Pub Date : 2008-05-14DOI: 10.1109/ISBI.2008.4541002
Luke M. A. Beaumont, James Wake, eld, J. Noble
Tracking of cell populations in vitro in time lapse microscopy images enables automatic high throughput spatiotemporal measurements of a range of cell cycle mechanics and dynamics. Both in clinical and academic environments, large scale cellular data analysis using such methods stands to facilitate a paradigm shift in approaches to understanding cell biology. In this paper, we present a novel approach to cell population tracking and segmentation. We employ the CONDENSATION algorithm in tandem with Fast Levels Sets and Exclusion Zones for robust tracking and pixel-accurate segmentation. The algorithm feeds its output to a lineage filter. The complete approach is validated in terms of its ability to track and identify nuclei, and by its success in detecting abnormalities in the length of mitosis.
{"title":"Spatiotemporal Bayesian cell population tracking and analysis with lineage construction","authors":"Luke M. A. Beaumont, James Wake, eld, J. Noble","doi":"10.1109/ISBI.2008.4541002","DOIUrl":"https://doi.org/10.1109/ISBI.2008.4541002","url":null,"abstract":"Tracking of cell populations in vitro in time lapse microscopy images enables automatic high throughput spatiotemporal measurements of a range of cell cycle mechanics and dynamics. Both in clinical and academic environments, large scale cellular data analysis using such methods stands to facilitate a paradigm shift in approaches to understanding cell biology. In this paper, we present a novel approach to cell population tracking and segmentation. We employ the CONDENSATION algorithm in tandem with Fast Levels Sets and Exclusion Zones for robust tracking and pixel-accurate segmentation. The algorithm feeds its output to a lineage filter. The complete approach is validated in terms of its ability to track and identify nuclei, and by its success in detecting abnormalities in the length of mitosis.","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":"122408862","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.2008.4540958
Christina A. Hallock, I. Ozgunes, Ramamurthy Bhagavatula, G. Rohde, J. C. Crowley, C. E. Onorato, A. Mavalankar, A. Chebira, C. H. Tan, Markus Püschel, J. Kovacevic
We propose a novel method for axonal bouton modeling and automated detection in populations of labeled neurons, as well as bouton distribution analysis for the study of neural circuit organization and plasticity. Since axonal boutons are the presynaptic specializations of neural synapses, their locations can be used to determine the organization of neural circuitry, and in time-lapse studies, neural circuit dynamics. We propose simple geometric models for axonal boutons that account for variations in size, position, rotation and curvature of the axon in the vicinity of the bouton. We then use the normalized cross-correlation between the model and image data as a test statistic for bouton detection and position estimation. Thus, the problem is cast as a statistical detection problem where we can tune the algorithm parameters to maximize the probability of detection for a given probability of false alarm. For example, we can detect 81% of boutons with 9% false alarm from noisy, out of focus, images. We also present a novel method to characterize the orientation and elongation of a distribution of labeled boutons and we demonstrate its performance by applying it to a labeled data set.
{"title":"Axonal bouton modeling, detection and distribution analysis for the study of neural circuit organization and plasticity","authors":"Christina A. Hallock, I. Ozgunes, Ramamurthy Bhagavatula, G. Rohde, J. C. Crowley, C. E. Onorato, A. Mavalankar, A. Chebira, C. H. Tan, Markus Püschel, J. Kovacevic","doi":"10.1109/ISBI.2008.4540958","DOIUrl":"https://doi.org/10.1109/ISBI.2008.4540958","url":null,"abstract":"We propose a novel method for axonal bouton modeling and automated detection in populations of labeled neurons, as well as bouton distribution analysis for the study of neural circuit organization and plasticity. Since axonal boutons are the presynaptic specializations of neural synapses, their locations can be used to determine the organization of neural circuitry, and in time-lapse studies, neural circuit dynamics. We propose simple geometric models for axonal boutons that account for variations in size, position, rotation and curvature of the axon in the vicinity of the bouton. We then use the normalized cross-correlation between the model and image data as a test statistic for bouton detection and position estimation. Thus, the problem is cast as a statistical detection problem where we can tune the algorithm parameters to maximize the probability of detection for a given probability of false alarm. For example, we can detect 81% of boutons with 9% false alarm from noisy, out of focus, images. We also present a novel method to characterize the orientation and elongation of a distribution of labeled boutons and we demonstrate its performance by applying it to a labeled data set.","PeriodicalId":184204,"journal":{"name":"2008 5th IEEE International Symposium on Biomedical Imaging: From Nano to Macro","volume":"47 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":"122454812","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.2008.4541161
E. Biot, Elizabeth Crowell, H. Hofte, Y. Maurin, S. Vernhettes, P. Andrey
In confocal cellular imaging, fluorescent markers are used to label target structures. These elements of interest frequently appear as spots within a noisy background. Several algorithms have been proposed to extract spots in such conditions. However, specific methods are required when other structures are also labelled. In this paper, a new spot extraction filter is proposed to discriminate between spots and other structures using the shape of the local spatial intensity distribution. Preliminary results in the analysis of the intra-cellular distribution of the KORRIGAN1 protein in plant cells are presented. Very low detection error rates are reported.
