Article
Engineering, Electrical & Electronic
Shaolin Ran, Xiaoyun Yang, Ming Liu, Yong Zhang, Cheng Cheng, Hongling Zhu, Ye Yuan
Summary: This article presents a homecare-oriented ECG diagnosis platform that utilizes a large-scale ECG dataset and a deep neural network model to achieve accurate diagnosis and continuous monitoring of various cardiac diseases. The platform also includes algorithm-hardware co-optimization for improved computation speed and adaptability.
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT
(2022)
Article
Computer Science, Artificial Intelligence
Abdolrahman Peimankar, Sadasivan Puthusserypady
Summary: This study proposes a deep learning model for heartbeat segmentation, which combines convolutional neural network and long short-term memory model to analyze ECG signals in real-time, achieving high sensitivity and precision.
EXPERT SYSTEMS WITH APPLICATIONS
(2021)
Article
Multidisciplinary Sciences
Jiewei Lai, Huixin Tan, Jinliang Wang, Lei Ji, Jun Guo, Baoshi Han, Yajun Shi, Qianjin Feng, Wei Yang
Summary: Cardiovascular disease is a global public health problem, and intelligent diagnostic approaches are important in ECG analysis. Convenient wearable ECG devices can detect transient arrhythmias and enable intervention during continuous monitoring. The researchers collected a large dataset of wearable 12-lead ECGs and developed a model that can classify 60 ECG diagnostic terms using self-supervised learning.
NATURE COMMUNICATIONS
(2023)
Article
Multidisciplinary Sciences
Jing Hua, Binbin Chu, Jiawen Zou, Jing Jia
Summary: Wearable devices are frequently employed in diagnosing arrhythmia, but the large amount of data generated during electrocardiogram (ECG) monitoring process can hinder the detection speed and accuracy. To address this issue, many studies have applied deep compressed sensing (DCS) technology to ECG monitoring, allowing under-sampling and reconstruction of ECG signals, thereby optimizing the diagnostic process. However, the reconstruction process is complex and costly. This paper proposes an enhanced classification scheme for deep compressed sensing models, which consists of four modules: pre-processing, compression, and classification. Experimental results demonstrate the robustness of the model, achieving high accuracy, sensitivity, and F1-score compared to other models.
Article
Multidisciplinary Sciences
Marius Reto Bigler, Christian Seiler
Summary: In this study, pre-trained CNNs were used to detect myocardial ischemia with similar accuracy to manual icECG ST-segment shift measurements, showing promising results for using deep learning methods in the diagnosis of ischemic heart disease. The CNNs focused on ST-segment and T-wave morphology for ischemia detection, providing insightful information for future research on ECG-based diagnostic tools.
Review
Cardiac & Cardiovascular Systems
Zeineb Bouzid, Salah S. Al-Zaiti, Raymond Bond, Ervin Sejdic
Summary: The electrocardiogram (ECG) records the electrical activity in the heart in real time, providing an important opportunity to detect various cardiac pathologies. Remote and wearable ECG devices have the potential to revolutionize the way we monitor and diagnose heart conditions, especially when paired with real-time notification techniques.
Article
Chemistry, Analytical
Do Hoon Kim, Gwangjin Lee, Seong Han Kim
Summary: This study proposes a scheme to detect arrhythmias in drivers during driving by using an electrocardiogram (ECG) signal stitching scheme. The proposed scheme extracts stable ECG signals and transforms them into full 10 s ECG signals for classification. Data preprocessing is performed before the ECG stitching algorithm is applied. To classify arrhythmias, transfer learning is performed using convolutional neural networks (CNN) on the transformed ECG signals. The best performance is achieved with the GoogleNet model using continuous wavelet transform (CWT) images, with a classification accuracy of 82.39%.
Article
Instruments & Instrumentation
Marco Chu, Hani E. Naguib
Summary: This study assessed the performance of various conductive composite polymers in collecting electrical signals from the heart, and found that adding 5% carbon nanotubes significantly increased the elastic modulus and conductivity of the composites. SBS-CNT composites at 5% and 10% showed the best performance in detecting ECG waves from the heart.
