AUTHOR=Li Hansheng , Kang Yuxin , Yang Wentao , Wu Zhuoyue , Shi Xiaoshuang , Liu Feihong , Liu Jianye , Hu Lingyu , Ma Qian , Cui Lei , Feng Jun , Yang Lin 

TITLE=A Robust Training Method for Pathological Cellular Detector via Spatial Loss Calibration

JOURNAL=Frontiers in Medicine

VOLUME=Volume 8 - 2021

YEAR=2021

URL=https://www.frontiersin.org/journals/medicine/articles/10.3389/fmed.2021.767625

DOI=10.3389/fmed.2021.767625

ISSN=2296-858X

ABSTRACT=Computer-aided diagnosis of pathological images usually requires to detect and examine all positive cells for accurate diagnosis. However, cellular datasets tend to be sparsely annotated due to the challenge of annotating all the cells. Obviously, sparse annotations may lead to a seriously miscalculated loss in training, limiting the performance of networks. Thus, efficient and reliable methods for training cellular detectors on sparse annotations are in higher demand than ever. Here, we propose a training method that utilizes regression boxes' spatial information to conduct loss calibration to reduce the miscalculated loss. Extensive experimental results show that our method can significantly boost detectors' performance trained on datasets with varying degrees of sparse annotations. Even if 90% of the annotations are missing, the performance of our method will hardly be affected. Furthermore, we find that the middle layers of the detector are closely related to the generalization performance. More generally, this study could elucidate the link between layers and generalization-performance, provide enlightenment for future research, such as designing and applying constraint rules to specific layers according to gradient analysis to achieve “scalpel-level" model training.