Pub Date : 2019-12-01DOI: 10.1109/IST48021.2019.9010104
Rui Wang, J. Zou, Runnan Qin, Liang Zhang
Addressing the problem that object instance detection has poor detection effect on occluded objects in unstructured environment when using deep learning network, we explore the use of the strategy of adversarial learning in this paper. A three-step pipeline is carried to build a novel learning framework denoted as Adversarial Generated Region-based Fully Convolutional Networks (AGR-FCN). Our method first training the noted deep model Region-based Fully Convolutional Networks (R-FCN), and then an Adversarial Mask Dropout Network (AMDN), which can generate occlusion features for training samples, is designed based on the trained R-FCN. Through the training strategy of adversarial learning between network R-FCN and network AMDN, the ability of network R-FCN to learn the features of occluded objects as well as its instance-level object detection performance is improved. Numerical experiments are conducted for instance detection to compare our proposed AGR-FCN with the original R-FCN on the self-made BHGI Database and the public database GMU Kitchen Dataset, which demonstrate that our proposed AGR-FCN outperforms original R-FCN and can achieve an average detection accuracy of nearly 90%.
{"title":"AGR-FCN: Adversarial Generated Region based on Fully Convolutional Networks for Single- and Multiple-Instance Object Detection","authors":"Rui Wang, J. Zou, Runnan Qin, Liang Zhang","doi":"10.1109/IST48021.2019.9010104","DOIUrl":"https://doi.org/10.1109/IST48021.2019.9010104","url":null,"abstract":"Addressing the problem that object instance detection has poor detection effect on occluded objects in unstructured environment when using deep learning network, we explore the use of the strategy of adversarial learning in this paper. A three-step pipeline is carried to build a novel learning framework denoted as Adversarial Generated Region-based Fully Convolutional Networks (AGR-FCN). Our method first training the noted deep model Region-based Fully Convolutional Networks (R-FCN), and then an Adversarial Mask Dropout Network (AMDN), which can generate occlusion features for training samples, is designed based on the trained R-FCN. Through the training strategy of adversarial learning between network R-FCN and network AMDN, the ability of network R-FCN to learn the features of occluded objects as well as its instance-level object detection performance is improved. Numerical experiments are conducted for instance detection to compare our proposed AGR-FCN with the original R-FCN on the self-made BHGI Database and the public database GMU Kitchen Dataset, which demonstrate that our proposed AGR-FCN outperforms original R-FCN and can achieve an average detection accuracy of nearly 90%.","PeriodicalId":117219,"journal":{"name":"2019 IEEE International Conference on Imaging Systems and Techniques (IST)","volume":"37 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121342001","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 : 2019-12-01DOI: 10.1109/IST48021.2019.9010280
M. S. Jabin, F. Magrabi, P. Hibbert, T. Schultz, W. Runciman
Beyond identifying and counting the things that go wrong, understanding how and why they go wrong requires qualitative research, especially for low-frequency events. The purpose of this study was to identify and characterize patient safety and quality issues related to health information technology (HIT) in medical imaging by collecting and analyzing incident reports through the lens of thematic analysis. In this article, we analyze 5 clusters: Staff related issues (16%), issues with diagnosis (15%), HIT incidents that involved “paper record” (12%), information and communication related (4%), and “action taken” related issues (4%). Human factors involved people failing to scan forms into the computer system (consents, requests, bookings, questionnaires, assessments, treatments and prescriptions), and another 4% involved failure to enter verbally imparted information into the system (about infectious patients, cancelled cases, and the status of reports). All of these problems had their genesis in human errors and violations. Human factors were found to cause more deleterious effects than technical factors. Of three instances of deaths caused by diagnostic issues, two were triggered by human factors, missed diagnosis. However, “staff or organizational outcome” was evenly distributed for both human and technical factors. It was therefore important to identify and characterize these incidents related to health information technology in medical imaging through the lens of thematic analysis, to provide a basis for improvements in preventing issues and improving clinical practice.
