Pub Date : 2020-09-01DOI: 10.1109/ATSIP49331.2020.9231848
M. Yahia, M. Mortula, Ameen Awwad, L. Albasha, Tarig Ali
Ultrasound imaging finds many applications in water environments. The water leakage milieu is considered in this study. However, the ultrasound images are contaminated by speckle noise which damages its quality. It complicates diagnosis and quantitative and visual measurements. In this work, the iterative minimum mean square error (IMMSE) method has been extended for de-speckling of ultrasound images. Therefore, by optimally setting the number of iterations, the IMMSE filtering approach outperforms traditional speckle filtering methods such as the mean, median and the improved Lee filters in speckle removal and edge preservation.
{"title":"Ultrasound Water Leakage Image Denoising By The Iterative Mmse Filter Abstract","authors":"M. Yahia, M. Mortula, Ameen Awwad, L. Albasha, Tarig Ali","doi":"10.1109/ATSIP49331.2020.9231848","DOIUrl":"https://doi.org/10.1109/ATSIP49331.2020.9231848","url":null,"abstract":"Ultrasound imaging finds many applications in water environments. The water leakage milieu is considered in this study. However, the ultrasound images are contaminated by speckle noise which damages its quality. It complicates diagnosis and quantitative and visual measurements. In this work, the iterative minimum mean square error (IMMSE) method has been extended for de-speckling of ultrasound images. Therefore, by optimally setting the number of iterations, the IMMSE filtering approach outperforms traditional speckle filtering methods such as the mean, median and the improved Lee filters in speckle removal and edge preservation.","PeriodicalId":384018,"journal":{"name":"2020 5th International Conference on Advanced Technologies for Signal and Image Processing (ATSIP)","volume":"45 4","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131746199","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 : 2020-09-01DOI: 10.1109/ATSIP49331.2020.9231597
Baraa Zayene, Chiraz Jlassi, N. Arous
Nowadays emotion recognition has become the most interesting topic due its important role in Human Computer Interaction (HCI). Speech emotion recognition is a part of this topic which is gaining more popularity in the last years. To recognize emotion, many methods have been developed using machine learning. In this work, we use a deep neural network which takes as input personalized features. To test our proposed system we used several databases with different languages to train and to evaluate our model.
{"title":"3D Convolutional Recurrent Global Neural Network for Speech Emotion Recognition","authors":"Baraa Zayene, Chiraz Jlassi, N. Arous","doi":"10.1109/ATSIP49331.2020.9231597","DOIUrl":"https://doi.org/10.1109/ATSIP49331.2020.9231597","url":null,"abstract":"Nowadays emotion recognition has become the most interesting topic due its important role in Human Computer Interaction (HCI). Speech emotion recognition is a part of this topic which is gaining more popularity in the last years. To recognize emotion, many methods have been developed using machine learning. In this work, we use a deep neural network which takes as input personalized features. To test our proposed system we used several databases with different languages to train and to evaluate our model.","PeriodicalId":384018,"journal":{"name":"2020 5th International Conference on Advanced Technologies for Signal and Image Processing (ATSIP)","volume":"63 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133644506","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 : 2020-09-01DOI: 10.1109/ATSIP49331.2020.9231759
Fatma Najar, S. Bourouis, M. Alshar'e, Roobaea Alroobaea, N. Bouguila, A. Al-Badi, Ines Channoufi
In this paper, we address the problem of human activities and facial expression recognition by investigating the effectiveness of Bayesian inference methods. Indeed, a novel method termed as Bayesian learning for finite multivariate generalized Gaussian mixture model is developed. The multivariate generalized Gaussian distribution is encouraged by its ability to model a large range of data and its shape flexibility. Our main contribution in this work is to develop a Markov Chain Monte Carlo within Metropolis-Hastings algorithm for proposed generative model. In this research, we tackle also some key issues related to machine learning and pattern recognition such as the statistical model’s parameters estimation. We demonstrate the merits of our developed learning framework over two challenging applications that concern human activity recognition and facial expression recognition.
