
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper.
    Nyholm, Tufve
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Jonsson, Joakim
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    MRI bias field correction with an implicitly trained CNN2022Ingår i: Proceedings of the 5th international conference on medical imaging with deep learning / [ed] Ender Konukoglu; Bjoern Menze; Archana Venkataraman; Christian Baumgartner; Qi Dou; Shadi Albarqouni, ML Research Press , 2022, s. 1125-1138Konferensbidrag (Refereegranskat)
    Abstract [en]

    In magnetic resonance imaging (MRI), bias fields are difficult to correct since they are inherently unknown. They cause intra-volume intensity inhomogeneities which limit the performance of subsequent automatic medical imaging tasks, \eg, tissue-based segmentation. Since the ground truth is unavailable, training a supervised machine learning solution requires approximating the bias fields, which limits the resulting method. We introduce implicit training which sidesteps the inherent lack of data and allows the training of machine learning solutions without ground truth. We describe how training a model implicitly for bias field correction allows using non-medical data for training, achieving a highly generalized model. The implicit approach was compared to a more traditional training based on medical data. Both models were compared to an optimized N4ITK method, with evaluations on six datasets. The implicitly trained model improved the homogeneity of all encountered medical data, and it generalized better for a range of anatomies, than the model trained traditionally. The model achieves a significant speed-up over an optimized N4ITK method—by a factor of 100, and after training, it also requires no parameters to tune. For tasks such as bias field correction - where ground truth is generally not available, but the characteristics of the corruption are known - implicit training promises to be a fruitful alternative for highly generalized solutions.

    Ladda ner fulltext (pdf)
    fulltext
  • 39.
    Simkó, Attila
    et al.
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper.
    Ruiter, Simone
    Department of Biomedical Engineering, Eindhoven University of Technology, Eindhoven, Netherlands.
    Löfstedt, Tommy
    Umeå universitet, Teknisk-naturvetenskapliga fakulteten, Institutionen för datavetenskap.
    Garpebring, Anders
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper.
    Nyholm, Tufve
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper.
    Bylund, Mikael
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper.
    Jonsson, Joakim
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper.
    Improving MR image quality with a multi-task model, using convolutional losses2023Ingår i: BMC Medical Imaging, ISSN 1471-2342, E-ISSN 1471-2342, Vol. 23, nr 1, artikel-id 148Artikel i tidskrift (Refereegranskat)
    Abstract [en]

    PURPOSE: During the acquisition of MRI data, patient-, sequence-, or hardware-related factors can introduce artefacts that degrade image quality. Four of the most significant tasks for improving MRI image quality have been bias field correction, super-resolution, motion-, and noise correction. Machine learning has achieved outstanding results in improving MR image quality for these tasks individually, yet multi-task methods are rarely explored.

    METHODS: In this study, we developed a model to simultaneously correct for all four aforementioned artefacts using multi-task learning. Two different datasets were collected, one consisting of brain scans while the other pelvic scans, which were used to train separate models, implementing their corresponding artefact augmentations. Additionally, we explored a novel loss function that does not only aim to reconstruct the individual pixel values, but also the image gradients, to produce sharper, more realistic results. The difference between the evaluated methods was tested for significance using a Friedman test of equivalence followed by a Nemenyi post-hoc test.

    RESULTS: Our proposed model generally outperformed other commonly-used correction methods for individual artefacts, consistently achieving equal or superior results in at least one of the evaluation metrics. For images with multiple simultaneous artefacts, we show that the performance of using a combination of models, trained to correct individual artefacts depends heavily on the order that they were applied. This is not an issue for our proposed multi-task model. The model trained using our novel convolutional loss function always outperformed the model trained with a mean squared error loss, when evaluated using Visual Information Fidelity, a quality metric connected to perceptual quality.

    CONCLUSION: We trained two models for multi-task MRI artefact correction of brain, and pelvic scans. We used a novel loss function that significantly improves the image quality of the outputs over using mean squared error. The approach performs well on real world data, and it provides insight into which artefacts it detects and corrects for. Our proposed model and source code were made publicly available.

