Addressing Sparse Annotation: a Novel Semantic Energy Loss for Tumor Cell Detection from Histopathologic Images

This repository contains the pytorch code for the paper:

Addressing Sparse Annotation: a Novel Semantic Energy Loss for Tumor Cell Detection from Histopathologic Images, BIBM2023. (PDF)

Citation

If you find this code useful in your research, please cite our work:

@article{Xu2023AddressingSA,
  title={Addressing Sparse Annotation: a Novel Semantic Energy Loss for Tumor Cell Detection from Histopathologic Images},
  author={Zhengyang Xu and XiangLong Du and Yuxin Kang and Hong Lv and Ming Li and Wentao Yang and Lei Cui and Hansheng Li and Yaqiong Xing and Jun Feng and Lin Yang},
  journal={2023 IEEE International Conference on Bioinformatics and Biomedicine (BIBM)},
  year={2023},
  pages={1588-1595},
  url={https://api.semanticscholar.org/CorpusID:267045003}
}

Introduction

Tumor cell detection plays a vital role in immunohistochemistry (IHC) quantitative analysis. While recent remarkable developments in fullysupervised deep learning have greatly contributed to the efficiency of this task, the necessity for manually annotating all cells of specific detection types remains impractical. Obviously, if we directly use full supervision to train these datasets, it can cause error in loss calculation due to the misclassification of unannotated cells as background. To address this issue, we observe that although some cells are omitted during the annotation process, these unannotated cells have a significant feature similarity with the annotated ones. Leveraging this characteristic, we propose a novel calibrated loss named Semantic Energy Loss (SEL). Specifically, our SEL automatically adjusts the loss to be lower for unannotated regions with similar semantic to the labeled ones, while penalizing regions with lager semantic difference. Besides, to prevent all regions from having similar semantics during training, we propose Stretched Feature Loss (SFL) that widen the semantic distance. We evaluate our method on two different IHC datasets and achieve significant performance improvements in both sparse and exhaustive annotation scenarios. Furthermore, we also validate that our method holds significant potential for detecting multiple types of cells.

Dependecies

In the environment configuration, we primarily utilize torch==1.7.1.

Usage

Data preparation

Before training, you need to prepare the training images and ground truth.

We place the training images in the Her2/img/train, and the corresponding ground truths processed with a Gaussian kernel are placed in the Her2/ground_truth_2classes/train.

Model training and test

To training a model, set related parameters and run python train.py

To evaluate the trained model on the test set, set related parameters in the file caculate_metric.py and run python test.py.

Due to time constraints, the code was not fully reviewed before release. If you have any questions or notice any missing code, I am available to assist you at any time.