Machine learning (ML) algorithms and their applications have become an integral part of daily life and are being used to support medical professionals in clinical decision support. The main applications of ML in the medical field are anomaly detection, medical imaging, clinical trial efficiency, and accelerated drug development. Medical data such as radiation scans, biosignal-based tests and medical laboratory tests are primarily used for anomaly detection. Our work mainly focuses on anomaly detection from chest computed tomography (CT) and electroencephalogram (EEG) signals. 

The 2019 novel coronavirus disease (COVID-19) has caused havoc worldwide, leading to many hospitalizations and mortalities. Severity-based treatment has saved many lives and the further spread of the virus has stopped. A chest CT scan is prescribed to identify the infections developed in the lungs. We propose a method to study the infection severity by extracting the infection regions, and classify the CT scans into three classes such as COVID-19, community-acquired Pneumonia (CAP) and normal. In particular we have developed some novel pre-processing steps which appear to enhance performance.

Epilepsy is a neurological disorder affecting about 50 million people worldwide. It is characterized by a prolonged peculiar burst of neuronal activity within different brain regions. These signatures can be captured efficiently by the scalp EEG signals. Trained neurologists are currently required to detect the seizure from the scalp EEG recordings, which is time-consuming. We propose a method to identify epileptic seizures from the EEG signals using a time delay neural network (TDNN) and long short-term memory (LSTM).