Pub Date : 2023-06-02DOI: 10.1109/ECBIOS57802.2023.10218615
Tanya Chanchalani, Gaurav R, Bhushan Kiran Munoli, Sinchitha H V, P. U
Cardiovascular diseases and Cardiac Arrhythmia are the most familiar reasons for death throughout the world over the last few decades across the world. However, it is difficult to examine patients in all cases accurately, and consultation with a patient for 24 hours by a doctor is not possible as it needs extra patience, expertise, and time. Thus, with ECG sensors, Arduino, and Raspberry Pi, we implemented machine learning models based on K-Nearest Neighbour, Logistic Regression, Support Vector Machine, and Random Forest for heart disease prediction based on the parameters and attributes related to cardiovascular disease. The datasets in this research are available publicly on the UCI website. The early diagnosis of cardiovascular diseases assists in making decisions on lifestyle changes in patients prone to high risk of heart diseases and minimizing the complications. The result of this research can be a milestone in medicine.
{"title":"Implementation of IoT-Based Healthcare Kit","authors":"Tanya Chanchalani, Gaurav R, Bhushan Kiran Munoli, Sinchitha H V, P. U","doi":"10.1109/ECBIOS57802.2023.10218615","DOIUrl":"https://doi.org/10.1109/ECBIOS57802.2023.10218615","url":null,"abstract":"Cardiovascular diseases and Cardiac Arrhythmia are the most familiar reasons for death throughout the world over the last few decades across the world. However, it is difficult to examine patients in all cases accurately, and consultation with a patient for 24 hours by a doctor is not possible as it needs extra patience, expertise, and time. Thus, with ECG sensors, Arduino, and Raspberry Pi, we implemented machine learning models based on K-Nearest Neighbour, Logistic Regression, Support Vector Machine, and Random Forest for heart disease prediction based on the parameters and attributes related to cardiovascular disease. The datasets in this research are available publicly on the UCI website. The early diagnosis of cardiovascular diseases assists in making decisions on lifestyle changes in patients prone to high risk of heart diseases and minimizing the complications. The result of this research can be a milestone in medicine.","PeriodicalId":334600,"journal":{"name":"2023 IEEE 5th Eurasia Conference on Biomedical Engineering, Healthcare and Sustainability (ECBIOS)","volume":" 16","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-06-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"120933560","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 : 2023-06-02DOI: 10.1109/ECBIOS57802.2023.10218646
Binghong Chen, Jenhui Chen
Using natural language processing (NLP) techniques, we conducted a preliminary diagnosis of the disease from the patient syndrome description. Because patients are not medical professionals, they cannot accurately describe all symptoms. To solve this issue, we build a medical knowledge graph (KG) by constructing symptom-disease relation triples for pre-processing the patient syndrome description. According to the medical KG, the descriptions were reconstructed into KG embedding representation. To avoid the knowledge noise issue, we investigate an inclusion-exclusion knowledge filtering approach (IKFA) for symptom-to-disease triples to load them to a pretrained language model (PLM), i.e., bidirectional encoder representations from Transformers (BERT). To train the IKFA, we built a medical diagnosis question-answer dataset (MDQA dataset), which contains large-scale and high-quality questions (patient symptom description) and answers (diagnosis) (Q&A) corpus with 1.63 million entries in the size of 213 MB. The KG was built based on 8,731 diseases with detailed syndrome descriptions in the size of 1.98 MB. The experimental results showed that the IKFA preliminarily diagnosed 8,731 different diseases based on the patient's initial symptom description with an accuracy of 0.9894.
