Incorporating Demographic and Heartbeat Features with Multichannel ECG for Cardiac Abnormality Detection using Parallel CNN and GAP Network

Abstract

Background: Early and correct diagnosis of cardiac arrhythmias from Multichannel Electrocardiogram (MECG) is a challenging problem. We address this problem through PhysioNet/Computing in Cardiology Challenge 2021. Method: The proposed method incorporates demographic features including patient age, gender and heartbeat features with MECG. Initially, MECG is cleaned from noise, followed by resampling and segmentation. Then R-peaks are extracted from Lead II signal using Pan Tompkins detector to obtain Heartbeat Features such as Heart Rate, RR Intervals, Mean QRS Amplitude, Hermite polynomial coefficients, statistical features, and Wave Amplitude based features. The demographic and heartbeat features combined with MECG are classified using a Parallel Convolution Neural Network with Global Average Pooling (PCNN-GAP) network. Results: Our team, skylark, achieved a score of 0.36, 0.41, 0.41, 0.45, and 0.49 for the 12-1ead, 6-1ead, 4-1ead, 3-lead, and 2-lead versions of the hidden test set with the Challenge evaluation metric. However, the results were not officially ranked because the training code may select the offline pre-trained models rather than using the training data. Therefore, the model may not adapt to new training instances.