Pub Date : 2014-07-31DOI: 10.1109/ISBI.2014.6868105
G. Michelin, L. Guignard, Ulla-Maj Fiúza, G. Malandain
Image-based studies of developing organs or embryos produce a huge quantity of data. To handle such high-throughput experimental protocols, automated computer-assisted methods are highly desirable. This article aims at designing an efficient cell segmentation method from microscopic images. The proposed approach is twofold: first, cell membranes are enhanced or extracted by the means of structure-based filters, and then perceptual grouping (i.e. tensor voting) allows to correct for segmentation gaps. To decrease the computational cost of this last step, we propose different methodologies to reduce the number of voters. Assessment on real data allows us to deduce the most efficient approach.
{"title":"Embryo cell membranes reconstruction by tensor voting","authors":"G. Michelin, L. Guignard, Ulla-Maj Fiúza, G. Malandain","doi":"10.1109/ISBI.2014.6868105","DOIUrl":"https://doi.org/10.1109/ISBI.2014.6868105","url":null,"abstract":"Image-based studies of developing organs or embryos produce a huge quantity of data. To handle such high-throughput experimental protocols, automated computer-assisted methods are highly desirable. This article aims at designing an efficient cell segmentation method from microscopic images. The proposed approach is twofold: first, cell membranes are enhanced or extracted by the means of structure-based filters, and then perceptual grouping (i.e. tensor voting) allows to correct for segmentation gaps. To decrease the computational cost of this last step, we propose different methodologies to reduce the number of voters. Assessment on real data allows us to deduce the most efficient approach.","PeriodicalId":440405,"journal":{"name":"2014 IEEE 11th International Symposium on Biomedical Imaging (ISBI)","volume":"233 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2014-07-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122390464","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 : 2014-07-31DOI: 10.1109/ISBI.2014.6867956
O. Michailovich, Y. Rathi
Among the existing methods of diffusion MRI, high angular resolution diffusion imaging (HARDI) excels in its ability to resolve the complex orientations of crossing and branching neural fibre tracts in the brain. Unfortunately, a widespread integration of HARDI into clinical workflows is still hindered by a few practical obstacles, chief among which relates to prohibitively long scan times required by current implementations of this protocol. In addition, the dependency of HARDI on rapid acquisition schemes, such as single-shot echo planar imaging, imposes limitations on the maximal spatial resolution that one can attain at an acceptable level of signal-to-noise ratio. A possible solution to the problem of limited spatial resolution of HARDI could be to modify the pattern of k-space encoding so as to maximally utilize the bandwidth efficiency of frequency encoding at the expense of using a smaller number of phase encoding steps. At the same time, a substantial reduction in the total acquisition time could be achieved through a subcritical sampling in the q-space. Although both the above mechanisms are bound to yield highly incomplete data, a stable and reliable reconstruction of the associated HARDI signals is still possible to achieve within the framework of compressed sensing. To solve this problem, we introduce an efficient reconstruction procedure, whose effectiveness is demonstrated through both in silico and in vivo experiments.
