Evaluation of Spectrum-Aided Vision Enhancer for Classification of Gastrointestinal Disease in Capsule Endoscopy

SAVE has been used to improve mucosal visualization, improving the identification of small mucosal abnormalities that conventional imaging techniques can overlook, using hyperspectral imaging techniques. The machine learning algorithms used were ResNet50, ResNet101, EfficientNetB2, EfficientNetB5 and EfficientNetV2, where EfficientNetV2 outperformed other models by producing an accuracy of 89 percent in SAVE images.

Autonomous Research Assistant

A fully local, AI-powered research assistant built using PDF parsing, chunking, semantic embedding with sentence-transformers, and vector search using ChromaDB. Designed to help researchers extract, search, and interact with large volumes of academic literature — with a Streamlit interface for natural question-answering.

NeuroV

NeuroV is a deep learning-based web application that detects the presence of brain tumors from MRI scans in .nii or .nii.gz format. The system is built with Flask for the web backend and PyTorch for the inference engine, and provides slice-level confidence visualization using Chart.js.

Implementation of CNN, ANN, LSTM and SNN on the EEG Brainwave dataset

Dataset EEG data (e.g., BCI Competition Dataset, DEAP, or synthetic EEG signals)

Please check my Github profile for more projects here.