Pub Date : 2022-12-07DOI: 10.1109/IECBES54088.2022.10079412
Doli Hazarika, S. Chanda, C. N. Gupta
Using a smartphone for Electroencephalogram (EEG) based research in the natural environment is a growing field of study. It brings attention to device portability, participant mobility, and system specifications. This article discusses the most recent developments in the field of EEG investigations using smartphones in natural environments, for healthy and clinical applications. We integrate the current trends in smartphone-based EEG studies, namely experimental paradigms, electrode/hardware compatibility, preprocessing frameworks, classifiers, and software apps. However, smartphone devices have inherent limitations like computational time and algorithm performance. Implementing artifact reduction and classification algorithms together in an android smartphone app is still speculative, and possible solutions are proposed. This review presents a holistic insight into our current understanding and challenges of the smartphone’s role in natural environment electroencephalogram trials. Clinical Relevance-These portable smartphone-based EEG systems will be useful in monitoring individuals with psychiatric diseases, in addition to human brain applications in a natural setting. With ubiquitous availability of internet on smartphones, telemedicine is another possible application.
{"title":"Smartphone-Based Natural Environment Electroencephalogram Experimentation-Opportunities and Challenges","authors":"Doli Hazarika, S. Chanda, C. N. Gupta","doi":"10.1109/IECBES54088.2022.10079412","DOIUrl":"https://doi.org/10.1109/IECBES54088.2022.10079412","url":null,"abstract":"Using a smartphone for Electroencephalogram (EEG) based research in the natural environment is a growing field of study. It brings attention to device portability, participant mobility, and system specifications. This article discusses the most recent developments in the field of EEG investigations using smartphones in natural environments, for healthy and clinical applications. We integrate the current trends in smartphone-based EEG studies, namely experimental paradigms, electrode/hardware compatibility, preprocessing frameworks, classifiers, and software apps. However, smartphone devices have inherent limitations like computational time and algorithm performance. Implementing artifact reduction and classification algorithms together in an android smartphone app is still speculative, and possible solutions are proposed. This review presents a holistic insight into our current understanding and challenges of the smartphone’s role in natural environment electroencephalogram trials. Clinical Relevance-These portable smartphone-based EEG systems will be useful in monitoring individuals with psychiatric diseases, in addition to human brain applications in a natural setting. With ubiquitous availability of internet on smartphones, telemedicine is another possible application.","PeriodicalId":146681,"journal":{"name":"2022 IEEE-EMBS Conference on Biomedical Engineering and Sciences (IECBES)","volume":"435 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126984399","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 : 2022-12-07DOI: 10.1109/IECBES54088.2022.10079592
Ashvaany Egambaram, N. Badruddin
Driver drowsiness is a well known problem that depreciates road safety that could cause road accidents, worldwide. Researchers are increasingly using the eye/eyelid images or the electroencephalogram’s (EEG) spectral information to detect drowsiness in drivers. However, no attempt has been made to detect drowsiness using the eye blink artifact features that contaminates EEG signals, which are typically regarded noise and undesired. Therefore, in this study, we have investigated whether the eye blink artifacts that were originally intended to be eliminated from EEG signals could be used to detect drowsiness among drivers. The eye blink artifacts and their features are extracted from EEG signals via the BLINKER algorithm. The deep learning classifiers, multilayer perceptron (MLP) and Recurrent Neural Network with Long-Short-Term-Memory (RNN-LSTM) are trained, validated, and tested to confirm if the eye blink artifacts can be used as an indicator of drowsiness. The investigation has demonstrated that using eye blink artifacts as an indicator of drowsiness is viable, with a classification accuracy of 94.91% achieved through RNN-LSTM.
