دانشگاه مازندران

 آمار

تعداد نشریات31
تعداد شماره‌ها499
تعداد مقالات4,854
تعداد مشاهده مقاله7,496,406
تعداد دریافت فایل اصل مقاله5,595,947
Mavaddati, Samira, koohi, Hamidreza, Kobti, Ziad. (1404). Detection of Cardiac Arrhythmias Using Transfer Learning and Deep CNN-LSTM Features in the Time–Frequency Domain. پژوهشنامه مدیریت اجرایی, 1(2), 24-31. doi: 10.22080/frai.2025.5665
Samira Mavaddati; Hamidreza koohi; Ziad Kobti. "Detection of Cardiac Arrhythmias Using Transfer Learning and Deep CNN-LSTM Features in the Time–Frequency Domain". پژوهشنامه مدیریت اجرایی, 1, 2, 1404, 24-31. doi: 10.22080/frai.2025.5665
Mavaddati, Samira, koohi, Hamidreza, Kobti, Ziad. (1404). 'Detection of Cardiac Arrhythmias Using Transfer Learning and Deep CNN-LSTM Features in the Time–Frequency Domain', پژوهشنامه مدیریت اجرایی, 1(2), pp. 24-31. doi: 10.22080/frai.2025.5665
Mavaddati, Samira, koohi, Hamidreza, Kobti, Ziad. Detection of Cardiac Arrhythmias Using Transfer Learning and Deep CNN-LSTM Features in the Time–Frequency Domain. پژوهشنامه مدیریت اجرایی, 1404; 1(2): 24-31. doi: 10.22080/frai.2025.5665

Detection of Cardiac Arrhythmias Using Transfer Learning and Deep CNN-LSTM Features in the Time–Frequency Domain

Future Research on AI and IoT
مقاله 4، دوره 1، شماره 2، آبان 2025، صفحه 24-31    اصل مقاله (703.77 K)
نوع مقاله: Research Article
شناسه دیجیتال (DOI): 10.22080/frai.2025.5665
نویسندگان
Samira Mavaddati* 1؛ Hamidreza koohi2؛ Ziad Kobti2
1University of Mazandarn
2University of Windsor
تاریخ دریافت: 24 تیر 1404،  تاریخ پذیرش: 14 مرداد 1404 
چکیده
Cardiac arrhythmias remain a major contributor to global mortality, underscoring the need for timely and accurate diagnosis. With the emergence of deep learning and advanced signal processing, automated detection systems now hold the potential to enhance early diagnosis and clinical decision-making. In this study, we present a novel hybrid framework that integrates transfer learning, convolutional neural networks (CNNs), and long short-term memory (LSTM) networks for effective detection of cardiac arrhythmias from electrocardiogram (ECG) signals. Initially, ECG signals are transformed into the time–frequency domain using continuous wavelet transform (CWT), generating scalogram images that capture both spectral and temporal dynamics. These scalograms serve as inputs to a pre-trained CNN model, enabling the extraction of high-level, domain-invariant features through transfer learning. An LSTM network sequentially processes the resulting deep features to capture temporal dependencies and rhythm patterns in heartbeats. A final fully connected layer performs the classification task, distinguishing between normal and arrhythmic conditions. Experimental results on benchmark ECG datasets demonstrate that the proposed method achieves an impressive classification accuracy of 98.76%, outperforming several existing approaches. This work highlights the promise of combining transfer learning with time–frequency analysis and sequential modeling for robust, real-time cardiac arrhythmia detection, potentially assisting clinicians in reducing diagnostic delays and improving patient outcomes.
کلیدواژه‌ها
Cardiac Arrhythmia Detection؛ Deep Learning؛ Transfer Learning؛ Time–Frequency Analysis؛ CNN-LSTM؛ Continuous Wavelet Transform
آمار
تعداد مشاهده مقاله: 17
تعداد دریافت فایل اصل مقاله: 30


سامانه مدیریت نشریات علمی. قدرت گرفته از سیناوب