{"title":"Efficient Statistical Learning Framework with Applications to Human Activity and Facial Expression Recognition","authors":"Fatma Najar, S. Bourouis, M. Alshar'e, Roobaea Alroobaea, N. Bouguila, A. Al-Badi, Ines Channoufi","doi":"10.1109/ATSIP49331.2020.9231759","DOIUrl":"https://doi.org/10.1109/ATSIP49331.2020.9231759","url":null,"abstract":"In this paper, we address the problem of human activities and facial expression recognition by investigating the effectiveness of Bayesian inference methods. Indeed, a novel method termed as Bayesian learning for finite multivariate generalized Gaussian mixture model is developed. The multivariate generalized Gaussian distribution is encouraged by its ability to model a large range of data and its shape flexibility. Our main contribution in this work is to develop a Markov Chain Monte Carlo within Metropolis-Hastings algorithm for proposed generative model. In this research, we tackle also some key issues related to machine learning and pattern recognition such as the statistical model’s parameters estimation. We demonstrate the merits of our developed learning framework over two challenging applications that concern human activity recognition and facial expression recognition.","PeriodicalId":384018,"journal":{"name":"2020 5th International Conference on Advanced Technologies for Signal and Image Processing (ATSIP)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130393663","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 : 2020-09-01DOI: 10.1109/ATSIP49331.2020.9231906
A. Daly, Hedi Yazid, B. Solaiman, N. Amara
Image registration is a crucial task in medical applications and is perceived as an optimization problem which has an important interest in clinical diagnosis. In this work, we propose an optimization strategy based on a specific design of genetic algorithm combined with the gradient descent optimizer within multi-resolution scheme. The performance of the proposed method was tested and evaluated on real multimodal registration scenarios from the Retrospective Image Registration Evaluation (RIR) database. Our method results were compared with those of existing registration methods, they are accurate and effective.
{"title":"Multimodal Medical Image Registration Based on a Hybrid Optimization Strategy","authors":"A. Daly, Hedi Yazid, B. Solaiman, N. Amara","doi":"10.1109/ATSIP49331.2020.9231906","DOIUrl":"https://doi.org/10.1109/ATSIP49331.2020.9231906","url":null,"abstract":"Image registration is a crucial task in medical applications and is perceived as an optimization problem which has an important interest in clinical diagnosis. In this work, we propose an optimization strategy based on a specific design of genetic algorithm combined with the gradient descent optimizer within multi-resolution scheme. The performance of the proposed method was tested and evaluated on real multimodal registration scenarios from the Retrospective Image Registration Evaluation (RIR) database. Our method results were compared with those of existing registration methods, they are accurate and effective.","PeriodicalId":384018,"journal":{"name":"2020 5th International Conference on Advanced Technologies for Signal and Image Processing (ATSIP)","volume":"50 19","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132389790","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 : 2020-09-01DOI: 10.1109/ATSIP49331.2020.9231765
Chaima Dachraoui, S. Labidi, A. Mouelhi
Multiple sclerosis is an inflammatory autoimmune disease that affects the central nervous system. We can consider that the Magnetic Resonance Imaging is a quantitative assessment and most objective approach for a better understanding of the pathology. Therefore MRI has emerged as a powerful tool for non-invasive diagnosis and description of the natural history of brain pathologies. A semi-automatic segmentation of multiple sclerosis lesions in brain MRI has been widely studied in recent years but in this paper we will be only limit on the automatic segmentation of these plaques disseminated in time and space. We quantitatively validate our results using data augmentation. Having a large dataset is crucial for the performance of our model. However, we can improve the performance of the model by augmenting the data that we already have.
{"title":"Computerized Image Segmentation of Multiple Sclerosis Lesions Using Fuzzy Level Set Model","authors":"Chaima Dachraoui, S. Labidi, A. Mouelhi","doi":"10.1109/ATSIP49331.2020.9231765","DOIUrl":"https://doi.org/10.1109/ATSIP49331.2020.9231765","url":null,"abstract":"Multiple sclerosis is an inflammatory autoimmune disease that affects the central nervous system. We can consider that the Magnetic Resonance Imaging is a quantitative assessment and most objective approach for a better understanding of the pathology. Therefore MRI has emerged as a powerful tool for non-invasive diagnosis and description of the natural history of brain pathologies. A semi-automatic segmentation of multiple sclerosis lesions in brain MRI has been widely studied in recent years but in this paper we will be only limit on the automatic segmentation of these plaques disseminated in time and space. We quantitatively validate our results using data augmentation. Having a large dataset is crucial for the performance of our model. However, we can improve the performance of the model by augmenting the data that we already have.","PeriodicalId":384018,"journal":{"name":"2020 5th International Conference on Advanced Technologies for Signal and Image Processing (ATSIP)","volume":"12 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123876911","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 : 2020-09-01DOI: 10.1109/ATSIP49331.2020.9231680
Aymen Mtibaa, Mohamed Amine Hmani, D. Petrovska-Delacrétaz, J. Boudy, A. Hamida, Claude Bauzou, Iacob Crucianu, I. Markopoulos, Emmanouil G Spanakis, Alexandru Nicolin, Christian Narr, M. Kockmann, Javier Pérez
Biometric recognition is nowadays widely used in different services and applications, making the user authentication easier and more secure than the traditional authentication system. Starting from this idea, the EU SpeechXRays project H2020 developed and evaluated in real-life environments a user recognition platform based on face and voice modalities. Since the proposed biometric solution was evaluated in real-life environments where biometric data recorded was not accessible because of the General Data Protection Regulation GDPR, the ground truth of the conducted evaluation was not available. To correctly report the performance evaluation, some methodologies were proposed to detect the errors caused by the absence of ground truth. This paper describes the biometric solution provided by the project and presents the biometric performance evaluation carried out in three real-life use case pilots on more than 2 000 users.
