Pub Date : 2018-09-01DOI: 10.23919/SPA.2018.8563393
Hugo Cordeiro, C. Meneses
This paper describes the impact of the signal bandwidth reduction in the identification of voice pathologies. The implemented systems evaluate the identification of 3 classes divided by healthy subjects, subjects diagnosed with physiological larynx pathologies and subjects diagnosed with neuromuscular larynx pathologies. Continuous speech signals are down-sampled to 4 kHz and the extracted spectral parameters are applied to a GMM classifier. No significant change in accuracy occurs, being possible to conclude that the low frequencies contain sufficient information to allow the classification of pathologies. A second objective is to test the effects of suppressing the voice activity detection and the increasing the analysis window length. In both cases the accuracy increases. In conclusion, a pathological voice identification system based on signals sampled at 4 kHz, without voice activity detection and with an analysis window length of 40 ms is proposed, getting 81.8% accuracy. The proposed system has also the advantage of reduces the storage memory and the processing time.
{"title":"Low band continuous speech system for voice pathologies identification","authors":"Hugo Cordeiro, C. Meneses","doi":"10.23919/SPA.2018.8563393","DOIUrl":"https://doi.org/10.23919/SPA.2018.8563393","url":null,"abstract":"This paper describes the impact of the signal bandwidth reduction in the identification of voice pathologies. The implemented systems evaluate the identification of 3 classes divided by healthy subjects, subjects diagnosed with physiological larynx pathologies and subjects diagnosed with neuromuscular larynx pathologies. Continuous speech signals are down-sampled to 4 kHz and the extracted spectral parameters are applied to a GMM classifier. No significant change in accuracy occurs, being possible to conclude that the low frequencies contain sufficient information to allow the classification of pathologies. A second objective is to test the effects of suppressing the voice activity detection and the increasing the analysis window length. In both cases the accuracy increases. In conclusion, a pathological voice identification system based on signals sampled at 4 kHz, without voice activity detection and with an analysis window length of 40 ms is proposed, getting 81.8% accuracy. The proposed system has also the advantage of reduces the storage memory and the processing time.","PeriodicalId":265587,"journal":{"name":"2018 Signal Processing: Algorithms, Architectures, Arrangements, and Applications (SPA)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2018-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123657369","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 : 2018-09-01DOI: 10.23919/SPA.2018.8563415
Faezeh Fallah, Bin Yang, S. Walter, F. Bamberg
In this paper, we proposed a deformation-Iregistration-free method for multilabel segmentation of fat-water MR images without need to prior localization or geometry estimation. This method employed a multiresolution (hierarchical) feature- and prior-based Random Walker graph and a hierarchical conditional random field (HCRF). To incorporate both aspatial (intra-patch) and spatial (inter-patch neighborhood) information into the image segmentation, the proposed random walker graph was made of a multiresolution spatial and a multiresolution aspatial (prior-based) sub-graph. Edge weights and prior probabilities of this graph as well as the energy terms of the HCRF were determined by a hierarchical random decision forest classifier. This classifier was trained using multiscale local and contextual features extracted from fat-water (2-channel) magnetic resonance (MR) images. The proposed method was trained and evaluated for simultaneous volumetric segmentation of vertebral bodies and intervertebral discs on fat-water MR images. These evaluations revealed its comparable accuracy to the state-of-the-art while demanding less computations and training data. The proposed method was, however, generic and extendible for segmenting any kind of tissues on other multichannel images.