{"title":"Inclusion-Exclusion Knowledge Filtering Approach for Conversation-Based Preliminary Diagnosis","authors":"Binghong Chen, Jenhui Chen","doi":"10.1109/ECBIOS57802.2023.10218646","DOIUrl":"https://doi.org/10.1109/ECBIOS57802.2023.10218646","url":null,"abstract":"Using natural language processing (NLP) techniques, we conducted a preliminary diagnosis of the disease from the patient syndrome description. Because patients are not medical professionals, they cannot accurately describe all symptoms. To solve this issue, we build a medical knowledge graph (KG) by constructing symptom-disease relation triples for pre-processing the patient syndrome description. According to the medical KG, the descriptions were reconstructed into KG embedding representation. To avoid the knowledge noise issue, we investigate an inclusion-exclusion knowledge filtering approach (IKFA) for symptom-to-disease triples to load them to a pretrained language model (PLM), i.e., bidirectional encoder representations from Transformers (BERT). To train the IKFA, we built a medical diagnosis question-answer dataset (MDQA dataset), which contains large-scale and high-quality questions (patient symptom description) and answers (diagnosis) (Q&A) corpus with 1.63 million entries in the size of 213 MB. The KG was built based on 8,731 diseases with detailed syndrome descriptions in the size of 1.98 MB. The experimental results showed that the IKFA preliminarily diagnosed 8,731 different diseases based on the patient's initial symptom description with an accuracy of 0.9894.","PeriodicalId":334600,"journal":{"name":"2023 IEEE 5th Eurasia Conference on Biomedical Engineering, Healthcare and Sustainability (ECBIOS)","volume":"268 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-06-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127375718","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 : 2023-06-02DOI: 10.1109/ECBIOS57802.2023.10218415
Nan Zhou, Xin Li, Dexiu Wang, Chengwen Jin, Weihao Wang, Xiaodong Cui
Presently, animal experimentation is a main teaching tool in basic medical education. Animal experiments can improve medical students' knowledge and its application with practical operation ability. However, based on the premise of the 3R principle, more scholars believe that animal experiments do not meet the needs of medical education and so need to be improved. With the current emergence of big data applications in artificial intelligence (AI) and computer science and technology, human physiological experiments are gradually replacing animal experiments. Although human experiments have drawbacks such as limiting their application in basic medical education. However, the data derived from computer and intelligent tool technology, are more intuitive and widely used than animal experiments. Combined with the application of AI, human computer experiments are gradually replacing animal experiments in medical practice and teaching. We discussed the principles and applications of human physiology in medical education combined with the experience in basic medical teaching. The discussion provided a reference value for medical basic education, especially the new teaching model of basic medical science using artificial intelligence computer simulation.
{"title":"Experimental Teaching System of Human Physiology and Artificial Intelligence Application in Basic Medical Education","authors":"Nan Zhou, Xin Li, Dexiu Wang, Chengwen Jin, Weihao Wang, Xiaodong Cui","doi":"10.1109/ECBIOS57802.2023.10218415","DOIUrl":"https://doi.org/10.1109/ECBIOS57802.2023.10218415","url":null,"abstract":"Presently, animal experimentation is a main teaching tool in basic medical education. Animal experiments can improve medical students' knowledge and its application with practical operation ability. However, based on the premise of the 3R principle, more scholars believe that animal experiments do not meet the needs of medical education and so need to be improved. With the current emergence of big data applications in artificial intelligence (AI) and computer science and technology, human physiological experiments are gradually replacing animal experiments. Although human experiments have drawbacks such as limiting their application in basic medical education. However, the data derived from computer and intelligent tool technology, are more intuitive and widely used than animal experiments. Combined with the application of AI, human computer experiments are gradually replacing animal experiments in medical practice and teaching. We discussed the principles and applications of human physiology in medical education combined with the experience in basic medical teaching. The discussion provided a reference value for medical basic education, especially the new teaching model of basic medical science using artificial intelligence computer simulation.","PeriodicalId":334600,"journal":{"name":"2023 IEEE 5th Eurasia Conference on Biomedical Engineering, Healthcare and Sustainability (ECBIOS)","volume":"66 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-06-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116149356","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 : 2023-06-02DOI: 10.1109/ECBIOS57802.2023.10218715
Ira Puspasari, T. Mengko, A. W. Setiawan, T. Adiono, M. Pramudyo
Processing heart sound signals, especially myocardial infarction (MI) signals, is crucial to identify essential features. The environment strongly influences the results of recording heart sound using a stethoscope on a patient in the hospital, the patient's condition, and other unpredictable noises. A crucial processing step of this signal is filtering. Noise removal in myocardial infarction signals has always been challenging in biomedical signal processing. We compare CEEMDAN and hard thresholding filtering methods. The signal result with the lowest MSE becomes the reference signal in LMSAF. The average MSE value in myocardial infarction signal noise reduction using LMSAF is 0.10, with an average time processing is 1.91 s. The normal signal temporal features on the systolic phase, namely T11: 0.81 s, and on the diastolic phase, namely T12: 0.33 s. The time duration of coronary artery disease (CAD) signal T11: 1.00 s, and T12: 0.46 s, CAD ST-elevation myocardial infarction (CAD STEMI) T-11: 0.99 s, and T12: 0.49 s, CAD non-ST-elevation myocardial infarction (CAD NSTEMI) T-11: 0.98 s, and T12: 0.51 s.
