Pub Date : 2008-05-14DOI: 10.1109/ISBI.2008.4541216
S. Sen, M. Foskey, J. Marron, M. Styner
The Support Vector Machine (SVM) is a powerful tool for classification. We generalize SVM to work with data objects that are naturally understood to be lying on curved manifolds, and not in the usual d-dimensional Euclidean space. Such data arise from medial representations (m-reps) in medical images, Diffusion Tensor-MRI (DT-MRI), diffeomorphisms, etc. Considering such data objects to be embedded in higher dimensional Euclidean space results in invalid projections (on the separating direction) while Kernel Embedding does not provide a natural separating direction. We use geodesic distances, defined on the manifold to formulate our methodology. This approach addresses the important issue of analyzing the change that accompanies the difference between groups by implicitly defining the notions of separating surface and separating direction on the manifold. The methods are applied in shape analysis with target data being m-reps of 3 dimensional medical images.
{"title":"Support vector machine for data on manifolds: An application to image analysis","authors":"S. Sen, M. Foskey, J. Marron, M. Styner","doi":"10.1109/ISBI.2008.4541216","DOIUrl":"https://doi.org/10.1109/ISBI.2008.4541216","url":null,"abstract":"The Support Vector Machine (SVM) is a powerful tool for classification. We generalize SVM to work with data objects that are naturally understood to be lying on curved manifolds, and not in the usual d-dimensional Euclidean space. Such data arise from medial representations (m-reps) in medical images, Diffusion Tensor-MRI (DT-MRI), diffeomorphisms, etc. Considering such data objects to be embedded in higher dimensional Euclidean space results in invalid projections (on the separating direction) while Kernel Embedding does not provide a natural separating direction. We use geodesic distances, defined on the manifold to formulate our methodology. This approach addresses the important issue of analyzing the change that accompanies the difference between groups by implicitly defining the notions of separating surface and separating direction on the manifold. The methods are applied in shape analysis with target data being m-reps of 3 dimensional medical images.","PeriodicalId":184204,"journal":{"name":"2008 5th IEEE International Symposium on Biomedical Imaging: From Nano to Macro","volume":"120 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2008-05-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114561312","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2008-05-14DOI: 10.1109/ISBI.2008.4541231
D. Afonso, J. Sanches, M. Lauterbach
The emerging functional MRI (magnetic resonance imaging), fMRI, imaging modality was developed to obtain non-invasive information regarding the neural processes behind pre-determined task. The data is gathered in such a way that the extraction certainty of the desired information is maximized. Still this is a difficult task due to low Signal-to-Noise Ratio (SNR), corrupting noise and artifacts from several sources. The most prevalent method, here called SPM-GLM uses a conventional statistical inference methodology based on the t-statistics, where it assumes a rather rigid shape on the BOLD hemodynamic response function (HRF), constant for the whole region of interest (ROI). A new algorithm, designed in a Bayesian framework, is presented in this paper, called SPM-MAP. The algorithm jointly detects the brain activated regions and the underlying HRF in an adaptative and local basis. This approach presents two main advantages: (1) the activity detection benefits from the method's high flexibility toward the HRF shape; (2) it provides local estimations for the HRF. The SPM-MAP algorithm is validated by using Monte Carlo tests with synthetic data and comparisons with the SPM-GLM are also performed. Tests using real data are also performed and results are compared with the ones provided by the SPM-GLM method tuned by the medical doctor.
