Breast cancer is the most commonly diagnosed cancer and the leading cause of cancer death among women all over the world. Content based mammographic mass retrieval can assist radiologists to retrieve biopsy-proven masses content similar with the diagnostic ones, which can help radiologists to improve the diagnostic performance. However, existing mammographic mass retrieval methods are based on single-view information although one mass has two different views in mammograms. In this paper, we propose a new multi-view based mammographic mass retrieval approach integrated with feature selection method. In our retrieval paradigm, the query example is a multi-view mass pair different from a single view mass in previous studies. Accordingly, in order to extract significant characteristics from the mass, a total of 99 handcrafted features are computed, and an optimal feature set is determined by Laplacian Score (LS) feature selection method. Initial experimental results show that the retrieval performance based on our approach is better than that based on single-view method.
{"title":"Mammographic Mass Retrieval Using Multi-view Information and Laplacian Score Feature Selection","authors":"Wei Liu, Yi-ran Wei, Cheng-qian Liu","doi":"10.1145/3399637.3399645","DOIUrl":"https://doi.org/10.1145/3399637.3399645","url":null,"abstract":"Breast cancer is the most commonly diagnosed cancer and the leading cause of cancer death among women all over the world. Content based mammographic mass retrieval can assist radiologists to retrieve biopsy-proven masses content similar with the diagnostic ones, which can help radiologists to improve the diagnostic performance. However, existing mammographic mass retrieval methods are based on single-view information although one mass has two different views in mammograms. In this paper, we propose a new multi-view based mammographic mass retrieval approach integrated with feature selection method. In our retrieval paradigm, the query example is a multi-view mass pair different from a single view mass in previous studies. Accordingly, in order to extract significant characteristics from the mass, a total of 99 handcrafted features are computed, and an optimal feature set is determined by Laplacian Score (LS) feature selection method. Initial experimental results show that the retrieval performance based on our approach is better than that based on single-view method.","PeriodicalId":248664,"journal":{"name":"Proceedings of the 2020 2nd International Conference on Intelligent Medicine and Image Processing","volume":"2 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-04-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122061954","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}
Jiaxin Sun, Xiaoxiang Huang, Yongmei Hu, Zhiping Liu
In order to accurately diagnosis the severity of heart disease, we proposed a feature selection method by fusing rough sets. We firstly use genetic algorithm and heuristic algorithm based on attribute importance to select features and get the classification accuracy by support vector machine (SVM). Then, we use the two algorithms fused with rough set to select features, and get the classification again. After comparing the classification performances which obtained respectively, we find the classification accuracy of the heuristic algorithm based on attribute importance which fused with rough set has reached 89.125%, which is very close to 90.125% of the optimal solution. The results demonstrate that our method is effective and efficient.
{"title":"A Severity Diagnosis Method for Heart Disease based on Fusion Rough Sets","authors":"Jiaxin Sun, Xiaoxiang Huang, Yongmei Hu, Zhiping Liu","doi":"10.1145/3399637.3399643","DOIUrl":"https://doi.org/10.1145/3399637.3399643","url":null,"abstract":"In order to accurately diagnosis the severity of heart disease, we proposed a feature selection method by fusing rough sets. We firstly use genetic algorithm and heuristic algorithm based on attribute importance to select features and get the classification accuracy by support vector machine (SVM). Then, we use the two algorithms fused with rough set to select features, and get the classification again. After comparing the classification performances which obtained respectively, we find the classification accuracy of the heuristic algorithm based on attribute importance which fused with rough set has reached 89.125%, which is very close to 90.125% of the optimal solution. The results demonstrate that our method is effective and efficient.","PeriodicalId":248664,"journal":{"name":"Proceedings of the 2020 2nd International Conference on Intelligent Medicine and Image Processing","volume":"36 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-04-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132850254","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}
Pathological lymph node segmentation plays an important role in clinical practice. Yet it is still a challenging problem owing to low contrast to surrounding structures. In this paper, we take a deep learning based approach for pathological lymph node segmentation task. Semantic segmentation architecture, DeepLabv3+, which has the advantage to segment objects in a multi-scale way, is adopted in this paper. Meanwhile, the focal loss function, which originally applied in object detection task to deal with the imbalance class number, is integrated into DeepLabv3+ architecture for the imbalance of voxel class between pathological lymph nodes and background. Compared to the cross entropy loss function and dice function, the focal loss function can improve the segmentation performance in terms of sensitivity and dice in the DeepLabv3+ segmentation architecture. Four-fold cross validation has been done on 63 volumes containing 214 malignant lymph nodes and the mean sensitivity of 87% and average Dice score of 75% are obtained for pathological lymph node segmentation.