{"title":"Denoising of Heart Sound Signal for Myocardial Infarction Detection Based on Adaptive Filtering","authors":"Ira Puspasari, T. Mengko, A. W. Setiawan, T. Adiono, M. Pramudyo","doi":"10.1109/ECBIOS57802.2023.10218715","DOIUrl":"https://doi.org/10.1109/ECBIOS57802.2023.10218715","url":null,"abstract":"Processing heart sound signals, especially myocardial infarction (MI) signals, is crucial to identify essential features. The environment strongly influences the results of recording heart sound using a stethoscope on a patient in the hospital, the patient's condition, and other unpredictable noises. A crucial processing step of this signal is filtering. Noise removal in myocardial infarction signals has always been challenging in biomedical signal processing. We compare CEEMDAN and hard thresholding filtering methods. The signal result with the lowest MSE becomes the reference signal in LMSAF. The average MSE value in myocardial infarction signal noise reduction using LMSAF is 0.10, with an average time processing is 1.91 s. The normal signal temporal features on the systolic phase, namely T11: 0.81 s, and on the diastolic phase, namely T12: 0.33 s. The time duration of coronary artery disease (CAD) signal T11: 1.00 s, and T12: 0.46 s, CAD ST-elevation myocardial infarction (CAD STEMI) T-11: 0.99 s, and T12: 0.49 s, CAD non-ST-elevation myocardial infarction (CAD NSTEMI) T-11: 0.98 s, and T12: 0.51 s.","PeriodicalId":334600,"journal":{"name":"2023 IEEE 5th Eurasia Conference on Biomedical Engineering, Healthcare and Sustainability (ECBIOS)","volume":"194 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-06-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116464218","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 : 2023-06-02DOI: 10.1109/ECBIOS57802.2023.10218438
Wilson O. Torres, Hannah S. Stuart
Conditions such as multiple sclerosis or arthritis impair normative hand function by diminishing motor control and limiting finger range of motion (ROM), respectively. Day-to-day variation of these symptoms makes it difficult for physicians to track clinically relevant changes in function. This is worsened in populations that lack general access to healthcare. A smartphone holds the potential to be used as a frequent self-screening platform for changes in hand function. We use a custom smartphone application for detecting deviations in force control due to tremors and differences in finger ROM through simple tapping and swiping gestures. We conduct a 17-participant cross-sectional study, which includes two participants with known hand tremors. From the smartphone data during tap-and-hold, we see that people with hand tremors demonstrate less touchscreen force control than normative subjects. During the swiping task, we find a statistically significant moderate correlation between the path length of the swiping gesture and the maximum proximal interphalangeal joint flexion angle of the index finger. We find that different processing methods for the swiping data can reveal additional correlations with metacarpophalangeal flexion. These results are a promising start for the smartphone as an accessible screening tool for tremors and changes in finger ROM.
{"title":"Tap and Swipe Smartphone Gestures Indicating Hand Tremor and Finger Joint Range in Motion","authors":"Wilson O. Torres, Hannah S. Stuart","doi":"10.1109/ECBIOS57802.2023.10218438","DOIUrl":"https://doi.org/10.1109/ECBIOS57802.2023.10218438","url":null,"abstract":"Conditions such as multiple sclerosis or arthritis impair normative hand function by diminishing motor control and limiting finger range of motion (ROM), respectively. Day-to-day variation of these symptoms makes it difficult for physicians to track clinically relevant changes in function. This is worsened in populations that lack general access to healthcare. A smartphone holds the potential to be used as a frequent self-screening platform for changes in hand function. We use a custom smartphone application for detecting deviations in force control due to tremors and differences in finger ROM through simple tapping and swiping gestures. We conduct a 17-participant cross-sectional study, which includes two participants with known hand tremors. From the smartphone data during tap-and-hold, we see that people with hand tremors demonstrate less touchscreen force control than normative subjects. During the swiping task, we find a statistically significant moderate correlation between the path length of the swiping gesture and the maximum proximal interphalangeal joint flexion angle of the index finger. We find that different processing methods for the swiping data can reveal additional correlations with metacarpophalangeal flexion. These results are a promising start for the smartphone as an accessible screening tool for tremors and changes in finger ROM.","PeriodicalId":334600,"journal":{"name":"2023 IEEE 5th Eurasia Conference on Biomedical Engineering, Healthcare and Sustainability (ECBIOS)","volume":"18 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-06-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128227217","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 : 2023-06-02DOI: 10.1109/ECBIOS57802.2023.10218560
Xueyu Guan, Yu-Heng Hsieh, Shyan-Ming Yuan
Wearable devices have significantly impacted precision medicine by providing real-time physiological and behavioral data. This data is crucial for accurate disease diagnosis, monitoring, and improved treatment outcomes. Wearable devices enable personalized health management and prevention programs, particularly being beneficial for chronic disease management. Overall, these devices offer more data and technical support for precision medicine, leading to better-individualized health management and treatment and ultimately improving medical outcomes and quality of life. However, the current device binding requires direct identifiers and grants manufacturers ownership of the generated data, limiting user control and raising privacy concerns. To address this, we propose a blockchain-based platform with two distinct blockchains: decentralized identity and physiological data. Users register decentralized identities on the first blockchain, which are then used for device binding on the second blockchain, enabling de-identified data collection. The platform generates user-specific smart contracts on the Physiological Data blockchain and ensures complete user control over data access. This system enhances the privacy, security, and credibility of users' physiological data, instilling confidence in the use of wearable devices.