{"title":"Identifying Clusters and Themes from Incidents Related to Health Information Technology in Medical Imaging as a Basis for Improvements in Practice","authors":"M. S. Jabin, F. Magrabi, P. Hibbert, T. Schultz, W. Runciman","doi":"10.1109/IST48021.2019.9010280","DOIUrl":"https://doi.org/10.1109/IST48021.2019.9010280","url":null,"abstract":"Beyond identifying and counting the things that go wrong, understanding how and why they go wrong requires qualitative research, especially for low-frequency events. The purpose of this study was to identify and characterize patient safety and quality issues related to health information technology (HIT) in medical imaging by collecting and analyzing incident reports through the lens of thematic analysis. In this article, we analyze 5 clusters: Staff related issues (16%), issues with diagnosis (15%), HIT incidents that involved “paper record” (12%), information and communication related (4%), and “action taken” related issues (4%). Human factors involved people failing to scan forms into the computer system (consents, requests, bookings, questionnaires, assessments, treatments and prescriptions), and another 4% involved failure to enter verbally imparted information into the system (about infectious patients, cancelled cases, and the status of reports). All of these problems had their genesis in human errors and violations. Human factors were found to cause more deleterious effects than technical factors. Of three instances of deaths caused by diagnostic issues, two were triggered by human factors, missed diagnosis. However, “staff or organizational outcome” was evenly distributed for both human and technical factors. It was therefore important to identify and characterize these incidents related to health information technology in medical imaging through the lens of thematic analysis, to provide a basis for improvements in preventing issues and improving clinical practice.","PeriodicalId":117219,"journal":{"name":"2019 IEEE International Conference on Imaging Systems and Techniques (IST)","volume":"85 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116012948","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 : 2019-12-01DOI: 10.1109/IST48021.2019.9010109
M. S. Jabin, F. Magrabi, P. Hibbert, T. Schultz, W. Runciman
The Joint Commission in the United States disseminated a Sentinel Event Alert because of the number of adverse outcomes from problems with health information technology (HIT). The HITs were trading off safety and quality against throughput or efficiency. The Alert urged healthcare providers to improve process measurement and provide leadership in mitigating the risks. In order to understand what problems compromise safety and efficiency, this study has accessed, deconstructed, categorized and analyzed Australian patient safety incident reports of the things that go wrong in medical imaging, and their impact on both patients and the medical imaging acquisition and processing systems. Data Sources comprised two sets of voluntary incident reports and convenience samples of interviews with radiology staff. A special targeted search was undertaken for identifying HIT related incidents so that they could be deconstructed with the health information technology classification system. This resulted in 436 HIT related incidents. Within these incidents, 623 HIT related issues were found. These included use or human factor related issues (40%), software and hardware related issues (30%) and machine related issues (30%). Although many technical problems and deficiencies were detected in the reports identified, we did not anticipate that more than half of the incidents would have involved failures of human performance. Identifying and characterizing the things that are going wrong, related to HIT through the lens of medical imaging incident reports can provide a basis for preventing issues and improving clinical practice.
{"title":"Identifying and Classifying Incidents Related to Health Information Technology in Medical Imaging as a Basis for Improvements in Practice","authors":"M. S. Jabin, F. Magrabi, P. Hibbert, T. Schultz, W. Runciman","doi":"10.1109/IST48021.2019.9010109","DOIUrl":"https://doi.org/10.1109/IST48021.2019.9010109","url":null,"abstract":"The Joint Commission in the United States disseminated a Sentinel Event Alert because of the number of adverse outcomes from problems with health information technology (HIT). The HITs were trading off safety and quality against throughput or efficiency. The Alert urged healthcare providers to improve process measurement and provide leadership in mitigating the risks. In order to understand what problems compromise safety and efficiency, this study has accessed, deconstructed, categorized and analyzed Australian patient safety incident reports of the things that go wrong in medical imaging, and their impact on both patients and the medical imaging acquisition and processing systems. Data Sources comprised two sets of voluntary incident reports and convenience samples of interviews with radiology staff. A special targeted search was undertaken for identifying HIT related incidents so that they could be deconstructed with the health information technology classification system. This resulted in 436 HIT related incidents. Within these incidents, 623 HIT related issues were found. These included use or human factor related issues (40%), software and hardware related issues (30%) and machine related issues (30%). Although many technical problems and deficiencies were detected in the reports identified, we did not anticipate that more than half of the incidents would have involved failures of human performance. Identifying and characterizing the things that are going wrong, related to HIT through the lens of medical imaging incident reports can provide a basis for preventing issues and improving clinical practice.","PeriodicalId":117219,"journal":{"name":"2019 IEEE International Conference on Imaging Systems and Techniques (IST)","volume":"140 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122859442","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 : 2019-12-01DOI: 10.1109/IST48021.2019.9010113
V. Sakkalis, M. Pediaditis, S. Sfakianakis
The main objective of this work is the development of an integrated non-invasive surveillance system with the aim of detecting potentially pathological conditions of infants of up to one year of age that are difficult to detect without continuous monitoring by a trained professional. Conditions such as apnea or choking events can be life-threatening to an infant. In this direction, we set the overall picture of the architecture of such an infant surveillance system and build a foundation allowing for real-time video analysis supporting fast feature extraction in order to be able to detect acute episodes with no time delay or any dropped frames. Functional and technical specifications of the envisaged system, as well as the simulation results are reported.
