Skip to content

Latest commit

 

History

History
104 lines (76 loc) · 3.68 KB

README.md

File metadata and controls

104 lines (76 loc) · 3.68 KB
Raw

RSNA Intracranial Hemorrhage Detection

This is the source code for the first place solution to the RSNA2019 Intracranial Hemorrhage Detection Challenge.

Journal Link | Cite

Solution write up: Link.

Solutuoin Overview

image

Dependencies

  • opencv-python==3.4.2
  • scikit-image==0.14.0
  • scikit-learn==0.19.1
  • scipy==1.1.0
  • torch==1.1.0
  • torchvision==0.2.1

CODE

  • 2DNet
  • 3DNet
  • SequenceModel

2D CNN Classifier

Pretrained models

Preprocessing

image

Prepare csv file:

download data.zip: https://drive.google.com/open?id=1buISR_b3HQDU4KeNc_DmvKTYJ1gvj5-3

  1. convert dcm to png
python3 prepare_data.py -dcm_path stage_1_train_images -png_path train_png
python3 prepare_data.py -dcm_path stage_1_test_images -png_path train_png
python3 prepare_data.py -dcm_path stage_2_test_images -png_path test_png
  1. train
python3 train_model.py -backbone DenseNet121_change_avg -img_size 256 -tbs 256 -vbs 128 -save_path DenseNet121_change_avg_256
python3 train_model.py -backbone DenseNet169_change_avg -img_size 256 -tbs 256 -vbs 128 -save_path DenseNet169_change_avg_256
python3 train_model.py -backbone se_resnext101_32x4d -img_size 256 -tbs 80 -vbs 40 -save_path se_resnext101_32x4d_256
  1. predict
python3 predict.py -backbone DenseNet121_change_avg -img_size 256 -tbs 4 -vbs 4 -spth DenseNet121_change_avg_256
python3 predict.py -backbone DenseNet169_change_avg -img_size 256 -tbs 4 -vbs 4 -spth DenseNet169_change_avg_256
python3 predict.py -backbone se_resnext101_32x4d -img_size 256 -tbs 4 -vbs 4 -spth se_resnext101_32x4d_256

After single models training, the oof files will be saved in ./SingleModelOutput(three folders for three pipelines).

After training the sequence model, the final submission will be ./FinalSubmission/final_version/submission_tta.csv

Sequence Models

Sequence Model 1

image

Sequence Model 2

image

Path Setup

Set data path in ./setting.py

download

download [csv.zip]

download [feature samples]

Sequence Model Training

CUDA_VISIBLE_DEVICES=0 python main.py

The final submissions are in the folder ../FinalSubmission/version2/submission_tta.csv

Final Submission

Private Leaderboard:

  • 0.04383

Reference

If you find our work useful in your research or if you use parts of this code please consider citing our paper:

title = {A Deep Learning Algorithm for Automatic Detection and Classification of Acute Intracranial Hemorrhages in Head CT Scans},
journal = {NeuroImage: Clinical},  
pages = {102785},  
year = {2021}

TODO

  • Pre-trained models
  • 2DCNN + SeqModel end-to-end training
  • 3DCNN training