{"title":"An Investigation to Detect Driver Drowsiness from Eye blink Artifacts Using Deep Learning Models","authors":"Ashvaany Egambaram, N. Badruddin","doi":"10.1109/IECBES54088.2022.10079592","DOIUrl":"https://doi.org/10.1109/IECBES54088.2022.10079592","url":null,"abstract":"Driver drowsiness is a well known problem that depreciates road safety that could cause road accidents, worldwide. Researchers are increasingly using the eye/eyelid images or the electroencephalogram’s (EEG) spectral information to detect drowsiness in drivers. However, no attempt has been made to detect drowsiness using the eye blink artifact features that contaminates EEG signals, which are typically regarded noise and undesired. Therefore, in this study, we have investigated whether the eye blink artifacts that were originally intended to be eliminated from EEG signals could be used to detect drowsiness among drivers. The eye blink artifacts and their features are extracted from EEG signals via the BLINKER algorithm. The deep learning classifiers, multilayer perceptron (MLP) and Recurrent Neural Network with Long-Short-Term-Memory (RNN-LSTM) are trained, validated, and tested to confirm if the eye blink artifacts can be used as an indicator of drowsiness. The investigation has demonstrated that using eye blink artifacts as an indicator of drowsiness is viable, with a classification accuracy of 94.91% achieved through RNN-LSTM.","PeriodicalId":146681,"journal":{"name":"2022 IEEE-EMBS Conference on Biomedical Engineering and Sciences (IECBES)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115932970","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 : 2022-12-07DOI: 10.1109/IECBES54088.2022.10079396
Hisham Khalil, Ahmed Mohamed Saad Emam Saad, U. Khairuddin
Human gait analysis has been one of the primary procedures for diagnosis in modern healthcare applications for various diseases. Instead of using expensive wearable sensors on patients, this research aims to assist in gait analysis and classification for medical diagnoses using computer vision solely. A long short-term memory (LSTM) neural network based on MediaPipe Pose for video-based human gait analysis is proposed to assist in diagnosing patients with neurodegenerative diseases, particularly cerebellar ataxia. The kinematic parameters were extracted from the pose estimation model on captured gait videos before deriving the spatiotemporal parameters for quantitative gait analysis. Data augmentation is applied to increase dataset size, and five-fold cross-validation is performed to verify the suitability of the developed dataset for training deep neural networks. The selected LSTM model achieves a testing accuracy of 99.8% with very high precision and recall metrics for ataxic and normal gait classes. The proposed methodology can be applied in broader applications for remote rehabilitation and patient monitoring. Clinical Relevance-The developed system can assist physicians in diagnosing cerebellar ataxic patients and monitoring gait rehabilitation process remotely via camera vision.
{"title":"Diagnosis of Cerebellar Ataxia Based on Gait Analysis Using Human Pose Estimation: A Deep Learning Approach","authors":"Hisham Khalil, Ahmed Mohamed Saad Emam Saad, U. Khairuddin","doi":"10.1109/IECBES54088.2022.10079396","DOIUrl":"https://doi.org/10.1109/IECBES54088.2022.10079396","url":null,"abstract":"Human gait analysis has been one of the primary procedures for diagnosis in modern healthcare applications for various diseases. Instead of using expensive wearable sensors on patients, this research aims to assist in gait analysis and classification for medical diagnoses using computer vision solely. A long short-term memory (LSTM) neural network based on MediaPipe Pose for video-based human gait analysis is proposed to assist in diagnosing patients with neurodegenerative diseases, particularly cerebellar ataxia. The kinematic parameters were extracted from the pose estimation model on captured gait videos before deriving the spatiotemporal parameters for quantitative gait analysis. Data augmentation is applied to increase dataset size, and five-fold cross-validation is performed to verify the suitability of the developed dataset for training deep neural networks. The selected LSTM model achieves a testing accuracy of 99.8% with very high precision and recall metrics for ataxic and normal gait classes. The proposed methodology can be applied in broader applications for remote rehabilitation and patient monitoring. Clinical Relevance-The developed system can assist physicians in diagnosing cerebellar ataxic patients and monitoring gait rehabilitation process remotely via camera vision.","PeriodicalId":146681,"journal":{"name":"2022 IEEE-EMBS Conference on Biomedical Engineering and Sciences (IECBES)","volume":"98 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116105405","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 : 2022-12-07DOI: 10.1109/IECBES54088.2022.10079262
M. F. El-Muhammady, H. Yusof, M. A. Rashidan, S. Sidek
Research in the area of Human-robot Interaction (HRI) has gained momentum in recent years. The robot-based intervention has now spread its wings to help the less fortunate specifically children who suffer from Autism Spectrum Disorder (ASD). These robot-based intervention studies utilized HRI in improving impaired skills, such as, social skills, motor skills, and behavior. Recently, robot-based therapies have shown encouraging outcomes in improving the social skills of Autism Spectrum Disorder Children (ASDC). Herein, this paper aims to review the studies of the use of adaptive HRI in improving social skills of ASDC. Due to rapid advancement in HRI studies, this paper limits the review from the past 10 years. The steps in the procedure, such as the choice of study participants and outcome measurements, are described in this paper. According to the analysis, all the studies deployed NAO robots as their HRI system. There was variability within the parameters measured to evaluate the efficacy of the HRI, however, the analysis discovered that robot-based interventions were practically helpful in enhancing the social skills for ASDC.