{"title":"A new filter for spot extraction in n-dimensional biological imaging","authors":"E. Biot, Elizabeth Crowell, H. Hofte, Y. Maurin, S. Vernhettes, P. Andrey","doi":"10.1109/ISBI.2008.4541161","DOIUrl":"https://doi.org/10.1109/ISBI.2008.4541161","url":null,"abstract":"In confocal cellular imaging, fluorescent markers are used to label target structures. These elements of interest frequently appear as spots within a noisy background. Several algorithms have been proposed to extract spots in such conditions. However, specific methods are required when other structures are also labelled. In this paper, a new spot extraction filter is proposed to discriminate between spots and other structures using the shape of the local spatial intensity distribution. Preliminary results in the analysis of the intra-cellular distribution of the KORRIGAN1 protein in plant cells are presented. Very low detection error rates are reported.","PeriodicalId":184204,"journal":{"name":"2008 5th IEEE International Symposium on Biomedical Imaging: From Nano to Macro","volume":"63 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":"122830296","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.2008.4541026
N. El-Zehiry, M. Casanova, Adel Said Elmaghraby
Minicolumnar disturbance is a common feature to both dyslexic and autistic brains. This paper is motivated by the persistent need to investigate the effect of minicolumnar disturbance on the magnetic resonance images of the brain. This will serve as a preliminary step to develop a non-invasive methodology to discriminate between the diseases based on the MRI findings. In this paper, we investigate the variability of the ratio between the corpus callosum cross sectional area and the total brain intracranial volume between two groups; a group of dyslexic patients and another group of normal controls. The results show that this ratio differs significantly between the two groups and that it can be used as a discriminatory feature between dyslexic brains and typically developed ones.
{"title":"Variability of the relative corpus callosum cross sectional area between dyslexic and normally developed brains","authors":"N. El-Zehiry, M. Casanova, Adel Said Elmaghraby","doi":"10.1109/ISBI.2008.4541026","DOIUrl":"https://doi.org/10.1109/ISBI.2008.4541026","url":null,"abstract":"Minicolumnar disturbance is a common feature to both dyslexic and autistic brains. This paper is motivated by the persistent need to investigate the effect of minicolumnar disturbance on the magnetic resonance images of the brain. This will serve as a preliminary step to develop a non-invasive methodology to discriminate between the diseases based on the MRI findings. In this paper, we investigate the variability of the ratio between the corpus callosum cross sectional area and the total brain intracranial volume between two groups; a group of dyslexic patients and another group of normal controls. The results show that this ratio differs significantly between the two groups and that it can be used as a discriminatory feature between dyslexic brains and typically developed ones.","PeriodicalId":184204,"journal":{"name":"2008 5th IEEE International Symposium on Biomedical Imaging: From Nano to Macro","volume":"12 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":"114300744","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.2008.4541123
Chavdar Papazov, V. J. Dercksen, H. Lamecker, H. Hege
Image-based 3D atlases have been proven to be very useful in biological and medical research. They serve as spatial reference systems that enable researchers to integrate experimental data in a spatially coherent way and thus to relate diverse data from different experiments. Typically such atlases consist of tissue-separating surfaces. The next step are 4D atlases that provide insight into temporal development and spatio- temporal relationships. Such atlases are based on time series of 3D images and related 3D models. We present work on temporal interpolation between such 3D atlases. Due to the morphogenesis of tissues during biological development, the topology of the non-manifold surfaces may vary between subsequent time steps. For animation therefore a smooth morphing between non-manifold surfaces with different topology is needed.
