https://github.com/koriavinash1/fetal-brain-segmentation

Fully automatic technique for fetal brain segmentation using deep convolutional neural network
https://github.com/koriavinash1/fetal-brain-segmentation

ai artificial-intelligence automatic-segmentation deep-convolutional-neural-networks deep-learning fetal-imaging fmri-analysis medical-image-analysis segmentation unet-image-segmentation

Last synced: about 2 months ago
JSON representation

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

• Host: GitHub
• URL: https://github.com/koriavinash1/fetal-brain-segmentation
• Owner: koriavinash1
• License: mit
• Created: 2018-01-27T16:17:26.000Z (over 7 years ago)
• Default Branch: master
• Last Pushed: 2018-08-05T09:18:50.000Z (almost 7 years ago)
• Last Synced: 2025-04-20T09:39:58.185Z (2 months ago)
• Topics: ai, artificial-intelligence, automatic-segmentation, deep-convolutional-neural-networks, deep-learning, fetal-imaging, fmri-analysis, medical-image-analysis, segmentation, unet-image-segmentation
• Language: Jupyter Notebook
• Homepage:
• Size: 3.12 MB
• Stars: 13
• Watchers: 3
• Forks: 6
• Open Issues: 3
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# Fetal-Brain-Segmentation

## Introduction

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

## Network Architecture

![pipeline](./images/network.png)

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

## Raw data

![data](./images/data.png)

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](./images/model_predictions.png)

### Dice score with epochs

![dice](./images/dice_score.png)



## 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 

* Avinash Kori ([email protected])