{"title":"Blockchain-Based Privacy Preserved Physiological Data Sharing Platform","authors":"Xueyu Guan, Yu-Heng Hsieh, Shyan-Ming Yuan","doi":"10.1109/ECBIOS57802.2023.10218560","DOIUrl":"https://doi.org/10.1109/ECBIOS57802.2023.10218560","url":null,"abstract":"Wearable devices have significantly impacted precision medicine by providing real-time physiological and behavioral data. This data is crucial for accurate disease diagnosis, monitoring, and improved treatment outcomes. Wearable devices enable personalized health management and prevention programs, particularly being beneficial for chronic disease management. Overall, these devices offer more data and technical support for precision medicine, leading to better-individualized health management and treatment and ultimately improving medical outcomes and quality of life. However, the current device binding requires direct identifiers and grants manufacturers ownership of the generated data, limiting user control and raising privacy concerns. To address this, we propose a blockchain-based platform with two distinct blockchains: decentralized identity and physiological data. Users register decentralized identities on the first blockchain, which are then used for device binding on the second blockchain, enabling de-identified data collection. The platform generates user-specific smart contracts on the Physiological Data blockchain and ensures complete user control over data access. This system enhances the privacy, security, and credibility of users' physiological data, instilling confidence in the use of wearable devices.","PeriodicalId":334600,"journal":{"name":"2023 IEEE 5th Eurasia Conference on Biomedical Engineering, Healthcare and Sustainability (ECBIOS)","volume":"44 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-06-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124743186","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 : 2023-06-02DOI: 10.1109/ECBIOS57802.2023.10218379
Yaochang Xi, Peijiang Chen, Chaochao Miao
An end-to-end fall detection method was developed using the YOLOv5s model to accurately locate a person and monitor their fall behavior in a crowd. We added the SE attention mechanism to the second and fourth CSP1_X structures in the network using feature extraction to locate a target more precisely. The spatial pyramid pooling and fully connected spatial convolution (SPPFCSPC) structure was designed to replace SPP to extract the information of the target in different scales effectively and enhance its feature expression ability and detection accuracy. Compared to the previous model, the precision, mean average precision (mAP), and recall rate of the YOLOv5s-2nd-4th-C3SE-SPPFCSPC model increased by 3., 6.2, and 2.9%, respectively. the mAP of the fall category increased by 7.3%. The developed model showed improved detection ability which surpassed that of the original YOLOv5s model.