{"title":"Hierarchical Feature-learning Graph-based Segmentation of Fat-Water MR Images","authors":"Faezeh Fallah, Bin Yang, S. Walter, F. Bamberg","doi":"10.23919/SPA.2018.8563415","DOIUrl":"https://doi.org/10.23919/SPA.2018.8563415","url":null,"abstract":"In this paper, we proposed a deformation-Iregistration-free method for multilabel segmentation of fat-water MR images without need to prior localization or geometry estimation. This method employed a multiresolution (hierarchical) feature- and prior-based Random Walker graph and a hierarchical conditional random field (HCRF). To incorporate both aspatial (intra-patch) and spatial (inter-patch neighborhood) information into the image segmentation, the proposed random walker graph was made of a multiresolution spatial and a multiresolution aspatial (prior-based) sub-graph. Edge weights and prior probabilities of this graph as well as the energy terms of the HCRF were determined by a hierarchical random decision forest classifier. This classifier was trained using multiscale local and contextual features extracted from fat-water (2-channel) magnetic resonance (MR) images. The proposed method was trained and evaluated for simultaneous volumetric segmentation of vertebral bodies and intervertebral discs on fat-water MR images. These evaluations revealed its comparable accuracy to the state-of-the-art while demanding less computations and training data. The proposed method was, however, generic and extendible for segmenting any kind of tissues on other multichannel images.","PeriodicalId":265587,"journal":{"name":"2018 Signal Processing: Algorithms, Architectures, Arrangements, and Applications (SPA)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2018-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125517478","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 : 2018-09-01DOI: 10.23919/SPA.2018.8563339
A. Klepaczko, P. Skulimowski, M. Strzelecki, L. Stefanczyk, E. Eikefjord, J. Rørvik, A. Lundervold
Magnetic resonance (MR) simulation is one of the possible approaches to test and develop new imaging protocols. It can assist in fast, on-demand verification of various hypotheses concerning the impact of different physical and/or technical factors on image appearance. In this paper, we perform numerical simulation of dynamic contrast-enhanced MR imaging. In particular, we present the implementation of the so-called balanced steady state free precession sequence and show its application in the synthesis of DCE-MR images mimicking perfusion-weighted examinations of the kidney. To this end, we designed a simplified digital phantom of renal parenchyma comprising of kidney cortex and medulla. The phantom was constructed based on manual segmentation of a real high-resolution CT image of the abdomen. The contrast agent kinetics was incorporated into the model by assigning time-varying $T_{1}$ relaxation time to the kidney tissue segments. The relevant T1 time courses were determined based on analysis of real DCE-MR studies. Eventually, the practical aspects of the designed simulator are illustrated in an example application, where selected image-derived perfusion characteristics are referred to physiological parameters of the kidney.
{"title":"Numerical simulation of the b-SSFP sequence in MR perfusion-weighted imaging of the kidney","authors":"A. Klepaczko, P. Skulimowski, M. Strzelecki, L. Stefanczyk, E. Eikefjord, J. Rørvik, A. Lundervold","doi":"10.23919/SPA.2018.8563339","DOIUrl":"https://doi.org/10.23919/SPA.2018.8563339","url":null,"abstract":"Magnetic resonance (MR) simulation is one of the possible approaches to test and develop new imaging protocols. It can assist in fast, on-demand verification of various hypotheses concerning the impact of different physical and/or technical factors on image appearance. In this paper, we perform numerical simulation of dynamic contrast-enhanced MR imaging. In particular, we present the implementation of the so-called balanced steady state free precession sequence and show its application in the synthesis of DCE-MR images mimicking perfusion-weighted examinations of the kidney. To this end, we designed a simplified digital phantom of renal parenchyma comprising of kidney cortex and medulla. The phantom was constructed based on manual segmentation of a real high-resolution CT image of the abdomen. The contrast agent kinetics was incorporated into the model by assigning time-varying $T_{1}$ relaxation time to the kidney tissue segments. The relevant T1 time courses were determined based on analysis of real DCE-MR studies. Eventually, the practical aspects of the designed simulator are illustrated in an example application, where selected image-derived perfusion characteristics are referred to physiological parameters of the kidney.","PeriodicalId":265587,"journal":{"name":"2018 Signal Processing: Algorithms, Architectures, Arrangements, and Applications (SPA)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2018-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116243890","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 : 2018-09-01DOI: 10.23919/SPA.2018.8563402
Michał Bednarek, K. Walas
Non-rigid objects 3D reconstruction is a complex problem, which solution has many practical implications for graphics, robotics or augmented reality. One of the approaches to the problem is template-based, where the reference mesh, deformed over time, is used. To achieve improvement in the quality of reconstruction of the non-rigid object, we implemented two novel concepts - Simulated Local Deformation (SLD) and focal length optimisation. In this paper, we combine them together. SLD successfully enlarges the number of correspondences, hence improving the reconstruction process. Additionally, 3D reconstruction also depends on reprojection error. To minimise this measure, we propose to optimise focal length simultaneously with the reconstruction process. As a result, we achieved improved reconstruction when compared to the state-of-the-art solution.
