Skip to content

Fully automatic technique for fetal brain segmentation using deep convolutional neural network

License

Notifications You must be signed in to change notification settings

koriavinash1/Fetal-Brain-Segmentation

Repository files navigation

Fetal-Brain-Segmentation

Introduction

This repository contains the implementation of 2D UNet architecture for fetal brain segmentation

Network Architecture

pipeline (https://arxiv.org/pdf/1505.04597.pdf)

Raw data

data

First figure shows raw MR image, second hand annotated groundtruth image and last shows weight map for spatial weighted cross entropy loss


Results

Model predictions

prediction

Dice score with epochs

dice


How to use?


git clone https://github.com/koriavinash1/Fetal-Brain-Segmentation.git
cd Fetal-Brain-Segmentation
pip install -r requirements.txt


Pre-Processing data

Run Generate_Procesed_Data notebook for generating pre-processed data

Folder structure

./src consists all source codes

config -> all initial configurations

data_loader -> multithread data loader

estimator -> model estimator class

network -> network architecture definition

runner -> main function

python runner.py for training and python predictor.py for testing the model


If any comments or issues, pull requests/issues are Welcomed....

Thankyou

Contact