| Autor: |
Qiao Xiao, Khuan Lee, Mokhtar, Siti Aisah, Ismail, Iskasymar, bin Md Pauzi, Ahmad Luqman, Poh Ying Lim |
| Předmět: |
|
| Zdroj: |
International Journal of Public Health & Clinical Sciences (IJPHCS); Jul/Aug2024, Vol. 11 Issue 4, p1-9, 9p |
| Abstrakt: |
Background: The field of automatic electrocardiogram (ECG) analysis has gained significant attention due to its potential for enhancing diagnostic accuracy and efficiency. An InceptionTime model enhanced with a lead-wise prior knowledge framework (LPKF) has been specifically developed to address the intricate challenge of multi-label ECG classification using the PTB-XL dataset. This model has demonstrated high performance, yielding superior Macro-AUC and F1 scores compared to other state-of-the-art studies. The objective of this study is to further assess the quality and interpretability of the LPKFenhanced InceptionTime model. Methods: The Grad-CAM technique has been adopted for analysis. The Grad-CAM method involves generating maps that highlight the regions of the ECG waveforms the model focuses on while making its predictions. This technique operates by computing the gradients of the target class scores with respect to the feature maps in the final convolutional layer. These gradients are then used to produce a localization map that indicates the most relevant regions contributing to the decision. Result: Grad-CAM visualizations of randomly selected ECG signals reveal the model's focus areas, which align well with clinical knowledge and provide clear, actionable insights into the decision-making process. Conclusion: The integration of Grad-CAM technique has verified the transparency and reliability of the LPKF-enhanced Inception Time approach, confirming its utility and effectiveness in clinical applications [ABSTRACT FROM AUTHOR] |
| Databáze: |
Complementary Index |
| Externí odkaz: |
|