{"title":"Simulated Local Deformation & Focal Length Optimisation For Improved Template-Based 3D Reconstruction of Non-Rigid Objects","authors":"Michał Bednarek, K. Walas","doi":"10.23919/SPA.2018.8563402","DOIUrl":"https://doi.org/10.23919/SPA.2018.8563402","url":null,"abstract":"Non-rigid objects 3D reconstruction is a complex problem, which solution has many practical implications for graphics, robotics or augmented reality. One of the approaches to the problem is template-based, where the reference mesh, deformed over time, is used. To achieve improvement in the quality of reconstruction of the non-rigid object, we implemented two novel concepts - Simulated Local Deformation (SLD) and focal length optimisation. In this paper, we combine them together. SLD successfully enlarges the number of correspondences, hence improving the reconstruction process. Additionally, 3D reconstruction also depends on reprojection error. To minimise this measure, we propose to optimise focal length simultaneously with the reconstruction process. As a result, we achieved improved reconstruction when compared to the state-of-the-art solution.","PeriodicalId":265587,"journal":{"name":"2018 Signal Processing: Algorithms, Architectures, Arrangements, and Applications (SPA)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2018-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129212692","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 : 2018-09-01DOI: 10.23919/SPA.2018.8563394
Fatih Serdar Sayin, Sertan Ozen, U. Baspinar
Cyber physical systems are gaining more place in daily life so interaction with the machines are increasing. Hand gestures are one of the tools for interaction with the machines and human - machines interfaces. Image processing, sensor based and sEMG based methods are the most popular for hand gesture recognition. sEMG based hand gesture recognition is chosen especially for graphical controller, hand rehabilitation software development and manipulation of robotic devices etc. In this study, classification of 5 hand motion, which are hand open, hand close, cylindrical grasp, Lateral pinch(key grasp) and index finger opening, have been realized. As a classifier, Artificial Neural Network(ANN) is used. The Data used for training and validation recorded from five subjects by using MYO® armband. Mean absolute value, slope sign change, waveform length, Willison amplitude and mean frequency features are used for classification. Classification performances were evaluated for all five subject together and each subject separately. In the study, we achieved 88.4% mean classification rate by using five subject's recordings.
{"title":"Hand Gesture Recognition by Using sEMG Signals for Human Machine Interaction Applications","authors":"Fatih Serdar Sayin, Sertan Ozen, U. Baspinar","doi":"10.23919/SPA.2018.8563394","DOIUrl":"https://doi.org/10.23919/SPA.2018.8563394","url":null,"abstract":"Cyber physical systems are gaining more place in daily life so interaction with the machines are increasing. Hand gestures are one of the tools for interaction with the machines and human - machines interfaces. Image processing, sensor based and sEMG based methods are the most popular for hand gesture recognition. sEMG based hand gesture recognition is chosen especially for graphical controller, hand rehabilitation software development and manipulation of robotic devices etc. In this study, classification of 5 hand motion, which are hand open, hand close, cylindrical grasp, Lateral pinch(key grasp) and index finger opening, have been realized. As a classifier, Artificial Neural Network(ANN) is used. The Data used for training and validation recorded from five subjects by using MYO® armband. Mean absolute value, slope sign change, waveform length, Willison amplitude and mean frequency features are used for classification. Classification performances were evaluated for all five subject together and each subject separately. In the study, we achieved 88.4% mean classification rate by using five subject's recordings.","PeriodicalId":265587,"journal":{"name":"2018 Signal Processing: Algorithms, Architectures, Arrangements, and Applications (SPA)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2018-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127380395","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 : 2018-09-01DOI: 10.23919/SPA.2018.8563427
P. Pawlowski, Karol Piniarski, A. Dabrowski
In this paper we present tests of the state-of-the-art lossy and lossless video codecs in order to show how to select best codecs for the advanced driver-assistance systems (ADAS). Additionally, we present how to properly adjust settings of lossy codecs for ADAS application to ensure high and constant quality of the compressed video.