{"title":"Visualizing morphogenesis and growth by temporal interpolation of surface-based 3D atlases","authors":"Chavdar Papazov, V. J. Dercksen, H. Lamecker, H. Hege","doi":"10.1109/ISBI.2008.4541123","DOIUrl":"https://doi.org/10.1109/ISBI.2008.4541123","url":null,"abstract":"Image-based 3D atlases have been proven to be very useful in biological and medical research. They serve as spatial reference systems that enable researchers to integrate experimental data in a spatially coherent way and thus to relate diverse data from different experiments. Typically such atlases consist of tissue-separating surfaces. The next step are 4D atlases that provide insight into temporal development and spatio- temporal relationships. Such atlases are based on time series of 3D images and related 3D models. We present work on temporal interpolation between such 3D atlases. Due to the morphogenesis of tissues during biological development, the topology of the non-manifold surfaces may vary between subsequent time steps. For animation therefore a smooth morphing between non-manifold surfaces with different topology is needed.","PeriodicalId":184204,"journal":{"name":"2008 5th IEEE International Symposium on Biomedical Imaging: From Nano to Macro","volume":"28 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":"114372719","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.2008.4541300
H. Ratiney, A. Bucur, M. Sdika, O. Beuf, F. Pilleul, S. Cavassila
In vivo hepatic 1H lineshapes modeled by the complex Voigt function are desirable to reduce systematic error and obtain accurate fits. However, the optimization procedure becomes challenging when the peak resonances overlap and the proportion of Gaussian to Lorentzian dampings is a priori unknown. In this context, nonlinear least-squares algorithms generally invoked in Magnetic Resonance Spectroscopy quantification are highly sensitive to the starting values and parameter bounds. To alleviate this sensitivity, multiple random starting values and parameter bounds settings are used to generate candidate solutions. The "best fit" fulfilling requirements on the cost function and damping factor final values is then selected among them. Monte Carlo studies and an in vivo hepatic 1H signal quantification demonstrated the relevance of the proposed strategy.
{"title":"Effective voigt model estimation using multiple random starting values and parameter bounds settings for in vivo hepatic 1H magnetic resonance spectroscopic data","authors":"H. Ratiney, A. Bucur, M. Sdika, O. Beuf, F. Pilleul, S. Cavassila","doi":"10.1109/ISBI.2008.4541300","DOIUrl":"https://doi.org/10.1109/ISBI.2008.4541300","url":null,"abstract":"In vivo hepatic 1H lineshapes modeled by the complex Voigt function are desirable to reduce systematic error and obtain accurate fits. However, the optimization procedure becomes challenging when the peak resonances overlap and the proportion of Gaussian to Lorentzian dampings is a priori unknown. In this context, nonlinear least-squares algorithms generally invoked in Magnetic Resonance Spectroscopy quantification are highly sensitive to the starting values and parameter bounds. To alleviate this sensitivity, multiple random starting values and parameter bounds settings are used to generate candidate solutions. The \"best fit\" fulfilling requirements on the cost function and damping factor final values is then selected among them. Monte Carlo studies and an in vivo hepatic 1H signal quantification demonstrated the relevance of the proposed strategy.","PeriodicalId":184204,"journal":{"name":"2008 5th IEEE International Symposium on Biomedical Imaging: From Nano to Macro","volume":"27 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":"114381632","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.2008.4540922
H. Necib, M. Dusart, B. Vanderlinden, I. Buvat
Patient follow-up based on PET scans is a promising approach for early assessment of tumor response and for detection of tumor recurrence. In this work, we introduce a parametric imaging method to detect and analyze the tumor changes between 2 consecutive PET scans. Fifteen pairs of consecutives PET/CT images obtained during the course of lung cancer patient monitoring were considered. For each pair, after CT- based registration of the PET images, the two PET datasets were subtracted. A biparametric graph of subtracted voxel values versus voxel values in the first scan was obtained. A model- based analysis of this graph was used to identify the tumor voxels in which significant changes occurred between the 2 scans, and yielded indices characterizing the changes. In our patients, the proposed approach correctly identified all tumor changes as confirmed using a conventional analysis. In addition, the parametric imaging approach can reveal heterogeneities in tumor response and does not require the preliminary identification of the tumors.
