Python code for the paper: Shu, Y., Smith, T. G., Arunachalam, S. P., Tolkacheva, E. G., & Cheng, C. (2023). Image-Decomposition-Enhanced Deep Learning for Detection of Rotor Cores in Cardiac Fibrillation. IEEE Transactions on Biomedical Engineering. DOI: 10.1109/TBME.2023.3292383
Empirical Mode Decomposition algorithm (EEMD) to decompose the image data and characterize the underlying dynamics, and the most representative component is then fed into a You-Only-Look-Once (YOLO) image-recognition architecture for heart disease detection.
A novel low-rank and joint-sparse decomposition (LJSD) is developed for effective reconstruction of epicardial electrical potential. Ensemble Empirical Mode Decomposition algorithm (EEMD) with a You-Only-Look-Once (YOLO) image-recognition architecture is applied for heart diseasr detection
This integrated EEMD-YOLO model achieves the highest rotor detection accuracy (96%) compared to other filtering-AI methods.
- CNN model (You-Only-Look-Once (YOLO) image-recognition architecture)
- Image partition and bounding box design