{"title":"Intervention of Autism Spectrum Disorder (ASD) in a New Perspective: A Review on the Deployment of Adaptive Human-Robot Interaction (HRI) System in Enhancing Social Skill Impairments","authors":"M. F. El-Muhammady, H. Yusof, M. A. Rashidan, S. Sidek","doi":"10.1109/IECBES54088.2022.10079262","DOIUrl":"https://doi.org/10.1109/IECBES54088.2022.10079262","url":null,"abstract":"Research in the area of Human-robot Interaction (HRI) has gained momentum in recent years. The robot-based intervention has now spread its wings to help the less fortunate specifically children who suffer from Autism Spectrum Disorder (ASD). These robot-based intervention studies utilized HRI in improving impaired skills, such as, social skills, motor skills, and behavior. Recently, robot-based therapies have shown encouraging outcomes in improving the social skills of Autism Spectrum Disorder Children (ASDC). Herein, this paper aims to review the studies of the use of adaptive HRI in improving social skills of ASDC. Due to rapid advancement in HRI studies, this paper limits the review from the past 10 years. The steps in the procedure, such as the choice of study participants and outcome measurements, are described in this paper. According to the analysis, all the studies deployed NAO robots as their HRI system. There was variability within the parameters measured to evaluate the efficacy of the HRI, however, the analysis discovered that robot-based interventions were practically helpful in enhancing the social skills for ASDC.","PeriodicalId":146681,"journal":{"name":"2022 IEEE-EMBS Conference on Biomedical Engineering and Sciences (IECBES)","volume":"55 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123540468","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 : 2022-12-07DOI: 10.1109/IECBES54088.2022.10079302
Rachel Weng Kei Boon, Y. Z. Chong, Yin Qing Tan, V. Sundar, Y. Selva, S. Chan
This study aimed to investigate the differences between muscle force generated by hard and soft cushioned running shoes during running. Six healthy Malaysian male amateur runners were recruited into this study (age: $29.67 pm 3.44$ years; height: $170.32 pm 3.36$ cm; body mass: $68.23 pm 4.90$ kg). Participants ran on the instrumented treadmill with embedded force plates at fixed speed of 12km/h. Running gait models were developed using open-source software – OpenSim, and muscle activations were estimated using the static optimization method. Pearson’s Correlation analysis was used to determine the correlation between simulated and experimental surface EMG data. Strong significant linear correlation was found in biceps femoris, rectus femoris and lateral gastrocnemius, while moderate significant linear correlation was found in tibialis anterior. Independent-sample t-test was used to investigate the differences between the muscle activation while run at hard and soft cushioned shoes. Results demonstrated that there were significant differences found in the rectus femoris (10-30%, p $lt0.001$; 40-60%, p=0.014). However, there is no significant differences been found in the biceps femoris, lateral gastrocnemius and tibialis anterior between the hard and soft cushioning groups. Further analysis on the effect of shoe cushioning can be studied as the running distance increases.
