Abstract

Objective: To evaluate interventricular septum (IVS) vertical displacement using single-point tracking (SPT) as a complementary approach to ECG for heart rate/rhythm categorization.
Methodology: This analytical cross-sectional study included 54 echocardiography videos (47 with ECG). Hybrid VGG-based SPT tracked IVS displacement; features (entropy, frequency, multifractal, time-domain) were classified using decision tree, logistic regression, SVM, and BiLSTM.
Results: Med-SPT improved tracking over raw SPT. After correcting data leakage (oversampling applied inside cross-validation folds), SVM achieved 91.2% accuracy (AUC 0.94) and BiLSTM 89.7% (AUC 0.92). ECG confirmation remains necessary near thresholds (58–62 or 88–92 bpm). Irregular rhythm detection is unreliable due to small sample size (n=2).
Conclusion: IVS-based SPT is a proof-of-concept complementary tool for heart rate categorization but requires external validation on larger cohorts before clinical deployment.

Key words: Interventricular septum, VGG neural network, single-point tracking, arrhythmias.