JiehuaYang / DLCA-1

A pytorch implementation of DLCA

DLCA

left: input (3D volume)

right: output (3D volume. Blue boxes：predictions, red box：GT label）

Table of contents

• Introduction
• Prerequisites
• Installation
• Usage
• Results

Introduction

This repository contains the source code and the trained model for our work Deep Learning for Detecting Cerebral Aneurysms on CT Angiography.

Prerequisites

• Ubuntu
• Python 3
• Pytorch 0.4.1
• NVIDIA GPU + CUDA CuDNN

This repository has been tested on NVIDIA TITAN Xp. Configurations (e.g batch size, image patch size) may need to be changed on different platforms.

Installation

• Clone this repo:

git clone https://github.com/CTA-detection/DLCA.git
cd DLCA

• Install dependencies:

pip install -r requirements.txt

Usage

1. Preprocess

• Run command as below.

python ./utils/pre_process.py --input="./raw_data/" --output="./train_data/"

2. Train

• Run command as below.

python train.py -j=16 -b=12 --input="train_data/" --output="./checkpoint/"

3. Inference

• Click the checkpoint link(GoogleDrive) to download trained model into "./checkpoint".
• Click the data link(GoogleDrive) to download test image "brain_CTA.nii.gz"
• Run command as below.

# an example with the image named "brain_CTA.nii.gz"
python inference.py -j=1 -b=1 --resume="./checkpoint/trained_model.ckpt" --input="./test_image/brain_CTA" --output="./prediction/brain_CTA"

Results

Sensitivity	False Positive per case
97.5%	13.8