{"title":"Pedestrian Fall Detection Using Improved YOLOv5","authors":"Yaochang Xi, Peijiang Chen, Chaochao Miao","doi":"10.1109/ECBIOS57802.2023.10218379","DOIUrl":"https://doi.org/10.1109/ECBIOS57802.2023.10218379","url":null,"abstract":"An end-to-end fall detection method was developed using the YOLOv5s model to accurately locate a person and monitor their fall behavior in a crowd. We added the SE attention mechanism to the second and fourth CSP1_X structures in the network using feature extraction to locate a target more precisely. The spatial pyramid pooling and fully connected spatial convolution (SPPFCSPC) structure was designed to replace SPP to extract the information of the target in different scales effectively and enhance its feature expression ability and detection accuracy. Compared to the previous model, the precision, mean average precision (mAP), and recall rate of the YOLOv5s-2nd-4th-C3SE-SPPFCSPC model increased by 3., 6.2, and 2.9%, respectively. the mAP of the fall category increased by 7.3%. The developed model showed improved detection ability which surpassed that of the original YOLOv5s model.","PeriodicalId":334600,"journal":{"name":"2023 IEEE 5th Eurasia Conference on Biomedical Engineering, Healthcare and Sustainability (ECBIOS)","volume":"159 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-06-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114734539","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 : 2023-06-02DOI: 10.1109/ECBIOS57802.2023.10218652
Y. Huang, Huang Shi, Wang Hao, Ruifeng Meng
This article presents a MMP algorithm that combines a 3-D path planning algorithm with a DWA obstacle avoidance algorithm, to enable obstacle-overcoming robots to navigate complex, unstructured scenes. To achieve this, a novel A-star algorithm is proposed that can switch to a greedy best-first strategy algorithm based on the characteristics of the scene. The path planning algorithm is integrated with the DWA algorithm, allowing for local dynamic obstacle avoidance while following the global planned path. Additionally, the algorithm enables the robot to correct its path after obstacle avoidance and overcoming. The feasibility and robustness of the algorithms are demonstrated through simulation experiments in a factory with several complex environments. The algorithms quickly generate a reasonable 3-D path and perform reliable local obstacle avoidance, while taking into account the characteristics of the scene and motion obstacles.
{"title":"Application of 3-D Path Planning and Obstacle Avoidance Algorithms on Obstacle-Overcoming Robots","authors":"Y. Huang, Huang Shi, Wang Hao, Ruifeng Meng","doi":"10.1109/ECBIOS57802.2023.10218652","DOIUrl":"https://doi.org/10.1109/ECBIOS57802.2023.10218652","url":null,"abstract":"This article presents a MMP algorithm that combines a 3-D path planning algorithm with a DWA obstacle avoidance algorithm, to enable obstacle-overcoming robots to navigate complex, unstructured scenes. To achieve this, a novel A-star algorithm is proposed that can switch to a greedy best-first strategy algorithm based on the characteristics of the scene. The path planning algorithm is integrated with the DWA algorithm, allowing for local dynamic obstacle avoidance while following the global planned path. Additionally, the algorithm enables the robot to correct its path after obstacle avoidance and overcoming. The feasibility and robustness of the algorithms are demonstrated through simulation experiments in a factory with several complex environments. The algorithms quickly generate a reasonable 3-D path and perform reliable local obstacle avoidance, while taking into account the characteristics of the scene and motion obstacles.","PeriodicalId":334600,"journal":{"name":"2023 IEEE 5th Eurasia Conference on Biomedical Engineering, Healthcare and Sustainability (ECBIOS)","volume":"8 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-06-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127695382","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 : 2023-06-02DOI: 10.1109/ECBIOS57802.2023.10218603
Chun-Cheng Peng, Bo-Han Liao
Brain tumors pose a significant health threat and cause severe damage to the body and its physiological functions. Traditional diagnostic methods for brain tumors involve expensive medical imaging scans and invasive surgical procedures, resulting in prolonged waiting times and recovery periods. Thus, we explore the potential of deep learning techniques and magnetic resonance imaging (MRI) for the diagnosis of brain tumors. With these technologies, we develop a diagnostic method that is faster, more accurate, and more reliable than current approaches. The proposed model employs preprocessing techniques and convolutional neural network (CNN) methods with the Adam optimizer. An average accuracy reaches 99.8% on the training set and 94.4% on the testing set. These results indicate that the classification of brain MRI is stable and reliable with the proposed method. This proposed approach outperforms four previous methods, demonstrating its superiority and potential for various applications in medical image analysis. In the future, improving overall performance and developing more advanced deep-learning models enables the medical community to diagnose diseases faster and more accurately.