{"title":"A generalized compressed sensing approach to high angular resolution diffusion imaging","authors":"O. Michailovich, Y. Rathi","doi":"10.1109/ISBI.2014.6867956","DOIUrl":"https://doi.org/10.1109/ISBI.2014.6867956","url":null,"abstract":"Among the existing methods of diffusion MRI, high angular resolution diffusion imaging (HARDI) excels in its ability to resolve the complex orientations of crossing and branching neural fibre tracts in the brain. Unfortunately, a widespread integration of HARDI into clinical workflows is still hindered by a few practical obstacles, chief among which relates to prohibitively long scan times required by current implementations of this protocol. In addition, the dependency of HARDI on rapid acquisition schemes, such as single-shot echo planar imaging, imposes limitations on the maximal spatial resolution that one can attain at an acceptable level of signal-to-noise ratio. A possible solution to the problem of limited spatial resolution of HARDI could be to modify the pattern of k-space encoding so as to maximally utilize the bandwidth efficiency of frequency encoding at the expense of using a smaller number of phase encoding steps. At the same time, a substantial reduction in the total acquisition time could be achieved through a subcritical sampling in the q-space. Although both the above mechanisms are bound to yield highly incomplete data, a stable and reliable reconstruction of the associated HARDI signals is still possible to achieve within the framework of compressed sensing. To solve this problem, we introduce an efficient reconstruction procedure, whose effectiveness is demonstrated through both in silico and in vivo experiments.","PeriodicalId":440405,"journal":{"name":"2014 IEEE 11th International Symposium on Biomedical Imaging (ISBI)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2014-07-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115168944","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 : 2014-07-31DOI: 10.1109/ISBI.2014.6868130
A. Giusti, Claudio Caccia, D. Ciresan, J. Schmidhuber, L. Gambardella
We consider the problem of detecting mitotic figures in breast cancer histology slides. We investigate whether the performance of state-of-the-art detection algorithms is comparable to the performance of humans, when they are compared under fair conditions: our test subjects were not previously exposed to the task, and were required to learn their own classification criteria solely by studying the same training set available to algorithms. We designed and implemented a standardized web-based test based on the publicly-available MITOS dataset, and compared results with the performance of the 6 top-scoring algorithms in the ICPR 2012 Mitosis Detection Contest. The problem is presented as a classification task on a balanced dataset. 45 different test subjects produced a total of 3009 classifications. The best individual (accuracy = 0.859 ± 0.012), is outperformed by the most accurate algorithm (accuracy = 0.873 ± 0.004). This suggests that state-of-the-art detection algorithms are likely limited by the size of the training set, rather than by lack of generalization ability.
{"title":"A comparison of algorithms and humans for mitosis detection","authors":"A. Giusti, Claudio Caccia, D. Ciresan, J. Schmidhuber, L. Gambardella","doi":"10.1109/ISBI.2014.6868130","DOIUrl":"https://doi.org/10.1109/ISBI.2014.6868130","url":null,"abstract":"We consider the problem of detecting mitotic figures in breast cancer histology slides. We investigate whether the performance of state-of-the-art detection algorithms is comparable to the performance of humans, when they are compared under fair conditions: our test subjects were not previously exposed to the task, and were required to learn their own classification criteria solely by studying the same training set available to algorithms. We designed and implemented a standardized web-based test based on the publicly-available MITOS dataset, and compared results with the performance of the 6 top-scoring algorithms in the ICPR 2012 Mitosis Detection Contest. The problem is presented as a classification task on a balanced dataset. 45 different test subjects produced a total of 3009 classifications. The best individual (accuracy = 0.859 ± 0.012), is outperformed by the most accurate algorithm (accuracy = 0.873 ± 0.004). This suggests that state-of-the-art detection algorithms are likely limited by the size of the training set, rather than by lack of generalization ability.","PeriodicalId":440405,"journal":{"name":"2014 IEEE 11th International Symposium on Biomedical Imaging (ISBI)","volume":"32 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2014-07-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131814664","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 : 2014-07-31DOI: 10.1109/ISBI.2014.6868133
F. S. Nezhad, H. S. Rad, H. Soltanian-Zadeh
Lack of anatomical details in diffusion weighted magnetic resonance images limits their utilization and treatment response monitoring, shadowing the useful information they contain. Contemporary methods of utilizing these images are based on manual selection of region of interest, raising concerns about susceptibility of manual ROI placement to human errors, and limiting the investigation in specific spatial regions. In contrary to the whole body bone marrow segmentation with the luxury to include all the diseased bone marrow, high profile analysis could be applied. In this paper, we propose an automatic method for segmentation of pelvic bone with possible bone metastasis in apparent diffusion coefficient (ADC) maps. This method is a multi-parametric registration-segmentation method, taking advantage of prior information of the pelvic anatomy. Intensity inhomogeneity in the bone structure caused by bone marrow metastasis challenges the segmentation process on anatomical MR images. Specifically, we first build a probability map which provides shape and volume constraints for the segmentation. Then, T1-weighted MR images are rigidly registered to the probability map, and then the registered T1-weighted image is non-rigidly registered to its' corresponding ADC maps. Finally, the probability map is coupled with a local level set framework for automatic pelvic bone segmentation of the T1-weighted images. The segmented bone is used as a mask on the ADC map. The method is validated on 10 pairs of ADC/T1 images of breast cancer with bone marrow metastases patients. Both quantitative and qualitative evaluation results demonstrate the validity of the proposed method.