{"title":"Methodologies of Audio-Visual Biometric Performance Evaluation for the H2020 SpeechXRays Project","authors":"Aymen Mtibaa, Mohamed Amine Hmani, D. Petrovska-Delacrétaz, J. Boudy, A. Hamida, Claude Bauzou, Iacob Crucianu, I. Markopoulos, Emmanouil G Spanakis, Alexandru Nicolin, Christian Narr, M. Kockmann, Javier Pérez","doi":"10.1109/ATSIP49331.2020.9231680","DOIUrl":"https://doi.org/10.1109/ATSIP49331.2020.9231680","url":null,"abstract":"Biometric recognition is nowadays widely used in different services and applications, making the user authentication easier and more secure than the traditional authentication system. Starting from this idea, the EU SpeechXRays project H2020 developed and evaluated in real-life environments a user recognition platform based on face and voice modalities. Since the proposed biometric solution was evaluated in real-life environments where biometric data recorded was not accessible because of the General Data Protection Regulation GDPR, the ground truth of the conducted evaluation was not available. To correctly report the performance evaluation, some methodologies were proposed to detect the errors caused by the absence of ground truth. This paper describes the biometric solution provided by the project and presents the biometric performance evaluation carried out in three real-life use case pilots on more than 2 000 users.","PeriodicalId":384018,"journal":{"name":"2020 5th International Conference on Advanced Technologies for Signal and Image Processing (ATSIP)","volume":"117 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124205731","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 : 2020-09-01DOI: 10.1109/ATSIP49331.2020.9231554
S. Hemissi, A. Alotaibi, S. Alotaibi
Currently, Hyperspectral signal processing is a crucial area of research. Respectively, various techniques have been investigated to apprehend features combination and multi-label classification issues. Indeed, significant consideration has been given to approaches supporting the use of a single type of features. Moreover, few efforts have been dedicated to model the multi-label aspect of hyperspectral pixels and to integrate simultaneously divergent kinds of interdependent features. In this paper, we propose a novel embedding multi-label learning approach integrating complementary weighted features. The proposed framework combines the singular statistical characteristics of each feature to accomplish a physically meaningful cooperative low-dimensional representation of extracted features. This will grant, in one hand, the refinement of classification process and the propagation of narrow class information to unlabeled sample, in the other hand, when only partial labeling knowledge is available. This paper makes the following contributions: (i) the extraction of multi-view features based on the 3D model of the spectral signature and (ii) an embedding multi-label based approach by better tackling unbalanced and dimensionality issues. A set of complementary spatial/spectral features is extracted in the experimental section from a series of hyperspectral images. The obtained results reflect the efficiency of the proposed classification schema while maintaining a reasonable computational complexity.
