MRI DETECTION SYSTEM
This project implements an end‑to‑end MRI tumor detection pipeline using transfer learning with the VGG16 convolutional backbone. Starting from raw scan images, it resizes and normalizes each slice, applies on‑the‑fly augmentations to enrich the training set, and then feeds them into a custom classifier head built atop the frozen VGG16 feature extractor (with the last few convolutional layers fine‑tuned for MRI texture). The model is compiled with an Adam optimizer and trained over multiple epochs using a data generator to manage memory, after which it’s evaluated on held‑out scans. Finally, a utility function loads new MRI images, runs inference to predict “Tumor: <type>” or “No Tumor,” and displays each scan with its confidence score—providing a streamlined tool for rapid, automated tumor screening.