{"title":"Segmentation of bone from ADC maps in pelvis area using local level-set and prior information","authors":"F. S. Nezhad, H. S. Rad, H. Soltanian-Zadeh","doi":"10.1109/ISBI.2014.6868133","DOIUrl":"https://doi.org/10.1109/ISBI.2014.6868133","url":null,"abstract":"Lack of anatomical details in diffusion weighted magnetic resonance images limits their utilization and treatment response monitoring, shadowing the useful information they contain. Contemporary methods of utilizing these images are based on manual selection of region of interest, raising concerns about susceptibility of manual ROI placement to human errors, and limiting the investigation in specific spatial regions. In contrary to the whole body bone marrow segmentation with the luxury to include all the diseased bone marrow, high profile analysis could be applied. In this paper, we propose an automatic method for segmentation of pelvic bone with possible bone metastasis in apparent diffusion coefficient (ADC) maps. This method is a multi-parametric registration-segmentation method, taking advantage of prior information of the pelvic anatomy. Intensity inhomogeneity in the bone structure caused by bone marrow metastasis challenges the segmentation process on anatomical MR images. Specifically, we first build a probability map which provides shape and volume constraints for the segmentation. Then, T1-weighted MR images are rigidly registered to the probability map, and then the registered T1-weighted image is non-rigidly registered to its' corresponding ADC maps. Finally, the probability map is coupled with a local level set framework for automatic pelvic bone segmentation of the T1-weighted images. The segmented bone is used as a mask on the ADC map. The method is validated on 10 pairs of ADC/T1 images of breast cancer with bone marrow metastases patients. Both quantitative and qualitative evaluation results demonstrate the validity of the proposed method.","PeriodicalId":440405,"journal":{"name":"2014 IEEE 11th International Symposium on Biomedical Imaging (ISBI)","volume":"188 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2014-07-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134253339","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 : 2014-07-31DOI: 10.1109/ISBI.2014.6867852
L. Zhan, N. Jahanshad, Yan Jin, T. Nir, Cassandra D. Leonardo, M. Bernstein, B. Borowski, C. Jack, P. Thompson
Large multi-site studies, such as the Alzheimer's disease Neuroimaging Initiative (ADNI) are designed to harmonize imaging protocols as far as possible across scanning sites. ADNI-2 collects diffusion-weighted images (DWI) at 14 sites, with a consistent scanner manufacturer (General Electric), magnetic field strength (3T) and consistent acquisition parameters - including voxel size and the number of gradient directions. Here we studied how the SNR, voxel-wise and ROI-based diffusion measures, and derived connectivity matrices and network properties depended on the scanner platform (with "HD" denoting version 16.x software and lower and DV being 20.x and higher). We found scanner platform effects on voxel-based FA, in several ROIs, but not on SNR or network properties. These results indicate the importance of accounting for any differences in scanner platform in multi-site DTI studies, even when the protocols are harmonized in all other respects.