{"title":"Multi-label learning embedding approach based on multi-temporal spectral signature for hyperspectral images classification","authors":"S. Hemissi, A. Alotaibi, S. Alotaibi","doi":"10.1109/ATSIP49331.2020.9231554","DOIUrl":"https://doi.org/10.1109/ATSIP49331.2020.9231554","url":null,"abstract":"Currently, Hyperspectral signal processing is a crucial area of research. Respectively, various techniques have been investigated to apprehend features combination and multi-label classification issues. Indeed, significant consideration has been given to approaches supporting the use of a single type of features. Moreover, few efforts have been dedicated to model the multi-label aspect of hyperspectral pixels and to integrate simultaneously divergent kinds of interdependent features. In this paper, we propose a novel embedding multi-label learning approach integrating complementary weighted features. The proposed framework combines the singular statistical characteristics of each feature to accomplish a physically meaningful cooperative low-dimensional representation of extracted features. This will grant, in one hand, the refinement of classification process and the propagation of narrow class information to unlabeled sample, in the other hand, when only partial labeling knowledge is available. This paper makes the following contributions: (i) the extraction of multi-view features based on the 3D model of the spectral signature and (ii) an embedding multi-label based approach by better tackling unbalanced and dimensionality issues. A set of complementary spatial/spectral features is extracted in the experimental section from a series of hyperspectral images. The obtained results reflect the efficiency of the proposed classification schema while maintaining a reasonable computational complexity.","PeriodicalId":384018,"journal":{"name":"2020 5th International Conference on Advanced Technologies for Signal and Image Processing (ATSIP)","volume":"208 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127598908","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 : 2020-09-01DOI: 10.1109/ATSIP49331.2020.9231559
R. Ayachi, Mouna Afif, Yahia Said, Abdessalem Ben Abdelaali
pedestrian detection is an important task that must be integrated into an advanced driving assisting system (ADAS). For a pedestrian detection task many rules must be respected like high performance, real-time processing, and lightweight size to fit into the embedded device of the ADAS. In this paper, we propose a pedestrian detection system based on a convolutional neural network (CNN). CNN is a deep learning model generally used for computer vision tasks like classification and detection because of its power in image processing and decision making. The proposed CNN model is named Yolov3 tiny. It was firstly used for general object detection. In this work, we applied the transfer learning technique on the proposed CNN model to make it suitable for pedestrian detection. The pedestrian detection dataset Caltech US was used to train and evaluate the proposed model. The model achieves an average precision of 76.7% and an inference time of 202 FPS.
{"title":"pedestrian detection for advanced driving assisting system: a transfer learning approach","authors":"R. Ayachi, Mouna Afif, Yahia Said, Abdessalem Ben Abdelaali","doi":"10.1109/ATSIP49331.2020.9231559","DOIUrl":"https://doi.org/10.1109/ATSIP49331.2020.9231559","url":null,"abstract":"pedestrian detection is an important task that must be integrated into an advanced driving assisting system (ADAS). For a pedestrian detection task many rules must be respected like high performance, real-time processing, and lightweight size to fit into the embedded device of the ADAS. In this paper, we propose a pedestrian detection system based on a convolutional neural network (CNN). CNN is a deep learning model generally used for computer vision tasks like classification and detection because of its power in image processing and decision making. The proposed CNN model is named Yolov3 tiny. It was firstly used for general object detection. In this work, we applied the transfer learning technique on the proposed CNN model to make it suitable for pedestrian detection. The pedestrian detection dataset Caltech US was used to train and evaluate the proposed model. The model achieves an average precision of 76.7% and an inference time of 202 FPS.","PeriodicalId":384018,"journal":{"name":"2020 5th International Conference on Advanced Technologies for Signal and Image Processing (ATSIP)","volume":"42 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114800052","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 : 2020-09-01DOI: 10.1109/ATSIP49331.2020.9231820
Marwa Hani, A. B. Slama, Nesrine Slokom, Hedi Trabelsi, I. Zghal, E. Sediki
Optical imaging for biological tissue is of great interest in the biomedical field because of its many qualities. the use of the waves between the visible and the infrared makes the optical imaging noninvasive and non-destructive. The diffusion of a large part of the light propagating in such objects represents a huge handicap since a long time. The recent application of low coherence interferometry has allowed the photons of backscattered light to be selectively collected and amplified, giving rise to the technique of optical coherence tomography (OCT). this article shows the combinatorial role of OCT and ocular ultrasound in the diagnosis of morning glory syndrome disease. In Ophthalmology, follow-up after surgery is very important especially for people who expect a retinal discharge which is impossible with the OCT, so we will quote in this paper the role of ultrasound in surgery follow-up postoperative. in this context we talk about the importance of ocular ultrasound in ophthalmology and Comparing this one with the OCT technique.