{"title":"Effect of Running Shoe Cushioning on Muscle Activation using OpenSim","authors":"Rachel Weng Kei Boon, Y. Z. Chong, Yin Qing Tan, V. Sundar, Y. Selva, S. Chan","doi":"10.1109/IECBES54088.2022.10079302","DOIUrl":"https://doi.org/10.1109/IECBES54088.2022.10079302","url":null,"abstract":"This study aimed to investigate the differences between muscle force generated by hard and soft cushioned running shoes during running. Six healthy Malaysian male amateur runners were recruited into this study (age: $29.67 pm 3.44$ years; height: $170.32 pm 3.36$ cm; body mass: $68.23 pm 4.90$ kg). Participants ran on the instrumented treadmill with embedded force plates at fixed speed of 12km/h. Running gait models were developed using open-source software – OpenSim, and muscle activations were estimated using the static optimization method. Pearson’s Correlation analysis was used to determine the correlation between simulated and experimental surface EMG data. Strong significant linear correlation was found in biceps femoris, rectus femoris and lateral gastrocnemius, while moderate significant linear correlation was found in tibialis anterior. Independent-sample t-test was used to investigate the differences between the muscle activation while run at hard and soft cushioned shoes. Results demonstrated that there were significant differences found in the rectus femoris (10-30%, p $lt0.001$; 40-60%, p=0.014). However, there is no significant differences been found in the biceps femoris, lateral gastrocnemius and tibialis anterior between the hard and soft cushioning groups. Further analysis on the effect of shoe cushioning can be studied as the running distance increases.","PeriodicalId":146681,"journal":{"name":"2022 IEEE-EMBS Conference on Biomedical Engineering and Sciences (IECBES)","volume":"59 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130870768","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 : 2022-12-07DOI: 10.1109/IECBES54088.2022.10079667
P. L. Thon, J. Than, R. M. Kassim, A. Yunus, N. Noor, P. Then
COVID-19 has been raging for almost three years ever since its first outbreak. It is without a doubt that it is a common human goal to end the pandemic and how it was before it started. Many efforts have been made to work toward this goal. In computer vision, works have been done to aid medical professionals into faster and more effective procedures when dealing with the disease. For example, disease diagnosis and severity prediction using chest imaging. At the same time, vision transformer is introduced and quickly stormed its way into one of the best deep learning models ever developed due to its ability to achieve good performance while being resources friendly. In this study, we investigated the performance of ViT on COVID19 severity classification using an open-source CXR images dataset. We applied different augmentation and transformation techniques to the dataset to see ViT’s ability to learn the features of the different severity levels of the disease. It is concluded that training ViT using the horizontally flipped images added to the original dataset gives the best overall accuracy of 0.862. To achieve explainability, we have also applied Grad-CAM to the best performing model to make sure it is looking at relevant region of the CXR image upon predicting the class label.
{"title":"Explainable COVID-19 Three Classes Severity Classification Using Chest X-Ray Images","authors":"P. L. Thon, J. Than, R. M. Kassim, A. Yunus, N. Noor, P. Then","doi":"10.1109/IECBES54088.2022.10079667","DOIUrl":"https://doi.org/10.1109/IECBES54088.2022.10079667","url":null,"abstract":"COVID-19 has been raging for almost three years ever since its first outbreak. It is without a doubt that it is a common human goal to end the pandemic and how it was before it started. Many efforts have been made to work toward this goal. In computer vision, works have been done to aid medical professionals into faster and more effective procedures when dealing with the disease. For example, disease diagnosis and severity prediction using chest imaging. At the same time, vision transformer is introduced and quickly stormed its way into one of the best deep learning models ever developed due to its ability to achieve good performance while being resources friendly. In this study, we investigated the performance of ViT on COVID19 severity classification using an open-source CXR images dataset. We applied different augmentation and transformation techniques to the dataset to see ViT’s ability to learn the features of the different severity levels of the disease. It is concluded that training ViT using the horizontally flipped images added to the original dataset gives the best overall accuracy of 0.862. To achieve explainability, we have also applied Grad-CAM to the best performing model to make sure it is looking at relevant region of the CXR image upon predicting the class label.","PeriodicalId":146681,"journal":{"name":"2022 IEEE-EMBS Conference on Biomedical Engineering and Sciences (IECBES)","volume":"85 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124136445","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 : 2022-12-07DOI: 10.1109/IECBES54088.2022.10079461
Y. Z. Chong, Kah Sheng Tiang, S. Chan
The application of telerehabilitation system has gained popularity and acceptance recently due to the restrictions in controlling the COVID-19 pandemic. This paper described the development of an elbow-wrist telerehabilitation system that complement the disrupted routine rehabilitation session. The developed system consists of a wearable exoskeleton system that assist in rehabilitation of the elbow and wrist joints for individuals with neurological conditions such as Parkinson’s and Spinal Cord Injuries that affects movements of the upper extremities. The two modes of operation available enables the adoption of the 5G technology in the near future. This system also potentially fulfills the requirement of Accessibility, Availability, Affordability, and Acceptability (4As) of Telerehabilitation System in Malaysia. Overall development cost of the system is approximately MYR 500. The system enable rehabilitation to be performed at home-setting with a cloud-based monitoring system that will provide long-term monitoring for clinician’s assessment. The project provides a proof-of-concept of such system in the Malaysian context.Clinical Relevance – This work demonstrated the proof-of concept of a 4A system is applicable in the Malaysian context.