{"title":"Towards an unobtrusive baby monitoring system for real-time detection of possibly hazardous situations*","authors":"V. Sakkalis, M. Pediaditis, S. Sfakianakis","doi":"10.1109/IST48021.2019.9010113","DOIUrl":"https://doi.org/10.1109/IST48021.2019.9010113","url":null,"abstract":"The main objective of this work is the development of an integrated non-invasive surveillance system with the aim of detecting potentially pathological conditions of infants of up to one year of age that are difficult to detect without continuous monitoring by a trained professional. Conditions such as apnea or choking events can be life-threatening to an infant. In this direction, we set the overall picture of the architecture of such an infant surveillance system and build a foundation allowing for real-time video analysis supporting fast feature extraction in order to be able to detect acute episodes with no time delay or any dropped frames. Functional and technical specifications of the envisaged system, as well as the simulation results are reported.","PeriodicalId":117219,"journal":{"name":"2019 IEEE International Conference on Imaging Systems and Techniques (IST)","volume":"81 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122909256","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 : 2019-12-01DOI: 10.1109/IST48021.2019.9010224
Ahmed A. Sleman, A. Soliman, M. Ghazal, H. Sandhu, S. Schaal, Adel Said Elmaghraby, A. El-Baz
The accurate segmentation of retinal layers of the eye in a 3-D Optical Coherence Tomography (OCT) data provides relevant clinical information. This paper introduces a 3D segmentation approach that uses an adaptive patient-specific retinal atlas as well as an appearance model for 3D OCT data. To reconstruct that atlas of 3D retinal scan, we first segment the central area of the macula at which we can clearly identify the fovea. Markov Gibbs Random Field (MGRF) including intensity, shape, and spatial information of 12 layers of retina were all used to segment the selected area of retinal fovea. A set of co-registered OCT scans that were gathered from 200 different individuals were used to build A 2D shape prior. This shape prior was adapted in a following step to the first order appearance and second order spatial interaction MGRF model. After segmenting the center of the macula “foveal area”, the labels and appearances of the layers that have been segmented were used to have the adjacent slices segmented as well. The last step was then repeated recursively until the a 3D OCT scan of the patient is segmented. This approach was tested on 35 individuals while some of them were normal and others were pathological, and then compared to a manually segmented ground truth and finally these results were verified by medical retina experts. Metrics such as Dice Similarity Coefficient (DSC), agreement coefficient (AC), and average deviation (AD) metrics were used to measure the performance of the proposed approach. Accomplished accuracy by the proposed approach shows promising results with noticeable advantages over the state-of-the-art 3D OCT approach.