{"title":"Classify Brain Tumors from MRI Images: Deep Learning-Based Approach","authors":"Chun-Cheng Peng, Bo-Han Liao","doi":"10.1109/ECBIOS57802.2023.10218603","DOIUrl":"https://doi.org/10.1109/ECBIOS57802.2023.10218603","url":null,"abstract":"Brain tumors pose a significant health threat and cause severe damage to the body and its physiological functions. Traditional diagnostic methods for brain tumors involve expensive medical imaging scans and invasive surgical procedures, resulting in prolonged waiting times and recovery periods. Thus, we explore the potential of deep learning techniques and magnetic resonance imaging (MRI) for the diagnosis of brain tumors. With these technologies, we develop a diagnostic method that is faster, more accurate, and more reliable than current approaches. The proposed model employs preprocessing techniques and convolutional neural network (CNN) methods with the Adam optimizer. An average accuracy reaches 99.8% on the training set and 94.4% on the testing set. These results indicate that the classification of brain MRI is stable and reliable with the proposed method. This proposed approach outperforms four previous methods, demonstrating its superiority and potential for various applications in medical image analysis. In the future, improving overall performance and developing more advanced deep-learning models enables the medical community to diagnose diseases faster and more accurately.","PeriodicalId":334600,"journal":{"name":"2023 IEEE 5th Eurasia Conference on Biomedical Engineering, Healthcare and Sustainability (ECBIOS)","volume":"12 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-06-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127843796","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 : 2023-06-02DOI: 10.1109/ECBIOS57802.2023.10218631
Hassan M. Qassim, W. Z. W. Hasan, H. R. Ramli, H. Harith, Liyanatul Najwa, Inchi Mat, Msf Salim
Elbow flexion and extension is a common rehabilitation routine that is widely performed by stroke patients to rehabilitate elbow joints. The biceps and triceps muscles are the responsible muscles for flexing and extending the elbow joint. Hence, analyzing the electrical activity of those muscles provides beneficial information on elbow motion intention and eventually can be used for controlling purposes of potential rehabilitation robots. We investigate the Electromyography (EMG) signals of the biceps and triceps of stroke patients and their roles in elbow flexion and extension. The investigation process involves collecting, processing, filtering, and segmenting the collected surface Electromyography (sEMG) signal to ultimately extract specific features. Then, the optimum feature for elbow motion prediction is identified to be later used for controlling purposes. Six time-domain features, specifically MAV, RMS, SD, SAV, SSC, and ZC, were chosen to evaluate their efficiency in predicting elbow joint motion. MAV, RMS, SD, and SAV are the features that showed similar behavior during elbow flexion and extension. However, SAV showed the highest variation in the magnitude when the muscle's state changed from contraction to relaxation and vice-versa. On the other hand, SSC and ZC features showed an arbitrary behavior, where no reliable results were achieved. Eight stroke patients participated in this study after obtaining the ethics approval and consent agreements. The clinical trials were conducted at the Department of Rehabilitation Medicine, Hospital Pengajar Universiti Putra Malaysia (HPUPM).
{"title":"Prediction of Elbow Joint Motion of Stroke Patients by Analyzing Biceps and Triceps Electromyography Signals","authors":"Hassan M. Qassim, W. Z. W. Hasan, H. R. Ramli, H. Harith, Liyanatul Najwa, Inchi Mat, Msf Salim","doi":"10.1109/ECBIOS57802.2023.10218631","DOIUrl":"https://doi.org/10.1109/ECBIOS57802.2023.10218631","url":null,"abstract":"Elbow flexion and extension is a common rehabilitation routine that is widely performed by stroke patients to rehabilitate elbow joints. The biceps and triceps muscles are the responsible muscles for flexing and extending the elbow joint. Hence, analyzing the electrical activity of those muscles provides beneficial information on elbow motion intention and eventually can be used for controlling purposes of potential rehabilitation robots. We investigate the Electromyography (EMG) signals of the biceps and triceps of stroke patients and their roles in elbow flexion and extension. The investigation process involves collecting, processing, filtering, and segmenting the collected surface Electromyography (sEMG) signal to ultimately extract specific features. Then, the optimum feature for elbow motion prediction is identified to be later used for controlling purposes. Six time-domain features, specifically MAV, RMS, SD, SAV, SSC, and ZC, were chosen to evaluate their efficiency in predicting elbow joint motion. MAV, RMS, SD, and SAV are the features that showed similar behavior during elbow flexion and extension. However, SAV showed the highest variation in the magnitude when the muscle's state changed from contraction to relaxation and vice-versa. On the other hand, SSC and ZC features showed an arbitrary behavior, where no reliable results were achieved. Eight stroke patients participated in this study after obtaining the ethics approval and consent agreements. The clinical trials were conducted at the Department of Rehabilitation Medicine, Hospital Pengajar Universiti Putra Malaysia (HPUPM).","PeriodicalId":334600,"journal":{"name":"2023 IEEE 5th Eurasia Conference on Biomedical Engineering, Healthcare and Sustainability (ECBIOS)","volume":"17 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-06-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114904490","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：