{"title":"Development of Elbow-Wrist Telerehabilitation System – An Affordable, Available, Accessible, and Acceptable (4As) System","authors":"Y. Z. Chong, Kah Sheng Tiang, S. Chan","doi":"10.1109/IECBES54088.2022.10079461","DOIUrl":"https://doi.org/10.1109/IECBES54088.2022.10079461","url":null,"abstract":"The application of telerehabilitation system has gained popularity and acceptance recently due to the restrictions in controlling the COVID-19 pandemic. This paper described the development of an elbow-wrist telerehabilitation system that complement the disrupted routine rehabilitation session. The developed system consists of a wearable exoskeleton system that assist in rehabilitation of the elbow and wrist joints for individuals with neurological conditions such as Parkinson’s and Spinal Cord Injuries that affects movements of the upper extremities. The two modes of operation available enables the adoption of the 5G technology in the near future. This system also potentially fulfills the requirement of Accessibility, Availability, Affordability, and Acceptability (4As) of Telerehabilitation System in Malaysia. Overall development cost of the system is approximately MYR 500. The system enable rehabilitation to be performed at home-setting with a cloud-based monitoring system that will provide long-term monitoring for clinician’s assessment. The project provides a proof-of-concept of such system in the Malaysian context.Clinical Relevance – This work demonstrated the proof-of concept of a 4A system is applicable in the Malaysian context.","PeriodicalId":146681,"journal":{"name":"2022 IEEE-EMBS Conference on Biomedical Engineering and Sciences (IECBES)","volume":"18 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121504953","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 : 2022-12-07DOI: 10.1109/IECBES54088.2022.10079411
H. Gan, M. H. Ramlee, Asnida Abdul Wahab, W. Mahmud, D. Setiadi
In recent years, convolutional neural network (CNN) becomes the mainstream image processing techniques for numerous medical imaging tasks such as segmentation, classification and detection. Nonetheless, CNN is limited to processing of fixed size input and demonstrates low generalizability to unseen features. Graph deep learning adopts graph concept and properties to capture rich information from complex data structure. Graph can effectively analyze the pairwise relationship between the target entities. Implementation of graph deep learning in medical imaging requires the conversion of grid-like image structure into graph representation. To date, the conversion mechanism remains underexplored. In this work, image-to-graph conversion via clustering has been proposed. Locally group homogeneous pixels have been grouped into a superpixel, which can be identified as node. Simple linear iterative clustering (SLIC) emerged as the suitable clustering technique to build superpixels as nodes for subsequent graph deep learning computation. The method was validated on knee, call and membrane image datasets. SLIC has reported Rand score of 0.92±0.015 and Silhouette coefficient of 0.85±0.02 for cell dataset, 0.62±0.02 (Rand score) and 0.61±0.07 (Silhouette coefficient) for membrane dataset and 0.82±0.025 (Rand score) and 0.67±0.02 (Silhouette coefficient) for knee dataset. Future works will investigate the performance of superpixel with enforcing connectivity as the prerequisite to develop graph deep learning for medical image segmentation.