{"title":"FMRI brain activity and underlying hemodynamics estimation in a new Bayesian framework","authors":"D. Afonso, J. Sanches, M. Lauterbach","doi":"10.1109/ISBI.2008.4541231","DOIUrl":"https://doi.org/10.1109/ISBI.2008.4541231","url":null,"abstract":"The emerging functional MRI (magnetic resonance imaging), fMRI, imaging modality was developed to obtain non-invasive information regarding the neural processes behind pre-determined task. The data is gathered in such a way that the extraction certainty of the desired information is maximized. Still this is a difficult task due to low Signal-to-Noise Ratio (SNR), corrupting noise and artifacts from several sources. The most prevalent method, here called SPM-GLM uses a conventional statistical inference methodology based on the t-statistics, where it assumes a rather rigid shape on the BOLD hemodynamic response function (HRF), constant for the whole region of interest (ROI). A new algorithm, designed in a Bayesian framework, is presented in this paper, called SPM-MAP. The algorithm jointly detects the brain activated regions and the underlying HRF in an adaptative and local basis. This approach presents two main advantages: (1) the activity detection benefits from the method's high flexibility toward the HRF shape; (2) it provides local estimations for the HRF. The SPM-MAP algorithm is validated by using Monte Carlo tests with synthetic data and comparisons with the SPM-GLM are also performed. Tests using real data are also performed and results are compared with the ones provided by the SPM-GLM method tuned by the medical doctor.","PeriodicalId":184204,"journal":{"name":"2008 5th IEEE International Symposium on Biomedical Imaging: From Nano to Macro","volume":"65 4 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2008-05-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129600628","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2008-05-14DOI: 10.1109/ISBI.2008.4540951
S. Lee, M. Cho, J. Choi
An X-ray micro-tomography system has been developed for in vivo small animal imaging studies. For efficient in vivo scanning, it has a rotating gantry on which the X-ray source and the flat panel X-ray detector are mounted. To reconstruct artifact-free images of a small local region inside the animal subject from the truncated projection data, the projection data from the large field of view (FOV) scan of the whole animal subject are combined with the projection data from the small FOV scan of the region of interest. For the acquisition of X-ray projection data, a 1248 times 1248 flat-panel x-ray detector with the pixel pitch of 100 mum has been used. The developed system has the spatial resolution of 12 lp/mm when the highest magnification ratio of 5:1 is applied to the zoom-in imaging.
{"title":"High-resolution local imaging using a micro-CT","authors":"S. Lee, M. Cho, J. Choi","doi":"10.1109/ISBI.2008.4540951","DOIUrl":"https://doi.org/10.1109/ISBI.2008.4540951","url":null,"abstract":"An X-ray micro-tomography system has been developed for in vivo small animal imaging studies. For efficient in vivo scanning, it has a rotating gantry on which the X-ray source and the flat panel X-ray detector are mounted. To reconstruct artifact-free images of a small local region inside the animal subject from the truncated projection data, the projection data from the large field of view (FOV) scan of the whole animal subject are combined with the projection data from the small FOV scan of the region of interest. For the acquisition of X-ray projection data, a 1248 times 1248 flat-panel x-ray detector with the pixel pitch of 100 mum has been used. The developed system has the spatial resolution of 12 lp/mm when the highest magnification ratio of 5:1 is applied to the zoom-in imaging.","PeriodicalId":184204,"journal":{"name":"2008 5th IEEE International Symposium on Biomedical Imaging: From Nano to Macro","volume":"41 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2008-05-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130606012","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2008-05-14DOI: 10.1109/ISBI.2008.4541037
Anders Ericsson, P. Aljabar, D. Rueckert
We present a method for the construction of patient-specific atlases of the brain. Traditional atlases of the brain aim to characterize the variability of a population of subjects. A common approach is to average the anatomies of a population after alignment to a common coordinate system. Subjects are typically given equal weights during averaging which results in atlases that are population-specific rather than subject specific. In this paper we propose a method for the construction of patient-specific atlas for a given query subject from a large population cohort. During the atlas construction we compute the similarity between the query subject and the subjects in the population cohort. This similarity measure can be based on image similarity or other meta-information (e.g. sex, age, ethnicity, medical history, etc). We show an example of the construction of brain atlases for different ages using a cohort of 575 subjects between the ages of 18 and 80.
