Abstract
BACKGROUND: Predicting myocardial functional recovery post-myocardial infarction remains a significant challenge. Researchers are currently exploring myocardial recovery potential by examining myocardial strain characteristics. This study hypothesizes that complexity metrics, which assess waveform randomness, reflect myocardial recovery capacity, hold prognostic value, and may aid in differentiating recovery states. METHODS: Thirty myocardial infarction patients underwent baseline and 1-year follow-up cardiac magnetic resonance, including cine and late gadolinium enhancement scans. Myocardial images were segmented into 16 segments per patient, totaling 480 segments, with 262 displaying initial scarring based on late gadolinium enhancement. Strain characteristics and complexity metrics, including frequency drift, were extracted, and daily segment recovery rates were computed to evaluate recovery. Based on daily segment recovery rates, scarred segments were stratified into High-recovery (37.8%), Low-recovery (38.9%), and Deteriorating (23.3%) groups. Receiver operating characteristic analysis demonstrated the prognostic efficacy of radial/circumferential frequency drift. RESULTS: Correlation analysis identified associations between three complexity metrics and daily segment recovery rates (|r|>0.2), among which radial frequency drift displayed the most robust correlation (r=-0.33, P<0.001). Furthermore, an ensemble classification model using One-vs.-One strategy effectively differentiated groups (High vs. Deteriorating: 0.826, High vs. Low: 0.802, Low vs. Deteriorating: 0.708) and the ensemble model further achieved 97.71% in distinguishing four states. DeLong test results indicated that integrating complexity indices with strain characteristics yielded marked improvements in predictive performance compared to models relying solely on strain parameters (P<0.05). CONCLUSIONS: The complexity metrics of myocardial motion curves exhibited robust prognostic capability. As pivotal indicators of myocardial motion stability, complexity metric could serve as potent tools for providing valuable prognostic information.