{"title":"Understanding scanner upgrade effects on brain integrity & connectivity measures","authors":"L. Zhan, N. Jahanshad, Yan Jin, T. Nir, Cassandra D. Leonardo, M. Bernstein, B. Borowski, C. Jack, P. Thompson","doi":"10.1109/ISBI.2014.6867852","DOIUrl":"https://doi.org/10.1109/ISBI.2014.6867852","url":null,"abstract":"Large multi-site studies, such as the Alzheimer's disease Neuroimaging Initiative (ADNI) are designed to harmonize imaging protocols as far as possible across scanning sites. ADNI-2 collects diffusion-weighted images (DWI) at 14 sites, with a consistent scanner manufacturer (General Electric), magnetic field strength (3T) and consistent acquisition parameters - including voxel size and the number of gradient directions. Here we studied how the SNR, voxel-wise and ROI-based diffusion measures, and derived connectivity matrices and network properties depended on the scanner platform (with \"HD\" denoting version 16.x software and lower and DV being 20.x and higher). We found scanner platform effects on voxel-based FA, in several ROIs, but not on SNR or network properties. These results indicate the importance of accounting for any differences in scanner platform in multi-site DTI studies, even when the protocols are harmonized in all other respects.","PeriodicalId":440405,"journal":{"name":"2014 IEEE 11th International Symposium on Biomedical Imaging (ISBI)","volume":"66 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2014-07-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134519538","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 : 2014-07-31DOI: 10.1109/ISBI.2014.6867797
Xi Jiang, Jinglei Lv, Dajiang Zhu, Tuo Zhang, Xiang Li, Xintao Hu, Lei Guo, Tianming Liu
It is widely believed that working memory process involves large-scale functional interactions among multiple brain networks. However, network-level functional interactions across large-scale brain networks in working memory have been rarely explored yet in the literature. In this paper, we propose a novel framework for modeling network-level functional interactions in working memory based on our publicly released 358 DICCCOL landmarks. First, 14 DICCCOLs are detected as group-wise activated ROIs via GLM and compose the `basic network' of working memory. Second, the time-frequency functional interaction patterns of each pair of activated DICCCOL and other DICCCOLs are calculated using cross-wavelet transform. Third, the common functional interaction patterns and corresponding brain networks are learned via effective online dictionary learning and sparse coding methods. Experimental results showed that multiple brain networks are involved in working memory processes. More importantly, each brain network interacts with the `basic network' via a specific functionally meaningful time-frequency interaction pattern.
{"title":"Discovering network-level functional interactions from working memory fMRI data","authors":"Xi Jiang, Jinglei Lv, Dajiang Zhu, Tuo Zhang, Xiang Li, Xintao Hu, Lei Guo, Tianming Liu","doi":"10.1109/ISBI.2014.6867797","DOIUrl":"https://doi.org/10.1109/ISBI.2014.6867797","url":null,"abstract":"It is widely believed that working memory process involves large-scale functional interactions among multiple brain networks. However, network-level functional interactions across large-scale brain networks in working memory have been rarely explored yet in the literature. In this paper, we propose a novel framework for modeling network-level functional interactions in working memory based on our publicly released 358 DICCCOL landmarks. First, 14 DICCCOLs are detected as group-wise activated ROIs via GLM and compose the `basic network' of working memory. Second, the time-frequency functional interaction patterns of each pair of activated DICCCOL and other DICCCOLs are calculated using cross-wavelet transform. Third, the common functional interaction patterns and corresponding brain networks are learned via effective online dictionary learning and sparse coding methods. Experimental results showed that multiple brain networks are involved in working memory processes. More importantly, each brain network interacts with the `basic network' via a specific functionally meaningful time-frequency interaction pattern.","PeriodicalId":440405,"journal":{"name":"2014 IEEE 11th International Symposium on Biomedical Imaging (ISBI)","volume":"42 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2014-07-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114382824","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 : 2014-07-31DOI: 10.1109/ISBI.2014.6868056
Wei Li, S. Sonntag, M. Becker, N. Marx, U. Steinseifer, D. Merhof
2D color Doppler imaging (CDI) is currently the clinical standard to assess the mitral regurgitation (MR) severity. However, due to technical and operational limitations, commonly used diagnostic approaches suffer from known shortcomings: inadequate reliability, poor reproducibility and heavy user-dependency. Aiming at improving the quality of medical assessment, an efficient numerical reconstruction of color Doppler images is presented. With help of a graphical user interface (GUI), virtual CDI of different system configurations and imaging parameters was conveniently generated in a reasonable time span. The numerical reconstruction was based on experimental results and computational fluid dynamics (CFD) simulation of a flow chamber with different orifices simulating variations of mitral insufficiency. This platform can be used to validate, evaluate and further develop existing diagnostic approaches of MR.