    Ladda ner fulltext (pdf)
    fulltext
  • 40. Srivastava, Vaibhav
    et al.
    Obudulu, Ogonna
    Umeå universitet, Teknisk-naturvetenskapliga fakulteten, Kemiska institutionen. Computational life science cluster (CLiC), Umeå University and Swedish University of Agricultural Sciences.
    Bygdell, Joakim
    Löfstedt, Tommy
    Umeå universitet, Teknisk-naturvetenskapliga fakulteten, Kemiska institutionen. Computational life science cluster (CLiC), Umeå University.
    Rydén, Patrik
    Umeå universitet, Teknisk-naturvetenskapliga fakulteten, Institutionen för matematik och matematisk statistik. Umeå universitet, Teknisk-naturvetenskapliga fakulteten, Kemiska institutionen. Computational life science cluster (CLiC), Umeå University.
    Nilsson, Robert
    Ahnlund, Maria
    Johansson, Annika
    Jonsson, Pär
    Umeå universitet, Teknisk-naturvetenskapliga fakulteten, Kemiska institutionen. Computational life science cluster (CLiC), Umeå University.
    Freyhult, Eva
    Umeå universitet, Teknisk-naturvetenskapliga fakulteten, Kemiska institutionen. Umeå universitet, Medicinska fakulteten, Institutionen för klinisk mikrobiologi, Klinisk bakteriologi. Computational life science cluster (CLiC), Umeå University.
    Qvarnström, Johanna
    Karlsson, Jan
    Umeå universitet, Teknisk-naturvetenskapliga fakulteten, Institutionen för fysiologisk botanik. Umeå universitet, Teknisk-naturvetenskapliga fakulteten, Umeå Plant Science Centre (UPSC).
    Melzer, Michael
    Moritz, Thomas
    Trygg, Johan
    Umeå universitet, Teknisk-naturvetenskapliga fakulteten, Kemiska institutionen. Computational life science cluster (CLiC), Umeå University.
    Hvidsten, Torgeir R
    Umeå universitet, Teknisk-naturvetenskapliga fakulteten, Institutionen för fysiologisk botanik. Umeå universitet, Teknisk-naturvetenskapliga fakulteten, Umeå Plant Science Centre (UPSC). Umeå universitet, Teknisk-naturvetenskapliga fakulteten, Kemiska institutionen. Computational life science cluster (CLiC), Umeå University and Department of Chemistry, Biotechnology; Food Science, Norwegian, University of Life Sciences, Ås Norwegian, Norway.
    Wingsle, Gunnar
    OnPLS integration of transcriptomic, proteomic and metabolomic data shows multi-level oxidative stress responses in the cambium of transgenic hipI- superoxide dismutase Populus plants2013Ingår i: BMC Genomics, E-ISSN 1471-2164, Vol. 14, artikel-id 893Artikel i tidskrift (Refereegranskat)
    Abstract [en]

    BACKGROUND: Reactive oxygen species (ROS) are involved in the regulation of diverse physiological processes in plants, including various biotic and abiotic stress responses. Thus, oxidative stress tolerance mechanisms in plants are complex, and diverse responses at multiple levels need to be characterized in order to understand them. Here we present system responses to oxidative stress in Populus by integrating data from analyses of the cambial region of wild-type controls and plants expressing high-isoelectric-point superoxide dismutase (hipI-SOD) transcripts in antisense orientation showing a higher production of superoxide. The cambium, a thin cell layer, generates cells that differentiate to form either phloem or xylem and is hypothesized to be a major reason for phenotypic perturbations in the transgenic plants. Data from multiple platforms including transcriptomics (microarray analysis), proteomics (UPLC/QTOF-MS), and metabolomics (GC-TOF/MS, UPLC/MS, and UHPLC-LTQ/MS) were integrated using the most recent development of orthogonal projections to latent structures called OnPLS. OnPLS is a symmetrical multi-block method that does not depend on the order of analysis when more than two blocks are analysed. Significantly affected genes, proteins and metabolites were then visualized in painted pathway diagrams.

    RESULTS: The main categories that appear to be significantly influenced in the transgenic plants were pathways related to redox regulation, carbon metabolism and protein degradation, e.g. the glycolysis and pentose phosphate pathways (PPP). The results provide system-level information on ROS metabolism and responses to oxidative stress, and indicate that some initial responses to oxidative stress may share common pathways.