SMART MATERIALS AND STRUCTURES
(2021)
Article
Physiology
Mengting Yang, Weichao Liu, Henggui Zhang
Summary: The study aims to develop a robust and efficient deep learning method embedded in portable ECG monitors for heartbeat classification. A novel lightweight architecture with weight-based loss was proposed to automatically identify different types of heartbeats, addressing classification bias in imbalanced ECG datasets. The algorithm achieved high accuracy, sensitivity, and F-1 score on the MIT-BIH Arrhythmia Database.
FRONTIERS IN PHYSIOLOGY
(2022)
Article
Chemistry, Analytical
Bambang Tutuko, Muhammad Naufal Rachmatullah, Annisa Darmawahyuni, Siti Nurmaini, Alexander Edo Tondas, Rossi Passarella, Radiyati Umi Partan, Ahmad Rifai, Ade Iriani Sapitri, Firdaus Firdaus
Summary: This study utilizes deep learning for automatic interpretation of ECG signals and focuses on developing a method for identifying atrial fibrillation conditions. The model shows high accuracy and sensitivity in recognizing AF, demonstrating its potential to identify AF conditions through ECG signal delineation.
Article
Engineering, Electrical & Electronic
Tianyu Liu, Yukang Yang, Wenhui Fan, Cheng Wu
Summary: A meta-transfer based few-shot learning method is proposed to handle arrhythmia classification with ECG signals from wearable devices, outperforming other comparative methods in accuracy according to extensive experiments.
DIGITAL SIGNAL PROCESSING
(2021)
Article
Engineering, Electrical & Electronic
Chaehyun Kim, Ahram Jang, Sebin Kim, Sungbo Cho, Donghun Lee, Young-Joon Kim
Summary: In this study, a simple and minimally invasive ECG measurement system for mice was developed. It consists of a miniaturized wearable ECG monitoring device and a smartphone application that displays real-time ECG waveform and heart rate. The system proved to be feasible through validation experiments.
IEEE SENSORS JOURNAL
(2023)
Article
Engineering, Electrical & Electronic
Saifur Rahman, Chandan Karmakar, John Yearwood, Marimuthu Palaniswami
Summary: This study proposes a tunable ECG noise localization system to detect noisy ECG segments in real-time at the IoT-enabled gateway, aiming to improve communication quality and clinical decision-making. Evaluation using publicly available and real-time ECG datasets confirms the effectiveness of the system in reducing data drop rate and increasing R-Peak detection.
IEEE SENSORS JOURNAL
(2022)
Article
Engineering, Multidisciplinary
Alessandra Galli, Giada Giorgi, Claudio Narduzzi
Summary: Wearable cardiac monitors play a positive role in the early detection of cardiovascular pathologies, but wireless transmission of ECG trace data faces challenges due to the large amount of data. This study proposes a signal analysis approach based on a Gaussian dictionary to compress ECG traces, achieving effective compression for wireless data transmission and accurate reconstruction of ECG traces.
Article
Engineering, Electrical & Electronic
Tianyu Liu, Yukang Yang, Wenhui Fan, Cheng Wu
Summary: This article proposes a meta-transfer based few-shot learning method for arrhythmia classification using ECG signals from wearable devices. By converting the ECG signals into spectrograms applicable to 2D-CNN models, and utilizing a special large-training scheme and meta-transfer scheme, the classification accuracy is improved.
DIGITAL SIGNAL PROCESSING
(2022)
Article
Automation & Control Systems
Ali Hadizadeh, Matin Hashemi, Mohammad Labbaf, Mostafa Parniani
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS
(2019)
Article
Computer Science, Hardware & Architecture
Matin Hashemi, Mohammad H. Foroozannejad, Soheil Ghiasi
ACM TRANSACTIONS ON EMBEDDED COMPUTING SYSTEMS
(2013)
Article
Computer Science, Hardware & Architecture
Mohammad H. Foroozannejad, Matin Hashemi, Alireza Mahini, Bevan M. Baas, Soheil Ghiasi
IEEE TRANSACTIONS ON COMPUTER-AIDED DESIGN OF INTEGRATED CIRCUITS AND SYSTEMS
(2014)
Article
Engineering, Biomedical
Alireza Amirshahi, Matin Hashemi
IEEE TRANSACTIONS ON BIOMEDICAL CIRCUITS AND SYSTEMS
(2019)
Article
Computer Science, Theory & Methods
Behrooz Zarebavani, Foad Jafarinejad, Matin Hashemi, Saber Salehkaleybar
IEEE TRANSACTIONS ON PARALLEL AND DISTRIBUTED SYSTEMS
(2020)
Article
Computer Science, Information Systems
Sepehr Dehdashtian, Matin Hashemi, Saber Salehkaleybar
Summary: In this work, a deep learning-based solution is proposed for the recovery of channel code parameters over a candidate set. The solution is capable of identifying parameters for various coding schemes, robust against channel impairments, and outperforms related works in detecting correct code parameters.
