GM-AE

Introduction

This repository is an implementation of the paper 'Parameterized Gompertz-guided Morphological AutoEncoder for Predicting Pulmonary Nodule Growth', accepted by TMI.

Gompertz network for volume and mass growth prediction on the baseline nodule over time.

Architecture of morphological autoencoder equipped with shape-aware (DS) and texture-aware (DN ) decoders.

Usage

Training Gompertz:

Run cd gompertz python main_ddp.py --resume --gpu 1,2 --batch-size 16 --epochs 60 --distributed

Training GM-AE：

cd gmae python main_ddp.py

Download trained models

Please download the trained models form baidu, key is 'qntc', and place the downloaded models according to directory structure.

Inference Gompertz:

Run cd gompertz python gompertz/inference.py

Inference GM-AE：

cd gmae python inference.py

Datasets

Details of data processing in STMixer. The paired nodules selected for this project is listed in './data/data.csv' (NLSTt), and please download raw CT scans from NLST

STATISTICS (IN Nn/Ns) OF BENCHMARK SPLITS OF THE NLSTT DATASET AND IN-HOUSE DATASET*.

Results

The center three columns show the predicted texture (left) and shape (right, solid yellow line) of time points Nt by PredNet, NoFoNet, and our GM-AE.

License

This repository is released under the Apache 2.0 license as found in the LICENSE file.

Citation

Please cite GM-AE in your publications if it helps your research.

@article{fang2023parameterized,
  title={Parameterized Gompertz-guided Morphological AutoEncoder for Predicting Pulmonary Nodule Growth},
  author={Fang, Jiansheng and Wang, Jingwen and Li, Anwei and Yan, Yuguang and Liu, Hongbo and Li, Jiajian and Yang, Huifang and Hou, Yonghe and Yang, Xuening and Yang, Ming and others},
  journal={IEEE Transactions on Medical Imaging},
  year={2023},
  publisher={IEEE}
}