{"title":"Selection and tests of lossless and lossy video codecs for advanced driver-assistance systems","authors":"P. Pawlowski, Karol Piniarski, A. Dabrowski","doi":"10.23919/SPA.2018.8563427","DOIUrl":"https://doi.org/10.23919/SPA.2018.8563427","url":null,"abstract":"In this paper we present tests of the state-of-the-art lossy and lossless video codecs in order to show how to select best codecs for the advanced driver-assistance systems (ADAS). Additionally, we present how to properly adjust settings of lossy codecs for ADAS application to ensure high and constant quality of the compressed video.","PeriodicalId":265587,"journal":{"name":"2018 Signal Processing: Algorithms, Architectures, Arrangements, and Applications (SPA)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2018-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130160531","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 : 2018-09-01DOI: 10.23919/SPA.2018.8563428
Damian Koszewski, B. Kostek
The aim of this paper is two-fold. Firstly, we attempt to check whether objective, low-level audio descriptors may serve as a comparison tool in music mix evaluation performed using different Digital Audio Workstations (DAWs). Secondly, we seek to answer whether differences in music mixes are objectively discernible when several sound processing engines of DAWs are used. The same tracks of a song exported from different Digital Audio Workstations constitute the basis for this research study. Several song mixes are built of 24 individual tracks with no added effects, employing both commercial and non-commercial DAWs. Then, a set of time- and frequency-domain audio descriptors is calculated to find similarities and differences between the music mixes. Informal listening tests are conducted to answer to what extent experts are able to evaluate differences in these mixes. Then data are analyzed to show that in most cases similar results are obtained regardless of the DAW employed.
{"title":"Low-level audio descriptors-based analysis of music mixes from different Digital Audio Workstations – case study","authors":"Damian Koszewski, B. Kostek","doi":"10.23919/SPA.2018.8563428","DOIUrl":"https://doi.org/10.23919/SPA.2018.8563428","url":null,"abstract":"The aim of this paper is two-fold. Firstly, we attempt to check whether objective, low-level audio descriptors may serve as a comparison tool in music mix evaluation performed using different Digital Audio Workstations (DAWs). Secondly, we seek to answer whether differences in music mixes are objectively discernible when several sound processing engines of DAWs are used. The same tracks of a song exported from different Digital Audio Workstations constitute the basis for this research study. Several song mixes are built of 24 individual tracks with no added effects, employing both commercial and non-commercial DAWs. Then, a set of time- and frequency-domain audio descriptors is calculated to find similarities and differences between the music mixes. Informal listening tests are conducted to answer to what extent experts are able to evaluate differences in these mixes. Then data are analyzed to show that in most cases similar results are obtained regardless of the DAW employed.","PeriodicalId":265587,"journal":{"name":"2018 Signal Processing: Algorithms, Architectures, Arrangements, and Applications (SPA)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2018-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130813529","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 : 2018-09-01DOI: 10.23919/SPA.2018.8563420
Baixiang Zhao, J. Soraghan, G. D. Caterina, L. Petropoulakis, D. Grose, T. Doshi
A novel algorithm for automatic 3D segmentation of magnetic resonance imaging (MRI) data for detection of head and neck cancerous lymph nodes (LN)) is presented in this paper. The proposed algorithm pre-processes the MRI data slices to enhance quality and reduce artefacts. A modified Fuzzy c-mean process is performed through all slices, followed by a probability map which refines the clustering results, to detect the approximate position of cancerous lymph nodes. Fourier interpolation is applied to create an isotropic 3D MRI volume. A new 3D level set method segments the tumour from the interpolated MRI volume. The proposed algorithm is tested on synthetic and real MRI data. The results show that the novel cancerous lymph nodes 3D volume extraction algorithm has over 0.9 Dice similarity score on synthetic data and 0.7 on real MRI data. The F-measure is 0.92 on synthetic data and 0.75 on real data.
{"title":"Automatic 3D segmentation of MRI data for detection of head and neck cancerous lymph nodes","authors":"Baixiang Zhao, J. Soraghan, G. D. Caterina, L. Petropoulakis, D. Grose, T. Doshi","doi":"10.23919/SPA.2018.8563420","DOIUrl":"https://doi.org/10.23919/SPA.2018.8563420","url":null,"abstract":"A novel algorithm for automatic 3D segmentation of magnetic resonance imaging (MRI) data for detection of head and neck cancerous lymph nodes (LN)) is presented in this paper. The proposed algorithm pre-processes the MRI data slices to enhance quality and reduce artefacts. A modified Fuzzy c-mean process is performed through all slices, followed by a probability map which refines the clustering results, to detect the approximate position of cancerous lymph nodes. Fourier interpolation is applied to create an isotropic 3D MRI volume. A new 3D level set method segments the tumour from the interpolated MRI volume. The proposed algorithm is tested on synthetic and real MRI data. The results show that the novel cancerous lymph nodes 3D volume extraction algorithm has over 0.9 Dice similarity score on synthetic data and 0.7 on real MRI data. The F-measure is 0.92 on synthetic data and 0.75 on real data.","PeriodicalId":265587,"journal":{"name":"2018 Signal Processing: Algorithms, Architectures, Arrangements, and Applications (SPA)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2018-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114116811","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 : 2018-09-01DOI: 10.23919/SPA.2018.8563426
A. Dabrowski
Imaging technologies and techniques of the human eye are used for both biometric and medical-diagnostic applications. Among various types of the eye images the following can be distinguished: iris images, fundus images, and various optical coherence tomography (OCT) scans. Contemporary processing approaches to all of these image types are reviewed and analyzed together with a discussion of their applications. Advanced image processing methods and algorithms, including the artificial intelligence approach, developed at the Division of Signal Processing and Electronic Systems of the Pozna� University of Technology for the considered applications, are presented. The proposed solutions are characterized by a good effectiveness and accuracy in the support of appropriate biometric and clinical decisions.
