Projects

  
NeuroValid: Transformer Validation in BCI
Comprehensive benchmark evaluating CvT & ViT on neurophysiologically isolated EEG signals (8-30Hz) using Optuna & RSA. (Under Review at BSPC)
BCI-XAI: Unmasking the “Clever Hans” Effect
Exposing artifact-dependency (EOG) in SOTA CNNs (EEGNet) for BCI rehabilitation using Grad-CAM and Saliency Maps. (Under Review at BSPC)
Imagined Speech Decoding
Developing a Spectro-Temporal Transformer architecture to decode imagined speech and isolate pure neural correlates from EMG noise.
Bone Age Estimation
Vision Transformer (ViT) implementation for bone age regression on hand X-ray images, reducing MAE by 12% compared to CNN baselines.
Sleep Apnea Event Detection
Detection of obstructive sleep apnea from single-channel ECG signals utilizing a hybrid CNN-Transformer architecture.
AI-Powered Depression Detection
Advanced detection of depression from EEG signals utilizing Transformer models for complex pattern recognition.
Neural Signal Processing Tutorials
Practical tutorials and scripts for EEG data handling, pre-processing, and artifact removal using MNE-Python and EEGLAB.
Hydrocephalus Prediction
Medical image analysis model for Hydrocephalus detection using Convolutional Neural Networks (CNNs).