{"title":"Focal Loss Function based DeepLabv3+ for Pathological Lymph Node Segmentation on PET/CT","authors":"Guoping Xu, Hanqiang Cao, Youli Dong, Chunyi Yue, Kexin Li, Yubing Tong","doi":"10.1145/3399637.3399651","DOIUrl":"https://doi.org/10.1145/3399637.3399651","url":null,"abstract":"Pathological lymph node segmentation plays an important role in clinical practice. Yet it is still a challenging problem owing to low contrast to surrounding structures. In this paper, we take a deep learning based approach for pathological lymph node segmentation task. Semantic segmentation architecture, DeepLabv3+, which has the advantage to segment objects in a multi-scale way, is adopted in this paper. Meanwhile, the focal loss function, which originally applied in object detection task to deal with the imbalance class number, is integrated into DeepLabv3+ architecture for the imbalance of voxel class between pathological lymph nodes and background. Compared to the cross entropy loss function and dice function, the focal loss function can improve the segmentation performance in terms of sensitivity and dice in the DeepLabv3+ segmentation architecture. Four-fold cross validation has been done on 63 volumes containing 214 malignant lymph nodes and the mean sensitivity of 87% and average Dice score of 75% are obtained for pathological lymph node segmentation.","PeriodicalId":248664,"journal":{"name":"Proceedings of the 2020 2nd International Conference on Intelligent Medicine and Image Processing","volume":"15 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-04-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132145935","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}
Y. Miao, Chunxu Shen, Weili Shi, Kecheng Zhang, Zhengang Jiang, Huamin Yang
To solve the problem that the traditional convolutional neural network uses the pooling layers to reduce the image feature dimensions, which leads to information loss and affects the accuracy of right ventricular segmentation, a right ventricular segmentation method based on U-Net improved network is proposed. The dense blocks are used to combine the bottom features of the contracting path, and shortcut connections are used to connect the low-level features and high-level features on the expanding path, which increase the reusability of features. Depthwise separable weighted convolutions are used to enhance the edge detail information and improve the possibility of information reconstruction. An improved shinkage loss function is proposed to solve the problem of unbalanced positive and negative samples. Finally, RVSC-MACCAI 2012 datasets are used in the comparison experiments of different models, and the results show the effectiveness of the improved algorithm with the Dice coefficient of 0.90 and Hausdorff distance of 6.42.
{"title":"A Right Ventricle Segmentation Method based on U-Net with Weighted Convolution and Dense Connection","authors":"Y. Miao, Chunxu Shen, Weili Shi, Kecheng Zhang, Zhengang Jiang, Huamin Yang","doi":"10.1145/3399637.3399652","DOIUrl":"https://doi.org/10.1145/3399637.3399652","url":null,"abstract":"To solve the problem that the traditional convolutional neural network uses the pooling layers to reduce the image feature dimensions, which leads to information loss and affects the accuracy of right ventricular segmentation, a right ventricular segmentation method based on U-Net improved network is proposed. The dense blocks are used to combine the bottom features of the contracting path, and shortcut connections are used to connect the low-level features and high-level features on the expanding path, which increase the reusability of features. Depthwise separable weighted convolutions are used to enhance the edge detail information and improve the possibility of information reconstruction. An improved shinkage loss function is proposed to solve the problem of unbalanced positive and negative samples. Finally, RVSC-MACCAI 2012 datasets are used in the comparison experiments of different models, and the results show the effectiveness of the improved algorithm with the Dice coefficient of 0.90 and Hausdorff distance of 6.42.","PeriodicalId":248664,"journal":{"name":"Proceedings of the 2020 2nd International Conference on Intelligent Medicine and Image Processing","volume":"12 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-04-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132876825","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}
Accurate segmentation of Organs at Risk (OAR) on Computed Tomography (CT) images is a crucial step in radiotherapy treatment planning. In this paper, we propose a novel W-Net structure combining a U-Net segmentation network and an adversarial network (GAN) to reconstruct the OAR. With the reconstruction loss, W-Net can better learn effective features and get more accuracy segmentation result than U-Net. We test our method in the SegTHOR challenge which focus on 4 thoracic OAR: esophagus, heart, trachea and aorta. The average Dice Similarity Coefficient (%) of W-Net and U-Net on these 4 OAR are 80.6 versus 79.6, 93.8 versus 93.4, 88.3 versus 88.1, and 91.5 versus 90.6. The Hausdorff Distance (HD) are 0.5905 versus 0.6923, 0.2055 versus 0.2215, 0.7162 versus 0.7374, and 0.8061 versus 0.9290.