    CONCLUSION: The proposed data evaluation strategy shows an efficient way of compiling complex, multi-platform datasets to obtain significant biological information.

    Ladda ner fulltext (pdf)
    fulltext
  • 41.
    Vu, Minh
    et al.
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Grimbergen, Guus
    Eindhoven University of Technology, 5612 AZ Eindhoven, the Netherlands.
    Simkó, Attila
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Nyholm, Tufve
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Löfstedt, Tommy
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik. Umeå universitet, Teknisk-naturvetenskapliga fakulteten, Kemiska institutionen.
    Localization Network and End-to-End Cascaded U-Nets for Kidney Tumor Segmentation2019Konferensbidrag (Övrigt vetenskapligt)
    Abstract [en]

    Kidney tumor segmentation emerges as a new frontier of computer vision in medical imaging. This is partly due to its challenging manual annotation and great medical impact. Within the scope of the Kidney Tumor Segmentation Challenge 2019, that is aiming at combined kidney and tumor segmentation, this work proposes a novel combination of 3D U-Nets---collectively denoted TuNet---utilizing the resulting kidney masks for the consecutive tumor segmentation. The proposed method achieves a Sørensen-Dice coefficient score of 0.902 for the kidney, and 0.408 for the tumor segmentation, computed from a five-fold cross-validation on the 210 patients available in the data.

  • 42.
    Vu, Minh H.
    et al.
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper.
    Garpebring, Anders
    Nyholm, Tufve
    Löfstedt, Tommy
    Compressing the activation maps in deep convolutional neural networks and the regularization effect of compressionManuskript (preprint) (Övrigt vetenskapligt)
  • 43.
    Vu, Minh H.
    et al.
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Grimbergen, Guus
    Department of Biomedical Engineering, Eindhoven University of Technology, Eindhoven 5612 AZ, the Netherlands.
    Nyholm, Tufve
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Löfstedt, Tommy
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Evaluation of multislice inputs to convolutional neural networks for medical image segmentation2020Ingår i: Medical physics (Lancaster), ISSN 0094-2405, Vol. 47, nr 12, s. 6216-6231Artikel i tidskrift (Refereegranskat)
    Abstract [en]

    Purpose: When using convolutional neural networks (CNNs) for segmentation of organs and lesions in medical images, the conventional approach is to work with inputs and outputs either as single slice [two-dimensional (2D)] or whole volumes [three-dimensional (3D)]. One common alternative, in this study denoted as pseudo-3D, is to use a stack of adjacent slices as input and produce a prediction for at least the central slice. This approach gives the network the possibility to capture 3D spatial information, with only a minor additional computational cost.

    Methods: In this study, we systematically evaluate the segmentation performance and computational costs of this pseudo-3D approach as a function of the number of input slices, and compare the results to conventional end-to-end 2D and 3D CNNs, and to triplanar orthogonal 2D CNNs. The standard pseudo-3D method regards the neighboring slices as multiple input image channels. We additionally design and evaluate a novel, simple approach where the input stack is a volumetric input that is repeatably convolved in 3D to obtain a 2D feature map. This 2D map is in turn fed into a standard 2D network. We conducted experiments using two different CNN backbone architectures and on eight diverse data sets covering different anatomical regions, imaging modalities, and segmentation tasks.

    Results: We found that while both pseudo-3D methods can process a large number of slices at once and still be computationally much more efficient than fully 3D CNNs, a significant improvement over a regular 2D CNN was only observed with two of the eight data sets. triplanar networks had the poorest performance of all the evaluated models. An analysis of the structural properties of the segmentation masks revealed no relations to the segmentation performance with respect to the number of input slices. A post hoc rank sum test which combined all metrics and data sets yielded that only our newly proposed pseudo-3D method with an input size of 13 slices outperformed almost all methods.

    Conclusion: In the general case, multislice inputs appear not to improve segmentation results over using 2D or 3D CNNs. For the particular case of 13 input slices, the proposed novel pseudo-3D method does appear to have a slight advantage across all data sets compared to all other methods evaluated in this work.