IEEE WIRELESS COMMUNICATIONS LETTERS
(2021)
Article
Engineering, Electrical & Electronic
Amir Ahangarzadeh, Matin Hashemi, S. Alireza Nezamalhosseini
Summary: This study proposes a modulation classification method based on convolutional neural networks, which achieves high accuracy in various channel conditions and signal situations without requiring a very deep network. Experimental results demonstrate that the proposed method can accurately classify different modulation types under different system impairment settings.
PHYSICAL COMMUNICATION
(2022)
Article
Engineering, Civil
Saeed Saadatnejad, Siyuan Li, Taylor Mordan, Alexandre Alahi
Summary: This study aims to improve the quality of generated images by rethinking the discriminator architecture, focusing on problems where images are generated given semantic inputs.
IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS
(2022)
Article
Transportation Science & Technology
Saeed Saadatnejad, Mohammadhossein Bahari, Pedram Khorsandi, Mohammad Saneian, Seyed-Mohsen Moosavi-Dezfooli, Alexandre Alahi
Summary: This paper introduces a new attack method to evaluate the social understanding of prediction models in terms of collision avoidance. The attack sheds light on the limitations of current models in social understanding and demonstrates its ability to improve the social understanding of state-of-the-art models.
TRANSPORTATION RESEARCH PART C-EMERGING TECHNOLOGIES
(2022)
Article
Computer Science, Artificial Intelligence
Amir Amirinezhad, Saber Salehkaleybar, Matin Hashemi
Summary: This study focuses on the experiment design problem of learning causal structures from interventional data. A deep reinforcement learning-based solution is proposed, which employs graph neural network and neural network for variable selection in interventions. The networks are jointly trained using Q-iteration algorithm, achieving the recovery of causal structures with minimum interventions.
Proceedings Paper
Computer Science, Artificial Intelligence
Mohammadhossein Bahari, Saeed Saadatnejad, Ahmad Rahimi, Mohammad Shaverdikondori, Amir Hossein Shahidzadeh, Seyed-Mohsen Moosavi-Dezfooli, Alexandre Alahi
Summary: Vehicle trajectory prediction is crucial in self-driving cars. This study introduces a method for generating realistic scenes to evaluate the generalization ability of existing prediction methods in new scenarios, and demonstrates that the generated scenes can enhance the robustness of current models.
2022 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR 2022)
(2022)
Proceedings Paper
Computer Science, Artificial Intelligence
Behnam Parsaeifard, Saeed Saadatnejad, Yuejiang Liu, Taylor Mordan, Alexandre Alahi
Summary: The study proposes learning decoupled representations for global and local pose forecasting tasks, suggesting it is better to stop predictions when uncertainty in human motion increases. The forecasting model outperforms existing methods by over 20% on the pose forecasting benchmark to date.
2021 IEEE/CVF INTERNATIONAL CONFERENCE ON COMPUTER VISION WORKSHOPS (ICCVW 2021)
(2021)
Proceedings Paper
Engineering, Electrical & Electronic
Mohammadhossein Bahari, Seyedramin Rasoulinezhad, Mahdi Amiri, Farzam Gilani, Saeed Saadatnejad, Seyed Alireza Nezamalhosseini, Mahdi Shabany
Summary: Centralized and cloud computing-based network architectures and massive MIMO technology show promise for future communication systems, but challenges remain in overcoming system complexity and energy consumption. Sophisticated algorithms like Joint Channel-and-Data (JCD) estimation are required for recovery of information, but their computation demands pose barriers to practical implementation.
2021 IEEE INTERNATIONAL SYMPOSIUM ON CIRCUITS AND SYSTEMS (ISCAS)
(2021)
