End to End System to Discover Brain Tumors for Medical Diagnosis

This repository contains the deep learning models for the project.

The model is based on the paper 3D MRI Brain Tumor Segmentation using Autoencoder Regularization

Credits goes to the original authors of the paper.

Setup

Create the environment using conda

$ conda env create -f environment.yml

Activate the environment using

$ conda activate model-server

Run the flask development server using

$ python app.py

Results

Managed to get upto 80% training accuracy in 800 epochs. This differs from the accuracy and number of epochs stated in the paper as learning rate scheduling and other optimization techniques were not implemented.

Appendix

Link to the project report: link