近年来，卷积神经网络(CNN)成为分割、分类、检测等众多医学成像任务的主流图像处理技术。然而，CNN仅限于处理固定大小的输入，对未见特征的泛化能力较低。图深度学习采用图的概念和属性，从复杂的数据结构中获取丰富的信息。图可以有效地分析目标实体之间的两两关系。在医学成像中实现图深度学习需要将网格状图像结构转换为图表示。迄今为止，这种转换机制仍未得到充分探索。在这项工作中，通过聚类提出了图像到图形的转换。局部组同质像素被分组成一个超像素，该超像素可以被识别为节点。简单线性迭代聚类(Simple linear iterative clustering, SLIC)是构建超像素作为后续图深度学习计算节点的合适聚类技术。该方法在膝关节、皮肤和膜图像数据集上进行了验证。据SLIC报道，细胞数据的Rand评分为0.92±0.015,Silhouette系数为0.85±0.02;膜数据的Rand评分为0.62±0.02,Silhouette系数为0.61±0.07;膝盖数据的Rand评分为0.82±0.025,Silhouette系数为0.67±0.02。未来的工作将研究超像素的性能，并将加强连通性作为开发用于医学图像分割的图深度学习的先决条件。
{"title":"Image-to-Graph Transformation via Superpixel Clustering to Build Nodes in Deep Learning for Graph","authors":"H. Gan, M. H. Ramlee, Asnida Abdul Wahab, W. Mahmud, D. Setiadi","doi":"10.1109/IECBES54088.2022.10079411","DOIUrl":"https://doi.org/10.1109/IECBES54088.2022.10079411","url":null,"abstract":"In recent years, convolutional neural network (CNN) becomes the mainstream image processing techniques for numerous medical imaging tasks such as segmentation, classification and detection. Nonetheless, CNN is limited to processing of fixed size input and demonstrates low generalizability to unseen features. Graph deep learning adopts graph concept and properties to capture rich information from complex data structure. Graph can effectively analyze the pairwise relationship between the target entities. Implementation of graph deep learning in medical imaging requires the conversion of grid-like image structure into graph representation. To date, the conversion mechanism remains underexplored. In this work, image-to-graph conversion via clustering has been proposed. Locally group homogeneous pixels have been grouped into a superpixel, which can be identified as node. Simple linear iterative clustering (SLIC) emerged as the suitable clustering technique to build superpixels as nodes for subsequent graph deep learning computation. The method was validated on knee, call and membrane image datasets. SLIC has reported Rand score of 0.92±0.015 and Silhouette coefficient of 0.85±0.02 for cell dataset, 0.62±0.02 (Rand score) and 0.61±0.07 (Silhouette coefficient) for membrane dataset and 0.82±0.025 (Rand score) and 0.67±0.02 (Silhouette coefficient) for knee dataset. Future works will investigate the performance of superpixel with enforcing connectivity as the prerequisite to develop graph deep learning for medical image segmentation.","PeriodicalId":146681,"journal":{"name":"2022 IEEE-EMBS Conference on Biomedical Engineering and Sciences (IECBES)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130894942","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 : 2022-12-07DOI: 10.1109/IECBES54088.2022.10079259
Yimin Yan, S. Samdin, K. Minhad
Epilepsy is a chronic non-communicable disease caused by abnormal firing activity of brain neurons in all age groups. This research studies two time-frequency domain analysis methods of EEG signals, short-time Fourier transform and continuous wavelet transform, using these two methods to analyze one piece of epilepsy EEG signals. The window size will affect the time resolution and frequency resolution for the short-time Fourier transform. The larger the window size, the lower the time resolution and the higher the frequency resolution, and vice versa. Therefore, it is vital to choose the most suitable window size. The best window size is 0.4s through experiments; for continuous wavelet transform, is a parameter that controls the scale of the Gaussian kernel, and $omega$ is the frequency of Morlet. The rule is obtained through experiments; when the results of $sigma times omega$ are between 2 and 4, the analysis results can simultaneously exhibit higher time, frequency resolution, and more details. No matter what the values of $sigma$ and $omega$ are, as long as the product of the two is the same, the analysis results are the same. Finally, this study obtained the seizures trend. The trend of epileptic seizures mainly started from the right side of the brain, moved to the left side, then to the forehead, and finally to the occipital brain region.