    Ladda ner fulltext (pdf)
    fulltext
  • 44.
    Vu, Minh H.
    et al.
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper.
    Tronchin, L.
    Nyholm, Tufve
    Löfstedt, Tommy
    Using synthetic images to augment small medical image datasetsManuskript (preprint) (Övrigt vetenskapligt)
  • 45.
    Vu, Minh Hoang
    et al.
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Löfstedt, Tommy
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Nyholm, Tufve
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Sznitman, Raphael
    ARTORG Center, University of Bern, 3008 Bern, Switzerland.
    A Question-Centric Model for Visual Question Answering in Medical Imaging2020Ingår i: IEEE Transactions on Medical Imaging, ISSN 0278-0062, E-ISSN 1558-254X, Vol. 39, nr 9, s. 2856-2868Artikel i tidskrift (Refereegranskat)
    Abstract [en]

    Deep learning methods have proven extremely effective at performing a variety of medical image analysis tasks. With their potential use in clinical routine, their lack of transparency has however been one of their few weak points, raising concerns regarding their behavior and failure modes. While most research to infer model behavior has focused on indirect strategies that estimate prediction uncertainties and visualize model support in the input image space, the ability to explicitly query a prediction model regarding its image content offers a more direct way to determine the behavior of trained models. To this end, we present a novel Visual Question Answering approach that allows an image to be queried by means of a written question. Experiments on a variety of medical and natural image datasets show that by fusing image and question features in a novel way, the proposed approach achieves an equal or higher accuracy compared to current methods.

  • 46.
    Vu, Minh Hoang
    et al.
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Norman, Gabriella
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Nyholm, Tufve
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Löfstedt, Tommy
    Umeå universitet, Teknisk-naturvetenskapliga fakulteten, Institutionen för datavetenskap.
    A Data-Adaptive Loss Function for Incomplete Data and Incremental Learning in Semantic Image Segmentation2022Ingår i: IEEE Transactions on Medical Imaging, ISSN 0278-0062, E-ISSN 1558-254X, Vol. 41, nr 6, s. 1320-1330Artikel i tidskrift (Refereegranskat)
    Abstract [en]

    In the last years, deep learning has dramatically improved the performances in a variety of medical image analysis applications. Among different types of deep learning models, convolutional neural networks have been among the most successful and they have been used in many applications in medical imaging.

    Training deep convolutional neural networks often requires large amounts of image data to generalize well to new unseen images. It is often time-consuming and expensive to collect large amounts of data in the medical image domain due to expensive imaging systems, and the need for experts to manually make ground truth annotations. A potential problem arises if new structures are added when a decision support system is already deployed and in use. Since the field of radiation therapy is constantly developing, the new structures would also have to be covered by the decision support system.

    In the present work, we propose a novel loss function to solve multiple problems: imbalanced datasets, partially-labeled data, and incremental learning. The proposed loss function adapts to the available data in order to utilize all available data, even when some have missing annotations. We demonstrate that the proposed loss function also works well in an incremental learning setting, where an existing model is easily adapted to semi-automatically incorporate delineations of new organs when they appear. Experiments on a large in-house dataset show that the proposed method performs on par with baseline models, while greatly reducing the training time and eliminating the hassle of maintaining multiple models in practice.

    Ladda ner fulltext (pdf)
    fulltext
  • 47.
    Vu, Minh Hoang
    et al.
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper.
    Nyholm, Tufve
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Löfstedt, Tommy
    Umeå universitet, Teknisk-naturvetenskapliga fakulteten, Institutionen för datavetenskap.
    Multi-decoder Networks with Multi-denoising Inputs for Tumor Segmentation2021Ingår i: Brainlesion: Glioma, Multiple Sclerosis, Stroke and Traumatic Brain Injuries / [ed] Alessandro Crimi, Spyridon Bakas, Springer, 2021, s. 412-423Konferensbidrag (Refereegranskat)
    Abstract [en]

    Automatic segmentation of brain glioma from multimodal MRI scans plays a key role in clinical trials and practice. Unfortunately, manual segmentation is very challenging, time-consuming, costly, and often inaccurate despite human expertise due to the high variance and high uncertainty in the human annotations. In the present work, we develop an end-to-end deep-learning-based segmentation method using a multi-decoder architecture by jointly learning three separate sub-problems using a partly shared encoder. We also propose to apply smoothing methods to the input images to generate denoised versions as additional inputs to the network. The validation performance indicates an improvement when using the proposed method. The proposed method was ranked 2nd in the task of Quantification of Uncertainty in Segmentation in the Brain Tumors in Multimodal Magnetic Resonance Imaging Challenge 2020.

