NeuroScan — Alzheimer’s MRI Classifier

A compact, high-accuracy MRI image classifier built with ResNet18. The model separates Alzheimer’s vs Normal brain MRIs using a fully custom PyTorch training pipeline.

⸻

Highlights • ResNet18 fine-tuned for binary MRI classification • Clean & minimal training code • Robust data augmentation • Saved best model for easy inference • Works on CPU or GPU

How It Works 1. MRI images are resized to 224×224 2. Images pass through a modified ResNet18 3. Final layer outputs two classes: • 0 → Normal • 1 → Alzheimer

Summary

This project shows a clean, reproducible approach to MRI-based Alzheimer’s classification using transfer learning. Everything — data loading, training, evaluation, and prediction — is contained inside a single polished workflow.