hanziying / bonet

Hand Pose Estimation for Pediatric Bone Age Assessment

BoNet

This repository provides a PyTorch implementation of BoNet, presented in the paper Hand Pose Estimation for Pediatric Bone Age Assessment. Oral presentation at MICCAI,2019. BoNet leverages local information and significantly outperforms state-of-the-art methods in Bone Age Assessment (BAA). We propose a new experimental framework with hand detection and hand pose estimation as new tasks to extract local information for BAA methods. We also introduce the Radiological Hand Pose Estimation (RHPE) dataset. Thanks to its fine-grained and precisely localized annotations, our dataset will allow to exploit local information to push forward automated BAA algorithms. To ensure reproducibility of our results and to promote further research on BAA we created the Bone Age Assessment Resources (BAAR).

Bone Age Assessment Resources

We created the Bone Age Assessment Resources (BAAR) as a platform for promoting the development of BAA algorithms. In the BAAR you can download the RSNA and RHPE datasets with keypoints, bounding box and boneage annotations for the training and validation sets. Additionally, you can explore an overview of the methods BCV has developed for this task. Finally, the BAAR include an evaluation server for the test set of RHPE and RSNA

Paper

Hand Pose Estimation for Pediatric Bone Age Assessment
María Escobar ^1* , Cristina González ^1* , Felipe Torres ¹,Laura Daza¹, Gustavo Triana², Pablo Arbeláez^1
*Equal contribution.
¹ Biomedical Computer Vision (BCV) Lab, Universidad de Los Andes.
² Radiology department, Fundación Santa Fe de Bogotá.

Dependencies

• Pytorch 1.2.0
• Pandas 0.24.2
• Horovod 0.18.0
• Tqdm 4.32.1
• Scipy 1.3.0

Usage

Cloning the repository

$ git clone https://github.com/BCV-Uniandes/Bonet.git
$ cd Bonet

Train setup:

Modify the routes in train_net.sh according to your local paths.

bash train_net.sh

Test setup:

Modify the routes in test_net.sh according to your local paths.

bash test_net.sh

Pretrained models

Pretrained models for BoNet are available here

Citation

@inproceedings{escobar2019hand,
  title={Hand Pose Estimation for Pediatric Bone Age Assessment},
  author={Escobar, Mar{\'\i}a and Gonz{\'a}lez, Cristina and Torres, Felipe and Daza, Laura and Triana, Gustavo and Arbel{\'a}ez, Pablo},
  booktitle={International Conference on Medical Image Computing and Computer-Assisted Intervention},
  pages={531--539},
  year={2019},
  organization={Springer}
}