Automatic detection of abnormal EEG signals using wavelet feature extraction and gradient boosting decision tree

Electroencephalography is frequently used for diagnostic evaluation of various brain-related disorders due to its excellent resolution, non-invasive nature and low cost. However, manual analysis EEG signals could be strenuous a time-consuming process experts. It requires long training time physicians develop expertise in it additionally experts have inter-rater agreement (IRA) among themselves. Therefore, many Computer Aided Diagnostic (CAD) based studies considered the automation interpreting alleviate workload support final diagnosis. In this paper, we present an automatic binary classification framework brain multichannel recordings. We propose use Wavelet Packet Decomposition (WPD) techniques decompose into frequency sub-bands extract set statistical features from each selected coefficients. Moreover, novel method reduce dimension feature space without compromising quality extracted features. The are classified using different Gradient Boosting Decision Tree (GBDT) frameworks, which CatBoost, XGBoost LightGBM. Temple University Hospital Abnormal Corpus V2.0.0 test our proposed technique. found that CatBoost classifier achieves accuracy 87.68%, outperforms state-of-the-art on same dataset by more than 1% 3% sensitivity. obtained results research provide important insights usefulness WPD extraction GBDT classifiers classification.