{"title":"Detection and characterization of the tumor change between two FDG PET scans using parametric imaging","authors":"H. Necib, M. Dusart, B. Vanderlinden, I. Buvat","doi":"10.1109/ISBI.2008.4540922","DOIUrl":"https://doi.org/10.1109/ISBI.2008.4540922","url":null,"abstract":"Patient follow-up based on PET scans is a promising approach for early assessment of tumor response and for detection of tumor recurrence. In this work, we introduce a parametric imaging method to detect and analyze the tumor changes between 2 consecutive PET scans. Fifteen pairs of consecutives PET/CT images obtained during the course of lung cancer patient monitoring were considered. For each pair, after CT- based registration of the PET images, the two PET datasets were subtracted. A biparametric graph of subtracted voxel values versus voxel values in the first scan was obtained. A model- based analysis of this graph was used to identify the tumor voxels in which significant changes occurred between the 2 scans, and yielded indices characterizing the changes. In our patients, the proposed approach correctly identified all tumor changes as confirmed using a conventional analysis. In addition, the parametric imaging approach can reveal heterogeneities in tumor response and does not require the preliminary identification of the tumors.","PeriodicalId":184204,"journal":{"name":"2008 5th IEEE International Symposium on Biomedical Imaging: From Nano to Macro","volume":"10 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":"122060578","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.2008.4541160
T. Markiewicz, C. Jochymski, R. Koktysz, W. Kozlowski
The paper presents program for automatic cell recognition and counting in selected immunohistochemical stains in the gastritis diseases. It is applied to cytoplasm reactivity markers, such as chromogranin A, serotonin and somatostatin antibodies. The program uses the sequential thresholding algorithm in combination with artificial neural network of support vector machine (SVM) type, to recognize the nuclei of the separated cells. The constructed algorithm imitates the human view of the image. The support vector machine is used for recognition of the immunoreactivity of the separated cell. The results corresponding to the exemplary images, confirm good accuracy, comparable to the human expert.
{"title":"Automatic cell recognition in immunohistochemical gastritis stains using sequential thresholding and SVM network","authors":"T. Markiewicz, C. Jochymski, R. Koktysz, W. Kozlowski","doi":"10.1109/ISBI.2008.4541160","DOIUrl":"https://doi.org/10.1109/ISBI.2008.4541160","url":null,"abstract":"The paper presents program for automatic cell recognition and counting in selected immunohistochemical stains in the gastritis diseases. It is applied to cytoplasm reactivity markers, such as chromogranin A, serotonin and somatostatin antibodies. The program uses the sequential thresholding algorithm in combination with artificial neural network of support vector machine (SVM) type, to recognize the nuclei of the separated cells. The constructed algorithm imitates the human view of the image. The support vector machine is used for recognition of the immunoreactivity of the separated cell. The results corresponding to the exemplary images, confirm good accuracy, comparable to the human expert.","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":"129911977","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.2008.4540987
Ting Chen, Xiaoxu Wang, Dimitris N. Metaxas, L. Axel
We propose a novel 3D motion estimation approach integrating the robust point matching (RPM) and meshless deformable models. In our study, we first use the Gabor filters to generate phase maps of short axis (SA) and long axis (LA) tagged MRI sequences. Then we use the RPM to track the heart motion sparsely at intersections of tag grids in these image sequences, using both intensity gradient and phase information. Next, the new meshless deformable model is used to recover the dense 3D motion of the myocardium temporally during the cardiac cycle. The deformable model is driven by external forces computed at tag intersections based on the RPM motion tracking and keeps a consistent but flexible topology during the deformation using internal constraint forces calculated by the moving least squares (MLS) method. The deformable model recovers the global deformation of the LV such as rotation, contraction and twisting by integrating global deformation parameters over the volume. The new model avoids the singularity problem of mesh-based deformable models and is capable of tracking deformation efficiently with the sparse external forces derived from tagging line intersections. We test the performance of the new approach on in vivo heart data of healthy subjects and patients. The experimental results show that our new method can fully recover the myocardium motion and strain in 3D.