在三维光学相干断层扫描(OCT)数据中准确分割人眼视网膜层提供相关的临床信息。本文介绍了一种3D分割方法，该方法使用自适应患者特异性视网膜图谱以及3D OCT数据的外观模型。为了重建三维视网膜扫描图谱，我们首先分割黄斑的中心区域，在那里我们可以清楚地识别中央凹。利用马尔可夫吉布斯随机场(Markov Gibbs Random Field, MGRF)，包括12层视网膜的强度、形状和空间信息，对视网膜中央凹选定区域进行分割。从200个不同的个体收集的一组共同注册的OCT扫描被用来预先构建一个二维形状。在接下来的步骤中，将这种形状先验适应于一阶外观和二阶空间相互作用的MGRF模型。对黄斑中心“中央凹区”进行分割后，利用已分割的层的标记和外观对相邻的切片进行分割。然后递归重复最后一步，直到患者的3D OCT扫描被分割。这种方法在35个人身上进行了测试，其中一些是正常的，另一些是病理的，然后与人工分割的地面真相进行比较，最后这些结果由医学视网膜专家验证。使用骰子相似系数(DSC)、一致系数(AC)和平均偏差(AD)等指标来衡量所提出方法的性能。所提出的方法的完成精度显示出有希望的结果，与最先进的3D OCT方法相比具有明显的优势。
{"title":"Retinal Layers OCT Scans 3-D Segmentation","authors":"Ahmed A. Sleman, A. Soliman, M. Ghazal, H. Sandhu, S. Schaal, Adel Said Elmaghraby, A. El-Baz","doi":"10.1109/IST48021.2019.9010224","DOIUrl":"https://doi.org/10.1109/IST48021.2019.9010224","url":null,"abstract":"The accurate segmentation of retinal layers of the eye in a 3-D Optical Coherence Tomography (OCT) data provides relevant clinical information. This paper introduces a 3D segmentation approach that uses an adaptive patient-specific retinal atlas as well as an appearance model for 3D OCT data. To reconstruct that atlas of 3D retinal scan, we first segment the central area of the macula at which we can clearly identify the fovea. Markov Gibbs Random Field (MGRF) including intensity, shape, and spatial information of 12 layers of retina were all used to segment the selected area of retinal fovea. A set of co-registered OCT scans that were gathered from 200 different individuals were used to build A 2D shape prior. This shape prior was adapted in a following step to the first order appearance and second order spatial interaction MGRF model. After segmenting the center of the macula “foveal area”, the labels and appearances of the layers that have been segmented were used to have the adjacent slices segmented as well. The last step was then repeated recursively until the a 3D OCT scan of the patient is segmented. This approach was tested on 35 individuals while some of them were normal and others were pathological, and then compared to a manually segmented ground truth and finally these results were verified by medical retina experts. Metrics such as Dice Similarity Coefficient (DSC), agreement coefficient (AC), and average deviation (AD) metrics were used to measure the performance of the proposed approach. Accomplished accuracy by the proposed approach shows promising results with noticeable advantages over the state-of-the-art 3D OCT approach.","PeriodicalId":117219,"journal":{"name":"2019 IEEE International Conference on Imaging Systems and Techniques (IST)","volume":"82 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114329763","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 : 2019-12-01DOI: 10.1109/IST48021.2019.9010107
A. Naglah, A. DeFilippis, F. Khalifa, N. Singam, B. Aladili, Mohammadi Ghazal, G. Giridharan, A. Khalil, Adel Said Elmaghraby, A. El-Baz
Acute myocardial infarction (MI) is complicated, and multiple etiologies can result in this clinical condition. Guidelines recognize two categories of MI: Thrombotic (Type 1) and non-thrombotic (Type 2), that have quite same prevalence but require unlike treatment. Unfortunately, diagnostic criteria to differentiate between Type 1 and Type 2 require invasive procedures. This results in inefficient and sub-optimal care of patients suspected of MI. This paper presents a novel machine-learning system that detects biomarkers of thrombus formation by analyzing the association between plasma metabolites with the formation of thrombosis in cohort of MI patients at multiple time-points. Study data are collected by a newly introduced non-targeted technique that evaluates the quantities of both known and unknown metabolites from blood samples. Our system uses recursive feature elimination (RFE) and multi-layer perceptron (MLP) neural network to detect associated metabolites at each time-point followed by weighted-voting algorithm using ensemble learning. Our experiment achieves an accuracy of 91%, sensitivity of 89%, and specificity of 94% for MI diagnosis.