{"title":"Efficient numerical reconstruction of color Doppler images of mitral regurgitation in vitro","authors":"Wei Li, S. Sonntag, M. Becker, N. Marx, U. Steinseifer, D. Merhof","doi":"10.1109/ISBI.2014.6868056","DOIUrl":"https://doi.org/10.1109/ISBI.2014.6868056","url":null,"abstract":"2D color Doppler imaging (CDI) is currently the clinical standard to assess the mitral regurgitation (MR) severity. However, due to technical and operational limitations, commonly used diagnostic approaches suffer from known shortcomings: inadequate reliability, poor reproducibility and heavy user-dependency. Aiming at improving the quality of medical assessment, an efficient numerical reconstruction of color Doppler images is presented. With help of a graphical user interface (GUI), virtual CDI of different system configurations and imaging parameters was conveniently generated in a reasonable time span. The numerical reconstruction was based on experimental results and computational fluid dynamics (CFD) simulation of a flow chamber with different orifices simulating variations of mitral insufficiency. This platform can be used to validate, evaluate and further develop existing diagnostic approaches of MR.","PeriodicalId":440405,"journal":{"name":"2014 IEEE 11th International Symposium on Biomedical Imaging (ISBI)","volume":"36 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2014-07-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114384222","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 : 2014-07-31DOI: 10.1109/ISBI.2014.6867930
Loizos Markides, D. Gillies
Decoding mental states from task-related fMRI data has recently been the focus of much research. Nevertheless, high levels of acquisition and physiological noise still makes inter-subject decoding a difficult and quite unstable process. Since all of the existing decoding approaches are applied on a volume-by-volume basis, it would be sensible to ensure that sudden signal changes reflect a true change of cognitive state rather than noise artefacts. Correction of the temporal signal can be achieved through temporal smoothing, which over the years has always been a debatable fMRI preprocessing step among the neuroscience community. In this paper, we present two methods for improving decoding accuracy by correcting the temporal dynamics of a number of functional regions, using parametrized temporal smoothing. We test our methods on a real fMRI dataset and we show that when temporal smoothing is applied separately in multiple scales and is both properly constrained and conditioned, it can remove sudden artefact-driven peaks and drops from the fMRI signal and thus improve the prediction accuracy of different tasks. Moreover, since our methods are performed independently from the decoding operations, they can be used in conjunction with any feature selection and classification algorithm.
{"title":"Improving brain decoding through constrained and parametrized temporal smoothing","authors":"Loizos Markides, D. Gillies","doi":"10.1109/ISBI.2014.6867930","DOIUrl":"https://doi.org/10.1109/ISBI.2014.6867930","url":null,"abstract":"Decoding mental states from task-related fMRI data has recently been the focus of much research. Nevertheless, high levels of acquisition and physiological noise still makes inter-subject decoding a difficult and quite unstable process. Since all of the existing decoding approaches are applied on a volume-by-volume basis, it would be sensible to ensure that sudden signal changes reflect a true change of cognitive state rather than noise artefacts. Correction of the temporal signal can be achieved through temporal smoothing, which over the years has always been a debatable fMRI preprocessing step among the neuroscience community. In this paper, we present two methods for improving decoding accuracy by correcting the temporal dynamics of a number of functional regions, using parametrized temporal smoothing. We test our methods on a real fMRI dataset and we show that when temporal smoothing is applied separately in multiple scales and is both properly constrained and conditioned, it can remove sudden artefact-driven peaks and drops from the fMRI signal and thus improve the prediction accuracy of different tasks. Moreover, since our methods are performed independently from the decoding operations, they can be used in conjunction with any feature selection and classification algorithm.","PeriodicalId":440405,"journal":{"name":"2014 IEEE 11th International Symposium on Biomedical Imaging (ISBI)","volume":"2012 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2014-07-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114544305","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 : 2014-07-31DOI: 10.1109/ISBI.2014.6867910
T. Becker, W. Kanje, D. Rapoport, Konstantin Thierbach, N. Scherf, Ingo Röder, A. M. Mamlouk
Mitosis detection poses a major challenge in cell tracking as mitoses are crucial events in the construction of genealogical trees. Making use of typical mitotic patterns that can be seen in phase contrast images of time lapse experiments, we propose a new benchmark data set CeTReS.B-MI consisting of mitotic and non-mitotic cells from the publicly accessible, fully labeled data set CeTReS.B. Using this data, two simple mitosis detectors (based on compactness and intensity) are used exemplarily to train, test and compare their ability to detect mitotic events. As a gold standard, we propose a linear support vector machine (SVM), which is able to separate the classes with a high accuracy (AUC=0.993). To illustrate the potential impact of a robust mitosis detection, the proposed classifiers are combined with two state of the art cell tracking algorithms. For both algorithms, performance does change when adding mitosis detection. Finally, this evaluation also emphasizes how easy implementation and comparison becomes, having suitable benchmark data at hand.