癫痫是一种慢性非传染性疾病，由所有年龄组的大脑神经元异常放电活动引起。本研究研究了脑电信号的两种时频域分析方法——短时傅里叶变换和连续小波变换，利用这两种方法对一幅癫痫脑电信号进行分析。窗口大小将影响短时傅里叶变换的时间分辨率和频率分辨率。窗口尺寸越大，时间分辨率越低，频率分辨率越高，反之亦然。因此，选择最合适的窗口大小至关重要。通过实验，最佳窗口尺寸为0.4s;对于连续小波变换，为控制高斯核尺度的参数，$omega$为Morlet频率。通过实验得出了这一规律;当$sigma times omega$的结果在2和4之间时，分析结果可以同时显示更高的时间、频率分辨率和更多的细节。无论$sigma$和$omega$的值是多少，只要两者的乘积相同，分析结果就相同。最后，本研究获得了癫痫发作趋势。癫痫发作的趋势主要从大脑右侧开始，向左侧移动，然后向前额移动，最后向枕脑区移动。
{"title":"Window-based Time-Frequency Methods for Analyzing Epileptic EEG Signals","authors":"Yimin Yan, S. Samdin, K. Minhad","doi":"10.1109/IECBES54088.2022.10079259","DOIUrl":"https://doi.org/10.1109/IECBES54088.2022.10079259","url":null,"abstract":"Epilepsy is a chronic non-communicable disease caused by abnormal firing activity of brain neurons in all age groups. This research studies two time-frequency domain analysis methods of EEG signals, short-time Fourier transform and continuous wavelet transform, using these two methods to analyze one piece of epilepsy EEG signals. The window size will affect the time resolution and frequency resolution for the short-time Fourier transform. The larger the window size, the lower the time resolution and the higher the frequency resolution, and vice versa. Therefore, it is vital to choose the most suitable window size. The best window size is 0.4s through experiments; for continuous wavelet transform, is a parameter that controls the scale of the Gaussian kernel, and $omega$ is the frequency of Morlet. The rule is obtained through experiments; when the results of $sigma times omega$ are between 2 and 4, the analysis results can simultaneously exhibit higher time, frequency resolution, and more details. No matter what the values of $sigma$ and $omega$ are, as long as the product of the two is the same, the analysis results are the same. Finally, this study obtained the seizures trend. The trend of epileptic seizures mainly started from the right side of the brain, moved to the left side, then to the forehead, and finally to the occipital brain region.","PeriodicalId":146681,"journal":{"name":"2022 IEEE-EMBS Conference on Biomedical Engineering and Sciences (IECBES)","volume":"32 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131408703","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 : 2022-12-07DOI: 10.1109/IECBES54088.2022.10079688
Wafaa Khazaal Shams, U. Qidwai
Brain computer interaction (BCI) based on electroencephalographic (EEG) signal helps people who suffering from disability to carry out their daily life. However, the numerus number of researches have done in this field, there are problems of high variance in accuracy and in efficiency among individuals. This paper presents a recognition method for eyes open (EO) and eyes closed(EC) of EEG signal using one channel P5. The model has tested to control a servo motor. A two types of feature are investigated; Energy of alpha power spectrum (EPSD) and relative alpha power (RAP). Further a linear discriminate analysis (LDA) and a nonlinear support vector machine (SVM) classifier are used. The used data are offline signals of 10 children age (4-5) years old. Results indicate the efficiency of EPSD hence the accuracy reaches to 95% for 2 sec time interval and 93.4% for 1 sec time interval. The RAP feature accuracy is 78.7%. The LDA has a significant performance compare to SVM. Both classifiers show high performance to detect EO event better than EC event. This study shows the ability of build EEG-BCI using one channel and with less computation process which can be affordable to most people with disability.
{"title":"Linear Classifier Approach to Detect Alpha Parietal Modulation for Brain Computer Interface","authors":"Wafaa Khazaal Shams, U. Qidwai","doi":"10.1109/IECBES54088.2022.10079688","DOIUrl":"https://doi.org/10.1109/IECBES54088.2022.10079688","url":null,"abstract":"Brain computer interaction (BCI) based on electroencephalographic (EEG) signal helps people who suffering from disability to carry out their daily life. However, the numerus number of researches have done in this field, there are problems of high variance in accuracy and in efficiency among individuals. This paper presents a recognition method for eyes open (EO) and eyes closed(EC) of EEG signal using one channel P5. The model has tested to control a servo motor. A two types of feature are investigated; Energy of alpha power spectrum (EPSD) and relative alpha power (RAP). Further a linear discriminate analysis (LDA) and a nonlinear support vector machine (SVM) classifier are used. The used data are offline signals of 10 children age (4-5) years old. Results indicate the efficiency of EPSD hence the accuracy reaches to 95% for 2 sec time interval and 93.4% for 1 sec time interval. The RAP feature accuracy is 78.7%. The LDA has a significant performance compare to SVM. Both classifiers show high performance to detect EO event better than EC event. This study shows the ability of build EEG-BCI using one channel and with less computation process which can be affordable to most people with disability.","PeriodicalId":146681,"journal":{"name":"2022 IEEE-EMBS Conference on Biomedical Engineering and Sciences (IECBES)","volume":"18 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122878488","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：