{"title":"Construction of a patient-specific atlas of the brain: Application to normal aging","authors":"Anders Ericsson, P. Aljabar, D. Rueckert","doi":"10.1109/ISBI.2008.4541037","DOIUrl":"https://doi.org/10.1109/ISBI.2008.4541037","url":null,"abstract":"We present a method for the construction of patient-specific atlases of the brain. Traditional atlases of the brain aim to characterize the variability of a population of subjects. A common approach is to average the anatomies of a population after alignment to a common coordinate system. Subjects are typically given equal weights during averaging which results in atlases that are population-specific rather than subject specific. In this paper we propose a method for the construction of patient-specific atlas for a given query subject from a large population cohort. During the atlas construction we compute the similarity between the query subject and the subjects in the population cohort. This similarity measure can be based on image similarity or other meta-information (e.g. sex, age, ethnicity, medical history, etc). We show an example of the construction of brain atlases for different ages using a cohort of 575 subjects between the ages of 18 and 80.","PeriodicalId":184204,"journal":{"name":"2008 5th IEEE International Symposium on Biomedical Imaging: From Nano to Macro","volume":"300 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2008-05-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131014339","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2008-05-14DOI: 10.1109/ISBI.2008.4541211
R. Narayanan, D. Shen, C. Davatzikos, E. Crawford, A. Barqawi, P. Werahera, D. Kumar, J. Suri
It is widely established that prostate cancer is a multifocal disease and cancerous lesions are not uniformly distributed within the gland. Current imaging methods cannot detect prostate cancer with sufficient sensitivity and specificity, especially localized cancers. A cancer atlas was previously demonstrated. However the atlas must be registered with a patient's ultrasound image in a clinical procedure. Here we present the fast registration of this atlas in a clinical setting so as to map cancer likelihoods in addition to optimized biopsy locations from the atlas space to the subject to maximize cancer detection accuracy. The registration was validated on 158 subjects with cancers annotated and the detection rate was found to be 84.81% and 89.87% for optimized 7 and 12 core biopsy schemes respectively. It took less than 8 seconds for the entire registration procedure.
{"title":"Application and validation of registration framework for real-time atlas guided biopsy","authors":"R. Narayanan, D. Shen, C. Davatzikos, E. Crawford, A. Barqawi, P. Werahera, D. Kumar, J. Suri","doi":"10.1109/ISBI.2008.4541211","DOIUrl":"https://doi.org/10.1109/ISBI.2008.4541211","url":null,"abstract":"It is widely established that prostate cancer is a multifocal disease and cancerous lesions are not uniformly distributed within the gland. Current imaging methods cannot detect prostate cancer with sufficient sensitivity and specificity, especially localized cancers. A cancer atlas was previously demonstrated. However the atlas must be registered with a patient's ultrasound image in a clinical procedure. Here we present the fast registration of this atlas in a clinical setting so as to map cancer likelihoods in addition to optimized biopsy locations from the atlas space to the subject to maximize cancer detection accuracy. The registration was validated on 158 subjects with cancers annotated and the detection rate was found to be 84.81% and 89.87% for optimized 7 and 12 core biopsy schemes respectively. It took less than 8 seconds for the entire registration procedure.","PeriodicalId":184204,"journal":{"name":"2008 5th IEEE International Symposium on Biomedical Imaging: From Nano to Macro","volume":"25 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2008-05-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130953256","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2008-05-14DOI: 10.1109/ISBI.2008.4540995
C. Ortíz-de-Solórzano, T. Pengo, Miguel Galarraga, A. Muñoz-Barrutia
We present the hardware and software specification of a quantitative, multidimensional and multispectral microscopy system designed for the detection of lung cancer using minimal samples of bronchoalveolar lavage (BAL). BAL samples were stained using FICTION: Fluorescence Immunophenotyping and Interphase Cytogenetics as a Tool for the Investigation of Neoplasms. Our system allows preliminary immunophenotypic detection of rare cancerous candidate cells, followed by accurate three-dimensional analysis of genomic integrity, to confirm or refute the initial assessment. Our results show that our automated analysis can accurately assist a human expert in the diagnostic evaluation of BAL samples.
{"title":"Automation of the detection of lung cancer cells in minimal samples of bronchioalveolar lavage","authors":"C. Ortíz-de-Solórzano, T. Pengo, Miguel Galarraga, A. Muñoz-Barrutia","doi":"10.1109/ISBI.2008.4540995","DOIUrl":"https://doi.org/10.1109/ISBI.2008.4540995","url":null,"abstract":"We present the hardware and software specification of a quantitative, multidimensional and multispectral microscopy system designed for the detection of lung cancer using minimal samples of bronchoalveolar lavage (BAL). BAL samples were stained using FICTION: Fluorescence Immunophenotyping and Interphase Cytogenetics as a Tool for the Investigation of Neoplasms. Our system allows preliminary immunophenotypic detection of rare cancerous candidate cells, followed by accurate three-dimensional analysis of genomic integrity, to confirm or refute the initial assessment. Our results show that our automated analysis can accurately assist a human expert in the diagnostic evaluation of BAL samples.","PeriodicalId":184204,"journal":{"name":"2008 5th IEEE International Symposium on Biomedical Imaging: From Nano to Macro","volume":"123 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2008-05-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131202952","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2008-05-14DOI: 10.1109/ISBI.2008.4541315
K. Ishii, Y. Funaki, Y. Kikuch, H. Yamazaki, S. Matsuyama, A. Terakawa, M. Fujiwara, R. Iwata, T. Kodama, Yukiko Watanabe, N. Tanizaki, D. Amano, Takashi Yamaguchi
Recently, we reported an animal semiconductor PET with the spatial resolution of 0.8 mm FWHM within the central 20 mm-diameter of FOV for the purpose of biomedical study using rats and mice. This ultra high spatial resolution was obtained by the use of small CdTe elements of 1.1mm x 1.0 mm x 5 mm. The FOV of this PET is 64 mm in diameter and 26 mm in axis. We applied to observe small tumors in mouse and succeeded to 1 Q obtain [ F]FDG images of mouse mammary carcinoma of ~1mm size.