计算机断层扫描(CT)图像中危险器官(OAR)的准确分割是制定放射治疗计划的关键步骤。在本文中，我们提出了一种新的W-Net结构，结合U-Net分割网络和对抗网络(GAN)来重建桨叶。利用重构损失，W-Net可以更好地学习有效特征，得到比U-Net更准确的分割结果。我们在SegTHOR挑战中测试了我们的方法，该挑战主要针对4个胸部桨:食道、心脏、气管和主动脉。W-Net和U-Net在这4个桨上的平均骰子相似系数(%)分别为80.6比79.6、93.8比93.4、88.3比88.1、91.5比90.6。豪斯多夫距离(HD)分别为0.5905 vs 0.6923, 0.2055 vs 0.2215, 0.7162 vs 0.7374, 0.8061 vs 0.9290。
{"title":"W-net: A Network Structure for Automatic Segmentation of Organs at Risk in Thorax Computed Tomography","authors":"Wenhui Zhao, Haibin Chen, Yao Lu","doi":"10.1145/3399637.3399642","DOIUrl":"https://doi.org/10.1145/3399637.3399642","url":null,"abstract":"Accurate segmentation of Organs at Risk (OAR) on Computed Tomography (CT) images is a crucial step in radiotherapy treatment planning. In this paper, we propose a novel W-Net structure combining a U-Net segmentation network and an adversarial network (GAN) to reconstruct the OAR. With the reconstruction loss, W-Net can better learn effective features and get more accuracy segmentation result than U-Net. We test our method in the SegTHOR challenge which focus on 4 thoracic OAR: esophagus, heart, trachea and aorta. The average Dice Similarity Coefficient (%) of W-Net and U-Net on these 4 OAR are 80.6 versus 79.6, 93.8 versus 93.4, 88.3 versus 88.1, and 91.5 versus 90.6. The Hausdorff Distance (HD) are 0.5905 versus 0.6923, 0.2055 versus 0.2215, 0.7162 versus 0.7374, and 0.8061 versus 0.9290.","PeriodicalId":248664,"journal":{"name":"Proceedings of the 2020 2nd International Conference on Intelligent Medicine and Image Processing","volume":"47 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-04-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131570881","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}
The paramedic uses his / her potential, i.e. knowledge, experience, and abilities. They must also be able to handle medical equipment and medical devices, know how to use the forms of control and care. A person infected with a highly dangerous disease. A situation can happen. Even in Europe. How to solve patient transport, how to protect his health and how to protect others from infection - all this is dealt with by a special ambulance car, which was developed in Zlín, where the University of Zlín also cooperated. The ambulance is an integral part of the Integrated Rescue System in the event of an emergency with a high-risk infection. For example, it may be MERS, SARS or Ebola. Performance of activities concerning maintenance, care and, in particular, control of medical devices, the priority of the medical rescue service is the actual performance of the activity of the emergency medical service. Every paramedic should have an accurate idea of how to treat, care and care for a particular medical device and take care of him. The ambulance is used by health care professionals to transport patients with a risk of infection when transferred to their destination. Ambulances run emergency medical services, hospitals, the International Red Cross and many other health organizations. Special features are military or fire-fighting ambulances, special hygiene products indirectly accessible, ambulatory rooms from the driver's cab. The crew arrives at their destination where the test practitioner wears a full-body protective suit and other aids such as glasses or gloves. The transport must start according to hygienic requirements. After the transfer of a sick patient, the medical ambulance must go through disinfection. Rescuers accept the strictest hygiene regulations: they can use disposable protective equipment or two-stage respiratory protection. Crews consistently use the barrier approach, using gloves that are deployed in three layers. Protective suits, so-called overalls, loose disposable. Rescuers use respirators with an ABEK1 or higher filter and paper and carbon filtering. The rescue airways are thus protected in two stages, namely a mechanical filter that captures particles and a chemical filter. They had glasses to protect their eyes, and they also started using face shields. Upon arrival at the base, decontamination is in progress, mechanical cleaning, application of disinfectant solutions and course ozone disinfection of the room. The ambulance is disinfected after every transported patient. Rescuers are also undergoing thorough cleaning to dispose of disposable protective equipment such as bio-waste. At the exit base, ambulances that run with an infectious ambulance have their entrance and their premises, including sanitary facilities, to prevent contact with other employees. Nowadays, when people are traveling at a crossroads when from one continent, people are transferred to another continent by plane in a few hours and the infection is spreadi
