Deep Learning Pipeline and Dataset for 3D ECG-Based Ischemia Detection

0_Preprocessing_and_IDs: This folder contains a single document (3DECG_preprocessing_and_recording_IDs.pdf) that includes the anonymous patient ECG recording identifiers (Precath and Postcath) extracted from the PTB Diagnostic ECG Database, the amplitude thresholds used for R-peak detection, and the corresponding preprocessing details required for data standardization and reproducibility.
1_Metrics_Computation: This folder contains the Python pipeline, explanatory documentation, and output files for the analysis of three-dimensional electrocardiogram (3D ECG) trajectories in patients with LAD ischemia before and after coronary revascularization. It includes the computation of geometric metrics (perimeter, curvature, almost-curvature, and torsion) along with statistical analyses.
2_DeepLearning_Classification: This folder includes the deep learning pipeline developed to extend the 3D ECG framework toward predictive modeling. It enables automated discrimination between ischemic and post-revascularization states based on 3D ECG-derived geometric descriptors, integrating residual MLP architectures, isotonic calibration, and patient-wise cross-validation for classification.