{"title":"The benchmark data SET CeTReS.B-MI for in vitro mitosis detection","authors":"T. Becker, W. Kanje, D. Rapoport, Konstantin Thierbach, N. Scherf, Ingo Röder, A. M. Mamlouk","doi":"10.1109/ISBI.2014.6867910","DOIUrl":"https://doi.org/10.1109/ISBI.2014.6867910","url":null,"abstract":"Mitosis detection poses a major challenge in cell tracking as mitoses are crucial events in the construction of genealogical trees. Making use of typical mitotic patterns that can be seen in phase contrast images of time lapse experiments, we propose a new benchmark data set CeTReS.B-MI consisting of mitotic and non-mitotic cells from the publicly accessible, fully labeled data set CeTReS.B. Using this data, two simple mitosis detectors (based on compactness and intensity) are used exemplarily to train, test and compare their ability to detect mitotic events. As a gold standard, we propose a linear support vector machine (SVM), which is able to separate the classes with a high accuracy (AUC=0.993). To illustrate the potential impact of a robust mitosis detection, the proposed classifiers are combined with two state of the art cell tracking algorithms. For both algorithms, performance does change when adding mitosis detection. Finally, this evaluation also emphasizes how easy implementation and comparison becomes, having suitable benchmark data at hand.","PeriodicalId":440405,"journal":{"name":"2014 IEEE 11th International Symposium on Biomedical Imaging (ISBI)","volume":"212 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2014-07-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114759310","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 : 2014-07-31DOI: 10.1109/ISBI.2014.6867978
D. Sorokin, Marco Tektonidis, K. Rohr, P. Matula
In live cell imaging it is essential to analyze the pure motion of sub-nuclear proteins without influence of the cell nucleus motion and deformation which is referred to as nucleus global motion. In this work, we propose a 2D contour-based image registration approach for compensation of the global motion of the nucleus. Compared to a previous contour-based approach, our approach employs an explicit rigid registration step to compensate the nucleus translation and rotation, it uses morphological contour matching for establishing more reliable correspondences between contours in consecutive frames, and utilizes the Navier equation for more realistically modeling the nucleus deformation. Our approach was successfully applied to real live cell microscopy image sequences and an experimental comparison with an existing contour-based registration method and an intensity-based registration method has been performed.
{"title":"Non-rigid contour-based temporal registration of 2D cell nuclei images using the Navier equation","authors":"D. Sorokin, Marco Tektonidis, K. Rohr, P. Matula","doi":"10.1109/ISBI.2014.6867978","DOIUrl":"https://doi.org/10.1109/ISBI.2014.6867978","url":null,"abstract":"In live cell imaging it is essential to analyze the pure motion of sub-nuclear proteins without influence of the cell nucleus motion and deformation which is referred to as nucleus global motion. In this work, we propose a 2D contour-based image registration approach for compensation of the global motion of the nucleus. Compared to a previous contour-based approach, our approach employs an explicit rigid registration step to compensate the nucleus translation and rotation, it uses morphological contour matching for establishing more reliable correspondences between contours in consecutive frames, and utilizes the Navier equation for more realistically modeling the nucleus deformation. Our approach was successfully applied to real live cell microscopy image sequences and an experimental comparison with an existing contour-based registration method and an intensity-based registration method has been performed.","PeriodicalId":440405,"journal":{"name":"2014 IEEE 11th International Symposium on Biomedical Imaging (ISBI)","volume":"13 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2014-07-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115171487","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：