{"title":"Computer-Aided Diagnosis of Acute Myocardial Infarction using Time-Dependent Plasma Metabolites","authors":"A. Naglah, A. DeFilippis, F. Khalifa, N. Singam, B. Aladili, Mohammadi Ghazal, G. Giridharan, A. Khalil, Adel Said Elmaghraby, A. El-Baz","doi":"10.1109/IST48021.2019.9010107","DOIUrl":"https://doi.org/10.1109/IST48021.2019.9010107","url":null,"abstract":"Acute myocardial infarction (MI) is complicated, and multiple etiologies can result in this clinical condition. Guidelines recognize two categories of MI: Thrombotic (Type 1) and non-thrombotic (Type 2), that have quite same prevalence but require unlike treatment. Unfortunately, diagnostic criteria to differentiate between Type 1 and Type 2 require invasive procedures. This results in inefficient and sub-optimal care of patients suspected of MI. This paper presents a novel machine-learning system that detects biomarkers of thrombus formation by analyzing the association between plasma metabolites with the formation of thrombosis in cohort of MI patients at multiple time-points. Study data are collected by a newly introduced non-targeted technique that evaluates the quantities of both known and unknown metabolites from blood samples. Our system uses recursive feature elimination (RFE) and multi-layer perceptron (MLP) neural network to detect associated metabolites at each time-point followed by weighted-voting algorithm using ensemble learning. Our experiment achieves an accuracy of 91%, sensitivity of 89%, and specificity of 94% for MI diagnosis.","PeriodicalId":117219,"journal":{"name":"2019 IEEE International Conference on Imaging Systems and Techniques (IST)","volume":"44 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121549134","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 : 2019-12-01DOI: 10.1109/IST48021.2019.9010078
N. Simos, Georgios C. Manikis, E. Papadaki, E. Kavroulakis, G. Bertsias, K. Marias
In this study we explored the robustness of machine learning algorithms for the classification of Neuropsychiatric systemic lupus erythematosus (NPSLE) patients and healthy controls using resting-state fMRI functional connectivity matrices. NPSLE, which is driven by systemic autoimmune inflammation in the context of lupus, involves a wide range of focal and diffuse central and peripheral nervous system symptoms and poses significant diagnostic challenges. Machine learning applications on clinical data may enhance the existing workflow for NPSLE classification as there is no established method of applying neuroimaging data to the diagnosis of NPSLE. Feature selection methods were applied prior to the classification process in order to perform the classification process on a lower dimension feature space. The Connectivity Matrix used consisted of pairwise regional functional associations of the fMRI signals (ROI to ROI correlations) within each of three predetermined brain networks in 41 NPSLE patients and 31 healthy control subjects. Support Vector Machines (SVM) was utilized in the final model. Results were evaluated using a nested cross validation methodology to prevent overfitting, and enhance generalization. Regions of Interest (ROI's) that contributed most in the final model were: Right Inferior Temporal, Thalamus, Left Angular Gyrus, Right Precuneus, Left Primary Motor Cortex, SMA, Left and Right Primary Motor Cortex. With a final F1 score of up to 77%, the results demonstrate the potential for the future implementation of similar methods in the diagnosis of NPSLE.