{"title":"Combined Optical Coherence Tomography and Ultrasound assisted analysis for retinal detachment in morning glory syndrome","authors":"Marwa Hani, A. B. Slama, Nesrine Slokom, Hedi Trabelsi, I. Zghal, E. Sediki","doi":"10.1109/ATSIP49331.2020.9231820","DOIUrl":"https://doi.org/10.1109/ATSIP49331.2020.9231820","url":null,"abstract":"Optical imaging for biological tissue is of great interest in the biomedical field because of its many qualities. the use of the waves between the visible and the infrared makes the optical imaging noninvasive and non-destructive. The diffusion of a large part of the light propagating in such objects represents a huge handicap since a long time. The recent application of low coherence interferometry has allowed the photons of backscattered light to be selectively collected and amplified, giving rise to the technique of optical coherence tomography (OCT). this article shows the combinatorial role of OCT and ocular ultrasound in the diagnosis of morning glory syndrome disease. In Ophthalmology, follow-up after surgery is very important especially for people who expect a retinal discharge which is impossible with the OCT, so we will quote in this paper the role of ultrasound in surgery follow-up postoperative. in this context we talk about the importance of ocular ultrasound in ophthalmology and Comparing this one with the OCT technique.","PeriodicalId":384018,"journal":{"name":"2020 5th International Conference on Advanced Technologies for Signal and Image Processing (ATSIP)","volume":"148 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124142852","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 : 2020-09-01DOI: 10.1109/ATSIP49331.2020.9231536
Hana Rmili, B. Solaiman, A. Mouelhi, R. Doghri, S. Labidi
Microscopic examination plays a significant role in the decision making for a reliable diagnosis of digestive neuroendocrine tumors (NETs), an immunohistochemical (IHC) analysis should be conducted by pathologists in order to identify cell morphology, tissue structure, and various histological disorders. The visual and manual assessment task, performed by experts, is tedious, time-consuming, and prone to inter-observer variability. Hence, there is an urgent need for developing an automatic nuclei segmentation approach which can provide an accurate number of cancerous histological tissues and overcome the issue of overlapping cells. In the proposed study, a morphological method for microscopic image segmentation is presented, this approach is mainly based on the choice of the appropriate color space, which highlights stained cells nuclei caused by stain variability and insufficient lighting conditions. Stromal cells, that differ from tumor cells in their particular form and small size, should be removed using shape criterion. Then marker-controlled watershed technique is applied in order to reduce the over-segmentation and to detach the connected cells in the resulting images. The proposed method is compared to ground truth segmentation, the results gave a Dice score of 0.959.
显微镜检查对于消化道神经内分泌肿瘤(NETs)的可靠诊断具有重要的决策作用，病理学家应进行免疫组化(IHC)分析，以识别细胞形态、组织结构和各种组织学紊乱。由专家执行的可视化和手动评估任务是乏味、耗时的，并且容易在观察者之间发生变化。因此，迫切需要开发一种能够提供准确的癌组织数量并克服细胞重叠问题的自动细胞核分割方法。本研究提出了一种显微图像分割的形态学方法，该方法主要基于选择合适的颜色空间，突出由于染色变异性和光照条件不足导致的染色细胞核。基质细胞与肿瘤细胞的形态不同，体积小，应按形状标准切除。然后采用标记控制分水岭技术，减少过度分割，分离图像中的连通细胞。将该方法与ground truth segmentation进行比较，得到的Dice得分为0.959。
{"title":"Nuclei Segmentation Approach for Digestive Neuroendocrine Tumors Analysis Using optimized Color Space Conversion","authors":"Hana Rmili, B. Solaiman, A. Mouelhi, R. Doghri, S. Labidi","doi":"10.1109/ATSIP49331.2020.9231536","DOIUrl":"https://doi.org/10.1109/ATSIP49331.2020.9231536","url":null,"abstract":"Microscopic examination plays a significant role in the decision making for a reliable diagnosis of digestive neuroendocrine tumors (NETs), an immunohistochemical (IHC) analysis should be conducted by pathologists in order to identify cell morphology, tissue structure, and various histological disorders. The visual and manual assessment task, performed by experts, is tedious, time-consuming, and prone to inter-observer variability. Hence, there is an urgent need for developing an automatic nuclei segmentation approach which can provide an accurate number of cancerous histological tissues and overcome the issue of overlapping cells. In the proposed study, a morphological method for microscopic image segmentation is presented, this approach is mainly based on the choice of the appropriate color space, which highlights stained cells nuclei caused by stain variability and insufficient lighting conditions. Stromal cells, that differ from tumor cells in their particular form and small size, should be removed using shape criterion. Then marker-controlled watershed technique is applied in order to reduce the over-segmentation and to detach the connected cells in the resulting images. The proposed method is compared to ground truth segmentation, the results gave a Dice score of 0.959.","PeriodicalId":384018,"journal":{"name":"2020 5th International Conference on Advanced Technologies for Signal and Image Processing (ATSIP)","volume":"364 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126703725","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：