{"title":"Contemporary technologies and techniques for processing of human eye images","authors":"A. Dabrowski","doi":"10.23919/SPA.2018.8563426","DOIUrl":"https://doi.org/10.23919/SPA.2018.8563426","url":null,"abstract":"Imaging technologies and techniques of the human eye are used for both biometric and medical-diagnostic applications. Among various types of the eye images the following can be distinguished: iris images, fundus images, and various optical coherence tomography (OCT) scans. Contemporary processing approaches to all of these image types are reviewed and analyzed together with a discussion of their applications. Advanced image processing methods and algorithms, including the artificial intelligence approach, developed at the Division of Signal Processing and Electronic Systems of the Pozna� University of Technology for the considered applications, are presented. The proposed solutions are characterized by a good effectiveness and accuracy in the support of appropriate biometric and clinical decisions.","PeriodicalId":265587,"journal":{"name":"2018 Signal Processing: Algorithms, Architectures, Arrangements, and Applications (SPA)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2018-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123187310","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 : 2018-09-01DOI: 10.23919/SPA.2018.8563392
Jakub Jurek, Mateusz Pelesz, Are Losnegård, L. Reisæter, A. Wojciechowski, A. Klepaczko, O. Halvorsen, C. Beisland, M. Kociński, A. Materka, J. Rørvik, A. Lundervold
Traditionally, analysis of Dynamic Contrast-Enhanced Magnetic Resonance Images (DCE MRI) requires pharmacokinetic modelling to derive quantitative physiological parameters of the tissue. Modelling, however, is a complex task and many competing models of contrast agent kinetics and tissue structure were proposed. Alternatively, raw DCE data could be analysed to find correlation with pathology in the tissue or other desired effects, for example by clustering. In this paper, we propose a new method for DCE MRI timeseries clustering. We model the data space as a Conditional Random Field (CRF) and optimize the objective function in order to find cluster labels for all timeseries. The method is unsupervised and fully automatic. We also propose a strategy to speed up the clustering process using Support Vector Machines. We demonstrate the utility of our method on two distinct problems: prostate cancer localization and healthy kidney compartment segmentation.
{"title":"CRF-Based Clustering of Pharmacokinetic Curves from Dynamic Contrast-Enhanced MR Images","authors":"Jakub Jurek, Mateusz Pelesz, Are Losnegård, L. Reisæter, A. Wojciechowski, A. Klepaczko, O. Halvorsen, C. Beisland, M. Kociński, A. Materka, J. Rørvik, A. Lundervold","doi":"10.23919/SPA.2018.8563392","DOIUrl":"https://doi.org/10.23919/SPA.2018.8563392","url":null,"abstract":"Traditionally, analysis of Dynamic Contrast-Enhanced Magnetic Resonance Images (DCE MRI) requires pharmacokinetic modelling to derive quantitative physiological parameters of the tissue. Modelling, however, is a complex task and many competing models of contrast agent kinetics and tissue structure were proposed. Alternatively, raw DCE data could be analysed to find correlation with pathology in the tissue or other desired effects, for example by clustering. In this paper, we propose a new method for DCE MRI timeseries clustering. We model the data space as a Conditional Random Field (CRF) and optimize the objective function in order to find cluster labels for all timeseries. The method is unsupervised and fully automatic. We also propose a strategy to speed up the clustering process using Support Vector Machines. We demonstrate the utility of our method on two distinct problems: prostate cancer localization and healthy kidney compartment segmentation.","PeriodicalId":265587,"journal":{"name":"2018 Signal Processing: Algorithms, Architectures, Arrangements, and Applications (SPA)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2018-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"117347829","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：