Exploring the Efficacy of Explainable Deep Learning in Identifying Neuromarkers for Precise Prediction of Epilepsy and Causal Connectivity Analysis

What is it about?

Causal connectivity among the brain regions has been recently exploited information for discriminating epileptiforms to detect epileptic seizures. Published investigations to detect ictal, inter-ictal, and preictal EEG reported the existence of long-range correlations of excitation within a functionally connected brain region and shifting of the focus of excitations from one region to another region and increase or decrease of intensity in certain frequency bands, which can be quantified using a suitable measure of Granger causality (GC). Deep neural networks obviated explicit pre-processing and feature extraction. The proposed work employs a temporal dilated convolution network to estimate causal connectivity relations among brain-regions in various frequency bands in a distributed manner. It implicitly learns varying overaggressive-lag-orders using stacked layers and covers long-range relationships using exponential update of layer-wise dilation-factor. Model training with several parameter combinations were conducted over 10 subjects. The proposed model outperformed the existing approaches and baseline model in terms of accuracy, sensitivity, and false positive rate. Class-wise dominating features were obtained using statistical significance analysis followed by family-wise error rate correction using the Benjamini-Hochberg method.

Featured Image

Perspectives