{"title":"3D cardiac motion tracking using Robust Point Matching and meshless deformable models","authors":"Ting Chen, Xiaoxu Wang, Dimitris N. Metaxas, L. Axel","doi":"10.1109/ISBI.2008.4540987","DOIUrl":"https://doi.org/10.1109/ISBI.2008.4540987","url":null,"abstract":"We propose a novel 3D motion estimation approach integrating the robust point matching (RPM) and meshless deformable models. In our study, we first use the Gabor filters to generate phase maps of short axis (SA) and long axis (LA) tagged MRI sequences. Then we use the RPM to track the heart motion sparsely at intersections of tag grids in these image sequences, using both intensity gradient and phase information. Next, the new meshless deformable model is used to recover the dense 3D motion of the myocardium temporally during the cardiac cycle. The deformable model is driven by external forces computed at tag intersections based on the RPM motion tracking and keeps a consistent but flexible topology during the deformation using internal constraint forces calculated by the moving least squares (MLS) method. The deformable model recovers the global deformation of the LV such as rotation, contraction and twisting by integrating global deformation parameters over the volume. The new model avoids the singularity problem of mesh-based deformable models and is capable of tracking deformation efficiently with the sparse external forces derived from tagging line intersections. We test the performance of the new approach on in vivo heart data of healthy subjects and patients. The experimental results show that our new method can fully recover the myocardium motion and strain in 3D.","PeriodicalId":184204,"journal":{"name":"2008 5th IEEE International Symposium on Biomedical Imaging: From Nano to Macro","volume":"96 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":"121178270","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.2008.4540939
Hirotaka Susukida, Fei Ma, M. Bajger
Mammogram segmentation tasks underpin a wide range of registration, temporal analysis and detection algorithms. Unfortunately, finding an accurate, robust and efficient segmentation still remains a challenging problem in mammography. A recent segmentation technique, based on minimum spanning trees (MST segmentation), is known to be robust to typical mammogram distortions and computationally efficient. This method captures both local and global image information but the balance requires choosing a parameter. So far no automatic procedure to estimate this parameter has been proposed and the value was determined experimentally. In this paper a segmentation evaluation criterion, based on a measure of image entropy, is used to automatically optimize the granularity of an MST-based segmentation. The method is tested on a set of 82 random images taken from a commonly used mammogram database. The results show a dramatic improvement in the accuracy of a MST segmentation tuned up using the entropy-based criterion.
{"title":"Automatic tuning of a graph-based image segmentation method for digital mammography applications","authors":"Hirotaka Susukida, Fei Ma, M. Bajger","doi":"10.1109/ISBI.2008.4540939","DOIUrl":"https://doi.org/10.1109/ISBI.2008.4540939","url":null,"abstract":"Mammogram segmentation tasks underpin a wide range of registration, temporal analysis and detection algorithms. Unfortunately, finding an accurate, robust and efficient segmentation still remains a challenging problem in mammography. A recent segmentation technique, based on minimum spanning trees (MST segmentation), is known to be robust to typical mammogram distortions and computationally efficient. This method captures both local and global image information but the balance requires choosing a parameter. So far no automatic procedure to estimate this parameter has been proposed and the value was determined experimentally. In this paper a segmentation evaluation criterion, based on a measure of image entropy, is used to automatically optimize the granularity of an MST-based segmentation. The method is tested on a set of 82 random images taken from a commonly used mammogram database. The results show a dramatic improvement in the accuracy of a MST segmentation tuned up using the entropy-based criterion.","PeriodicalId":184204,"journal":{"name":"2008 5th IEEE International Symposium on Biomedical Imaging: From Nano to Macro","volume":"28 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":"114078885","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}