CwA-T: A Channelwise AutoEncoder with Transformer for EEG Abnormality Detection

Brain disorders such as Alzheimer's disease, epilepsy, Parkinson's disease have attracted significant research interest due to their profound impact on patients' quality of life and healthcare systems globally [1, 2]. Timely and accurate diagnosis is crucial for effective intervention and management, necessitating reliable tools capable of capturing the dynamic changes in brain activity. Electroencephalography (EEG), a cost-effective and non-invasive method for real-time monitoring of brain function, has become a cornerstone in clinical practice for detecting brain disorders. By measuring electrical activity in the brain, EEG provides valuable insights into neural dynamics, particularly for conditions like epilepsy and Alzheimer's disease, where the identification of abnormal patterns is critical for diagnosis and treatment. Recent advances in deep learning (DL) have significantly enhanced the capabilities of computer-aided diagnosis (CAD) systems for EEG analysis. These systems excel at extracting complex, high-dimensional features from raw EEG signals, improving diagnostic accuracy across various applications [3, 4, 5].