最近，我们报道了一种在视场中心直径20 mm范围内空间分辨率为0.8 mm FWHM的动物半导体PET，用于大鼠和小鼠的生物医学研究。这种超高空间分辨率是通过使用1.1mm x 1.0 mm x 5mm的小碲化镉元素获得的。该PET的视场直径为64毫米，轴长为26毫米。我们应用于小鼠小肿瘤的观察，成功获得了~1mm大小的小鼠乳腺癌的[F]FDG图像。
{"title":"FDG imaging of 1mm tumor with an ultra high resolution animal PET","authors":"K. Ishii, Y. Funaki, Y. Kikuch, H. Yamazaki, S. Matsuyama, A. Terakawa, M. Fujiwara, R. Iwata, T. Kodama, Yukiko Watanabe, N. Tanizaki, D. Amano, Takashi Yamaguchi","doi":"10.1109/ISBI.2008.4541315","DOIUrl":"https://doi.org/10.1109/ISBI.2008.4541315","url":null,"abstract":"Recently, we reported an animal semiconductor PET with the spatial resolution of 0.8 mm FWHM within the central 20 mm-diameter of FOV for the purpose of biomedical study using rats and mice. This ultra high spatial resolution was obtained by the use of small CdTe elements of 1.1mm x 1.0 mm x 5 mm. The FOV of this PET is 64 mm in diameter and 26 mm in axis. We applied to observe small tumors in mouse and succeeded to 1 Q obtain [ F]FDG images of mouse mammary carcinoma of ~1mm size.","PeriodicalId":184204,"journal":{"name":"2008 5th IEEE International Symposium on Biomedical Imaging: From Nano to Macro","volume":"16 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2008-05-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131152053","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2008-05-14DOI: 10.1109/ISBI.2008.4541304
Sang Ho Lee, Jong Hyo Kim, J. Park, J. Chang, Sang Joon Park, Y. Jung, S. Tak, W. Moon
This study introduces a novel texture analysis scheme applied to perfusion volumes in dynamic contrast-enhanced (DCE) breast MRI to provide a method of lesion discrimination. DCE MRI was applied to 24 lesions (12 malignant, 12 benign). Automatic segmentation was performed for extraction of a lesion volume, which was divided into whole, rim and core volume partitions. Lesion perfusion volumes were classified using three-time-points (3TP) method of computer-aided diagnosis. Receiver operating characteristic curve (ROC) analysis was performed for differentiation of benign and malignant lesions using texture features of perfusion volumes classified by the 3TP method. When using the texture features of perfusion volumes divided into rim and core lesion volume, the texture features to have more improved accuracy appeared than using whole lesion volume. This result suggests that lesion classification using texture features of local perfusion volumes is helpful in selecting meaningful texture features for differentiation of benign and malignant lesions.