{"title":"Vacuum Ambulance for Transporting Accessible Patient","authors":"Marta Blahová, M. Hromada","doi":"10.1145/3399637.3399648","DOIUrl":"https://doi.org/10.1145/3399637.3399648","url":null,"abstract":"The paramedic uses his / her potential, i.e. knowledge, experience, and abilities. They must also be able to handle medical equipment and medical devices, know how to use the forms of control and care. A person infected with a highly dangerous disease. A situation can happen. Even in Europe. How to solve patient transport, how to protect his health and how to protect others from infection - all this is dealt with by a special ambulance car, which was developed in Zlín, where the University of Zlín also cooperated. The ambulance is an integral part of the Integrated Rescue System in the event of an emergency with a high-risk infection. For example, it may be MERS, SARS or Ebola. Performance of activities concerning maintenance, care and, in particular, control of medical devices, the priority of the medical rescue service is the actual performance of the activity of the emergency medical service. Every paramedic should have an accurate idea of how to treat, care and care for a particular medical device and take care of him. The ambulance is used by health care professionals to transport patients with a risk of infection when transferred to their destination. Ambulances run emergency medical services, hospitals, the International Red Cross and many other health organizations. Special features are military or fire-fighting ambulances, special hygiene products indirectly accessible, ambulatory rooms from the driver's cab. The crew arrives at their destination where the test practitioner wears a full-body protective suit and other aids such as glasses or gloves. The transport must start according to hygienic requirements. After the transfer of a sick patient, the medical ambulance must go through disinfection. Rescuers accept the strictest hygiene regulations: they can use disposable protective equipment or two-stage respiratory protection. Crews consistently use the barrier approach, using gloves that are deployed in three layers. Protective suits, so-called overalls, loose disposable. Rescuers use respirators with an ABEK1 or higher filter and paper and carbon filtering. The rescue airways are thus protected in two stages, namely a mechanical filter that captures particles and a chemical filter. They had glasses to protect their eyes, and they also started using face shields. Upon arrival at the base, decontamination is in progress, mechanical cleaning, application of disinfectant solutions and course ozone disinfection of the room. The ambulance is disinfected after every transported patient. Rescuers are also undergoing thorough cleaning to dispose of disposable protective equipment such as bio-waste. At the exit base, ambulances that run with an infectious ambulance have their entrance and their premises, including sanitary facilities, to prevent contact with other employees. Nowadays, when people are traveling at a crossroads when from one continent, people are transferred to another continent by plane in a few hours and the infection is spreadi","PeriodicalId":248664,"journal":{"name":"Proceedings of the 2020 2nd International Conference on Intelligent Medicine and Image Processing","volume":"91 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-04-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127126107","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}
Liver cancer is one of the most common malignancies, which has extremely high mortality rate. Gene sequencing can reveal genetic variants of hepatocytes. The CDK1 gene has the potential to target anti-tumor. Therefore, the prediction of CDK1 gene mutation is of great significance for the diagnosis and treatment. In this paper, a new method for predicting CDK1 gene mutation is proposed. A novel tumor image enhancement converts the CT images into low-exposure images, high-exposure images and tumor detail-enhanced images. These images are effective to enhance interstitial and necrotic area, tumor parenchyma, tumor texture and edge features, respectively. CDK1 gene mutation prediction is modeled with deep neural network. A multi-strategy fusion loss function, which solves the imbalance of sample categories and hard samples, is used to improve the prediction performance. Comparative experiments are designed to verify the effectiveness of the proposed methods. The CDK1 gene mutation prediction after enhancement improves the accuracy of the classifier, which was 0.2 higher than others. The model with multi-strategy fusion loss function outperformed 0.116 in AUC than compared loss function. The proposed enhancement method is capable to improve the performance of classification. The multi-strategy fusion loss function comprehensively improves the indicators of the classifier.