{"title":"Machine Learning Classification of Neuropsychiatric Systemic Lupus Erythematosus patients using resting-state fMRI functional connectivity","authors":"N. Simos, Georgios C. Manikis, E. Papadaki, E. Kavroulakis, G. Bertsias, K. Marias","doi":"10.1109/IST48021.2019.9010078","DOIUrl":"https://doi.org/10.1109/IST48021.2019.9010078","url":null,"abstract":"In this study we explored the robustness of machine learning algorithms for the classification of Neuropsychiatric systemic lupus erythematosus (NPSLE) patients and healthy controls using resting-state fMRI functional connectivity matrices. NPSLE, which is driven by systemic autoimmune inflammation in the context of lupus, involves a wide range of focal and diffuse central and peripheral nervous system symptoms and poses significant diagnostic challenges. Machine learning applications on clinical data may enhance the existing workflow for NPSLE classification as there is no established method of applying neuroimaging data to the diagnosis of NPSLE. Feature selection methods were applied prior to the classification process in order to perform the classification process on a lower dimension feature space. The Connectivity Matrix used consisted of pairwise regional functional associations of the fMRI signals (ROI to ROI correlations) within each of three predetermined brain networks in 41 NPSLE patients and 31 healthy control subjects. Support Vector Machines (SVM) was utilized in the final model. Results were evaluated using a nested cross validation methodology to prevent overfitting, and enhance generalization. Regions of Interest (ROI's) that contributed most in the final model were: Right Inferior Temporal, Thalamus, Left Angular Gyrus, Right Precuneus, Left Primary Motor Cortex, SMA, Left and Right Primary Motor Cortex. With a final F1 score of up to 77%, the results demonstrate the potential for the future implementation of similar methods in the diagnosis of NPSLE.","PeriodicalId":117219,"journal":{"name":"2019 IEEE International Conference on Imaging Systems and Techniques (IST)","volume":"73 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131970232","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 : 2019-12-01DOI: 10.1109/IST48021.2019.9010097
Haocheng Ma, Lihui Peng
In order to automatically measure the flow rate, alcohol strength and roughly determine the quality of distilled liquor, a system was developed with two transparent standard containers, two load cell sensors, and a camera. This paper presents the imaging part of the measurement system, including the optical path as well as a set of image processing methods to detect the position of the liquid level and calculate the amount of the bubbles on the top of the liquid. Four ArUco markers are used to locate the containers in the captured image and the containers are cropped out. Then the liquid level is detected and the area of bubbles are segmented using statistical information of the pixels in each container. According to the test results, the proposed methods archive accurate and real-time detection on an embedded processor and is robust to the change of the illumination, flowrate, liquid level and camera position.
{"title":"Vision Based Liquid Level Detection and Bubble Area Segmentation in Liquor Distillation","authors":"Haocheng Ma, Lihui Peng","doi":"10.1109/IST48021.2019.9010097","DOIUrl":"https://doi.org/10.1109/IST48021.2019.9010097","url":null,"abstract":"In order to automatically measure the flow rate, alcohol strength and roughly determine the quality of distilled liquor, a system was developed with two transparent standard containers, two load cell sensors, and a camera. This paper presents the imaging part of the measurement system, including the optical path as well as a set of image processing methods to detect the position of the liquid level and calculate the amount of the bubbles on the top of the liquid. Four ArUco markers are used to locate the containers in the captured image and the containers are cropped out. Then the liquid level is detected and the area of bubbles are segmented using statistical information of the pixels in each container. According to the test results, the proposed methods archive accurate and real-time detection on an embedded processor and is robust to the change of the illumination, flowrate, liquid level and camera position.","PeriodicalId":117219,"journal":{"name":"2019 IEEE International Conference on Imaging Systems and Techniques (IST)","volume":"27 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127815512","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 : 2019-12-01DOI: 10.1109/IST48021.2019.9010123
H. Abdeltawab, F. Khalifa, F. Taher, G. Beache, Tamer Mohamed, Adel Said Elmaghraby, M. Ghazal, R. Keynton, A. El-Baz
A new method for the automatic segmentation and quantitative assessment of the left ventricle (LV) is proposed in this paper. The method is composed of two steps. First, a fully convolutional U-net is used for the segmentation of the epi- and endo-cardial boundaries of the LV from cine MR images. This step incorporates a novel loss function that accounts for the class imbalance problem caused by the binary cross entropy (BCE) loss function. Our novel loss function maximizes the segmentation accuracy and penalizes the effect of the class-imbalance caused by BCE. In the second step, the ventricular volume curves are constructed from which LV function parameter is estimated (i.e., ejection fraction). Our method demonstrated a statistical significance in the segmentation of the epi- and endo-cardial boundaries (Dice score of 0.94 and 0.96, respectively) compared with the BCE loss (Dice score of 0.89 and 0.86, respectively). Furthermore, a high positive correlation of 0.97 between the estimated ejection fraction and the gold standard was obtained.