{"title":"Texture analysis of lesion perfusion volumes in dynamic contrast-enhanced breast MRI","authors":"Sang Ho Lee, Jong Hyo Kim, J. Park, J. Chang, Sang Joon Park, Y. Jung, S. Tak, W. Moon","doi":"10.1109/ISBI.2008.4541304","DOIUrl":"https://doi.org/10.1109/ISBI.2008.4541304","url":null,"abstract":"This study introduces a novel texture analysis scheme applied to perfusion volumes in dynamic contrast-enhanced (DCE) breast MRI to provide a method of lesion discrimination. DCE MRI was applied to 24 lesions (12 malignant, 12 benign). Automatic segmentation was performed for extraction of a lesion volume, which was divided into whole, rim and core volume partitions. Lesion perfusion volumes were classified using three-time-points (3TP) method of computer-aided diagnosis. Receiver operating characteristic curve (ROC) analysis was performed for differentiation of benign and malignant lesions using texture features of perfusion volumes classified by the 3TP method. When using the texture features of perfusion volumes divided into rim and core lesion volume, the texture features to have more improved accuracy appeared than using whole lesion volume. This result suggests that lesion classification using texture features of local perfusion volumes is helpful in selecting meaningful texture features for differentiation of benign and malignant lesions.","PeriodicalId":184204,"journal":{"name":"2008 5th IEEE International Symposium on Biomedical Imaging: From Nano to Macro","volume":"5 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2008-05-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128281847","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2008-05-14DOI: 10.1109/ISBI.2008.4541159
J. Rose, C. Revol-Muller, J. Langlois, M. Janier, C. Odet
We propose an automated region growing integrating adaptive shape prior in order to segment biomedical images. In our work, the segmentation method is improved by taking into account a shape reference model by non-linear way. Thus, the proposed method is driven by statistical data computed from the evolving region and by a priori shape information given by the model. An improvement of the method is proposed by adapting automatically the degree of integration of shape prior for each pixel of the image. The proposed method was applied for segmenting 3D micro-CT image of mouse skull in the framework of small animal imaging. The method gives promising results and appears to be well adapted to the context.
{"title":"3D region growing integrating adaptive shape prior","authors":"J. Rose, C. Revol-Muller, J. Langlois, M. Janier, C. Odet","doi":"10.1109/ISBI.2008.4541159","DOIUrl":"https://doi.org/10.1109/ISBI.2008.4541159","url":null,"abstract":"We propose an automated region growing integrating adaptive shape prior in order to segment biomedical images. In our work, the segmentation method is improved by taking into account a shape reference model by non-linear way. Thus, the proposed method is driven by statistical data computed from the evolving region and by a priori shape information given by the model. An improvement of the method is proposed by adapting automatically the degree of integration of shape prior for each pixel of the image. The proposed method was applied for segmenting 3D micro-CT image of mouse skull in the framework of small animal imaging. The method gives promising results and appears to be well adapted to the context.","PeriodicalId":184204,"journal":{"name":"2008 5th IEEE International Symposium on Biomedical Imaging: From Nano to Macro","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2008-05-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128971442","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2008-05-14DOI: 10.1109/ISBI.2008.4541148
R. Neji, G. Fleury, J. Deux, A. Rahmouni, G. Bassez, A. Vignaud, N. Paragios
In this paper we propose a classification-based method towards the segmentation of diffusion tensor images. We use support vector machines to classify diffusion tensors and we extend linear classification to the non linear case. To this end, we discuss and evaluate three different classes of kernels on the space of symmetric definite positive matrices that are well suited for the classification of tensor data. We impose spatial constraints by means of a Markov random field model that takes into account the result of SVM classification. Experimental results are provided for diffusion tensor images of human skeletal muscles. They demonstrate the potential of our method in discriminating the different muscle groups.
{"title":"Support vector driven Markov random fields towards DTI segmentation of the human skeletal muscle","authors":"R. Neji, G. Fleury, J. Deux, A. Rahmouni, G. Bassez, A. Vignaud, N. Paragios","doi":"10.1109/ISBI.2008.4541148","DOIUrl":"https://doi.org/10.1109/ISBI.2008.4541148","url":null,"abstract":"In this paper we propose a classification-based method towards the segmentation of diffusion tensor images. We use support vector machines to classify diffusion tensors and we extend linear classification to the non linear case. To this end, we discuss and evaluate three different classes of kernels on the space of symmetric definite positive matrices that are well suited for the classification of tensor data. We impose spatial constraints by means of a Markov random field model that takes into account the result of SVM classification. Experimental results are provided for diffusion tensor images of human skeletal muscles. They demonstrate the potential of our method in discriminating the different muscle groups.","PeriodicalId":184204,"journal":{"name":"2008 5th IEEE International Symposium on Biomedical Imaging: From Nano to Macro","volume":"35 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2008-05-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128818442","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：