{"title":"Liver Tumor Image Enhancement and CDK1 Gene Mutation Prediction Method","authors":"Yang Zhou, Huiyan Jiang, Yan Zhang","doi":"10.1145/3399637.3399638","DOIUrl":"https://doi.org/10.1145/3399637.3399638","url":null,"abstract":"Liver cancer is one of the most common malignancies, which has extremely high mortality rate. Gene sequencing can reveal genetic variants of hepatocytes. The CDK1 gene has the potential to target anti-tumor. Therefore, the prediction of CDK1 gene mutation is of great significance for the diagnosis and treatment. In this paper, a new method for predicting CDK1 gene mutation is proposed. A novel tumor image enhancement converts the CT images into low-exposure images, high-exposure images and tumor detail-enhanced images. These images are effective to enhance interstitial and necrotic area, tumor parenchyma, tumor texture and edge features, respectively. CDK1 gene mutation prediction is modeled with deep neural network. A multi-strategy fusion loss function, which solves the imbalance of sample categories and hard samples, is used to improve the prediction performance. Comparative experiments are designed to verify the effectiveness of the proposed methods. The CDK1 gene mutation prediction after enhancement improves the accuracy of the classifier, which was 0.2 higher than others. The model with multi-strategy fusion loss function outperformed 0.116 in AUC than compared loss function. The proposed enhancement method is capable to improve the performance of classification. The multi-strategy fusion loss function comprehensively improves the indicators of the classifier.","PeriodicalId":248664,"journal":{"name":"Proceedings of the 2020 2nd International Conference on Intelligent Medicine and Image Processing","volume":"45 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-04-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125529664","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}
In this paper, we describe the deep learning-based Bengali handwritten grapheme classification. Specifically, our recognition approach is based on the convolutional neural networks (CNNs) as deep CNNs have achieved splendid performance on many different visual recognition tasks. Moreover, we employ GridMask-based data augmentation to improve the recognition performance further. We compare the GridMask-based data augmentation with conventional data augmentations (such as flip, rotation, mixup) on three widely-used CNN architectures: ResNet101, DenseNet169 and EfficientNet B0. Extensive experiments demonstrate GridMask can utilize the information removal to improve the robustness of the neural networks, and the boost of hierarchical macro-averaged recall on the validation set suggest that GridMask data augmentation can be efficiently used for the Bengali handwritten grapheme analysis without any prior grapheme segmentation.
{"title":"GridMask Based Data Augmentation For Bengali Handwritten Grapheme Classification","authors":"Jiayu Yang","doi":"10.1145/3399637.3399650","DOIUrl":"https://doi.org/10.1145/3399637.3399650","url":null,"abstract":"In this paper, we describe the deep learning-based Bengali handwritten grapheme classification. Specifically, our recognition approach is based on the convolutional neural networks (CNNs) as deep CNNs have achieved splendid performance on many different visual recognition tasks. Moreover, we employ GridMask-based data augmentation to improve the recognition performance further. We compare the GridMask-based data augmentation with conventional data augmentations (such as flip, rotation, mixup) on three widely-used CNN architectures: ResNet101, DenseNet169 and EfficientNet B0. Extensive experiments demonstrate GridMask can utilize the information removal to improve the robustness of the neural networks, and the boost of hierarchical macro-averaged recall on the validation set suggest that GridMask data augmentation can be efficiently used for the Bengali handwritten grapheme analysis without any prior grapheme segmentation.","PeriodicalId":248664,"journal":{"name":"Proceedings of the 2020 2nd International Conference on Intelligent Medicine and Image Processing","volume":"3 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-04-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125542240","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}
This paper explores an asynchronous noise-suppression technique to be used in conjunction with asynchronous Gaussian blob tracking on dynamic vision sensor (DVS) data, specifically for space-based object tracking. The technique presented treats each sensor pixel as a spiking cell whose activity can be filtered out of the resulting sensor event stream by user-defined threshold values. In the space environment, radiation effects can introduce both transient and persistent noise into the DVS event stream. For space applications, targets of interest may be no larger than a single pixel and can be indistinguishable from sensor noise. In this paper, the asynchronous approach is experimentally compared to a conventional approach applied to reconstructed frame data for both performance and accuracy metrics. The results of this research show that the asynchronous approach can produce comparable or superior tracking accuracy while also drastically reducing the execution time of the process by seven times on average.