{"title":"Automatic Segmentation and Functional Assessment of the Left Ventricle using U-net Fully Convolutional Network","authors":"H. Abdeltawab, F. Khalifa, F. Taher, G. Beache, Tamer Mohamed, Adel Said Elmaghraby, M. Ghazal, R. Keynton, A. El-Baz","doi":"10.1109/IST48021.2019.9010123","DOIUrl":"https://doi.org/10.1109/IST48021.2019.9010123","url":null,"abstract":"A new method for the automatic segmentation and quantitative assessment of the left ventricle (LV) is proposed in this paper. The method is composed of two steps. First, a fully convolutional U-net is used for the segmentation of the epi- and endo-cardial boundaries of the LV from cine MR images. This step incorporates a novel loss function that accounts for the class imbalance problem caused by the binary cross entropy (BCE) loss function. Our novel loss function maximizes the segmentation accuracy and penalizes the effect of the class-imbalance caused by BCE. In the second step, the ventricular volume curves are constructed from which LV function parameter is estimated (i.e., ejection fraction). Our method demonstrated a statistical significance in the segmentation of the epi- and endo-cardial boundaries (Dice score of 0.94 and 0.96, respectively) compared with the BCE loss (Dice score of 0.89 and 0.86, respectively). Furthermore, a high positive correlation of 0.97 between the estimated ejection fraction and the gold standard was obtained.","PeriodicalId":117219,"journal":{"name":"2019 IEEE International Conference on Imaging Systems and Techniques (IST)","volume":"29 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131672270","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 : 2019-12-01DOI: 10.1109/IST48021.2019.9010236
Giorgos Kakamoukas, Panayiotis Sariciannidis, G. Livanos, M. Zervakis, Dimitris Ramnalis, Vasilis Polychronos, Thomi Karamitsou, A. Folinas, N. Tsitsiokas
Smart Farming (SF) or Precision Agriculture (PA) use precise and efficient approaches for monitoring and processing information from farms, crops, forestry, and livestock aiming at more productive and sustainable rural development. Internet of Things (IoT) is the ecosystem that can provide effective real-time information gathering and processing mechanisms, while supporting cloud access and decision-making mechanisms. Despite the notable progress in the SF field, the ability of these systems to adapt into different types of crops in order to constitute a ready-to-use tool for agricultural stakeholders remains a challenge. In this paper we present a flexible and easy-to-adopt architecture for applying modern IoT-enabled technologies in the context of SF. The proposed architecture encloses Wireless Sensor Networks (WSNs), meteorological stations and Unmanned Aerial Vehicles (UAVs) along with an information processing system that leverages machine learning and computing technologies. The innovation of the proposed architecture lies in the creation of an integrated monitoring and decision support system aiming at production increasing, efficient allocation of resources and protection of plant capital from exogenous (weather and pests) and endogenous (diseases) factors.
{"title":"A Multi-collective, IoT-enabled, Adaptive Smart Farming Architecture","authors":"Giorgos Kakamoukas, Panayiotis Sariciannidis, G. Livanos, M. Zervakis, Dimitris Ramnalis, Vasilis Polychronos, Thomi Karamitsou, A. Folinas, N. Tsitsiokas","doi":"10.1109/IST48021.2019.9010236","DOIUrl":"https://doi.org/10.1109/IST48021.2019.9010236","url":null,"abstract":"Smart Farming (SF) or Precision Agriculture (PA) use precise and efficient approaches for monitoring and processing information from farms, crops, forestry, and livestock aiming at more productive and sustainable rural development. Internet of Things (IoT) is the ecosystem that can provide effective real-time information gathering and processing mechanisms, while supporting cloud access and decision-making mechanisms. Despite the notable progress in the SF field, the ability of these systems to adapt into different types of crops in order to constitute a ready-to-use tool for agricultural stakeholders remains a challenge. In this paper we present a flexible and easy-to-adopt architecture for applying modern IoT-enabled technologies in the context of SF. The proposed architecture encloses Wireless Sensor Networks (WSNs), meteorological stations and Unmanned Aerial Vehicles (UAVs) along with an information processing system that leverages machine learning and computing technologies. The innovation of the proposed architecture lies in the creation of an integrated monitoring and decision support system aiming at production increasing, efficient allocation of resources and protection of plant capital from exogenous (weather and pests) and endogenous (diseases) factors.","PeriodicalId":117219,"journal":{"name":"2019 IEEE International Conference on Imaging Systems and Techniques (IST)","volume":"28 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115047318","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：