  • 48.
    Vu, Minh Hoang
    et al.
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Nyholm, Tufve
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Löfstedt, Tommy
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    TuNet: End-to-End Hierarchical Brain Tumor Segmentation Using Cascaded Networks2020Ingår i: Brainlesion: Glioma, Multiple Sclerosis, Stroke and Traumatic Brain Injuries: 5th International Workshop, BrainLes 2019, Held in Conjunction with MICCAI 2019, Shenzhen, China, October 17, 2019, Revised Selected Papers, Part I / [ed] Alessandro Crimi and Spyridon Bakas, Cham: Springer, 2020, s. 174-186Konferensbidrag (Refereegranskat)
    Abstract [en]

    Glioma is one of the most common types of brain tumors; it arises in the glial cells in the human brain and in the spinal cord. In addition to having a high mortality rate, glioma treatment is also very expensive. Hence, automatic and accurate segmentation and measurement from the early stages are critical in order to prolong the survival rates of the patients and to reduce the costs of the treatment. In the present work, we propose a novel end-to-end cascaded network for semantic segmentation in the Brain Tumors in Multimodal Magnetic Resonance Imaging Challenge 2019 that utilizes the hierarchical structure of the tumor sub-regions with ResNet-like blocks and Squeeze-and-Excitation modules after each convolution and concatenation block. By utilizing cross-validation, an average ensemble technique, and a simple post-processing technique, we obtained dice scores of 88.06, 80.84, and 80.29, and Hausdorff Distances (95th percentile) of 6.10, 5.17, and 2.21 for the whole tumor, tumor core, and enhancing tumor, respectively, on the online test set. The proposed method was ranked among the top in the task of Quantification of Uncertainty in Segmentation.

  • 49.
    Vu, Minh Hoang
    et al.
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Sznitman, Raphael
    ARTORG Center, University of Bern, Bern, Switzerland.
    Nyholm, Tufve
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik.
    Löfstedt, Tommy
    Umeå universitet, Medicinska fakulteten, Institutionen för strålningsvetenskaper, Radiofysik. Umeå universitet, Teknisk-naturvetenskapliga fakulteten, Kemiska institutionen.
    Ensemble of Streamlined Bilinear Visual Question Answering Models for the ImageCLEF 2019 Challenge in the Medical Domain2019Ingår i: CLEF 2019: Working Notes of CLEF 2019 - Conference and Labs of the Evaluation Forum / [ed] Linda Cappellato, Nicola Ferro, David E. Losada, and Henning Müller, 2019, Vol. 2380Konferensbidrag (Övrigt vetenskapligt)
    Abstract [en]

    This paper describes the contribution by participants from Umeå University, Sweden, in collaboration with the University of Bern, Switzerland, for the Medical Domain Visual Question Answering challenge hosted by ImageCLEF 2019. We proposed a novel Visual Question Answering approach that leverages a bilinear model to aggregateand synthesize extracted image and question features. While we did not make use of any additional training data, our model used an attention scheme to focus on the relevant input context and was further boosted by using an ensemble of trained models. We show here that the proposed approach performs at state-of-the-art levels, and provides an improvement over several existing methods. The proposed method was ranked 3rd in the Medical Domain Visual Question Answering challenge of ImageCLEF 2019.

    Ladda ner fulltext (pdf)
    fulltext
1 - 49 av 49
RefereraExporteraLänk till träfflistan
Permanent länk
Referera
Referensformat
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Annat format
Fler format
Språk
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Annat språk
Fler språk
Utmatningsformat
  • html
  • text
  • asciidoc
  • rtf