{"title":"Event-Based Noise Filtration with Point-of-Interest Detection and Tracking for Space Situational Awareness","authors":"Nikolaus Salvatore, A. George","doi":"10.1145/3399637.3399656","DOIUrl":"https://doi.org/10.1145/3399637.3399656","url":null,"abstract":"This paper explores an asynchronous noise-suppression technique to be used in conjunction with asynchronous Gaussian blob tracking on dynamic vision sensor (DVS) data, specifically for space-based object tracking. The technique presented treats each sensor pixel as a spiking cell whose activity can be filtered out of the resulting sensor event stream by user-defined threshold values. In the space environment, radiation effects can introduce both transient and persistent noise into the DVS event stream. For space applications, targets of interest may be no larger than a single pixel and can be indistinguishable from sensor noise. In this paper, the asynchronous approach is experimentally compared to a conventional approach applied to reconstructed frame data for both performance and accuracy metrics. The results of this research show that the asynchronous approach can produce comparable or superior tracking accuracy while also drastically reducing the execution time of the process by seven times on average.","PeriodicalId":248664,"journal":{"name":"Proceedings of the 2020 2nd International Conference on Intelligent Medicine and Image Processing","volume":"230 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-04-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132587201","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}
The near-infrared optical imaging technology based on deep learning has attracted much attention in the field of imaging reconstruction due to its small amount of calculation, fast reconstruction speed and so on. Modeling sample datasets selection are directly related to the accuracy and stability of the training model. Aiming at the influence of randomly selecting data samples on the effect of optical reconstruction based on deep learning, this paper proposes a method for selecting data samples based on equal distance cross-selection to achieve data equalization distribution. Based on the stacked auto-encoder neural network, the imaging model of 350 data samples was established, and the remaining 80 data samples were predicted. The results show that the prediction accuracy of anomaly reconstruction is 77.2% under the method of randomly selection sample datasets, while the training datasets and the prediction datasets were processed by the data equalization distribution selection method, the SAE method achieved the prediction accuracy of anomaly reconstruction of 96.25%. The method of data equalization distribution selection to collect modeling sample datasets can not only improve the accuracy of optical imaging detection effectively, but also have a certain guiding significance for the selection method of optical reconstruction sample datasets based on deep learning.
{"title":"Data Equalization Distribution Improves the Near-infrared Tissue Reconstruction based on Stacked Auto-encoder","authors":"Huiquan Wang, Tian Feng, Nian Wu","doi":"10.1145/3399637.3399649","DOIUrl":"https://doi.org/10.1145/3399637.3399649","url":null,"abstract":"The near-infrared optical imaging technology based on deep learning has attracted much attention in the field of imaging reconstruction due to its small amount of calculation, fast reconstruction speed and so on. Modeling sample datasets selection are directly related to the accuracy and stability of the training model. Aiming at the influence of randomly selecting data samples on the effect of optical reconstruction based on deep learning, this paper proposes a method for selecting data samples based on equal distance cross-selection to achieve data equalization distribution. Based on the stacked auto-encoder neural network, the imaging model of 350 data samples was established, and the remaining 80 data samples were predicted. The results show that the prediction accuracy of anomaly reconstruction is 77.2% under the method of randomly selection sample datasets, while the training datasets and the prediction datasets were processed by the data equalization distribution selection method, the SAE method achieved the prediction accuracy of anomaly reconstruction of 96.25%. The method of data equalization distribution selection to collect modeling sample datasets can not only improve the accuracy of optical imaging detection effectively, but also have a certain guiding significance for the selection method of optical reconstruction sample datasets based on deep learning.","PeriodicalId":248664,"journal":{"name":"Proceedings of the 2020 2nd International Conference on Intelligent Medicine and Image Processing","volume":"14 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-04-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126014911","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：