Unleashing the strengths of unlabelled data in deep learning-assisted pan-cancer abdominal organ quantification: the FLARE22 challenge
Jun Ma, Yao Zhang, Song Gu, Ge Cheng, Shihao Mae, Adamo Young, Cheng Zhu, Xin Yang, Kangkang Meng, Ziyan Huang, Fan Zhang, YuanKe Pan, Shoujin Huang, Jiacheng Wang, Mingze Sun, Rongguo Zhang, Dengqiang Jia, Jae Won Choi, Natália Alves, Bram de Wilde, Gregor Koehler, Haoran Lai, Ershuai Wang, Manuel Wiesenfarth, Qiongjie Zhu, Guoqiang Dong, Jian He, Junjun He, Hua Yang, Bingding Huang, Mengye Lyu, Yongkang Ma, Heng Guo, Weixin Xu, Klaus Maier-Hein, Yajun Wu, Bo Wang
Abstract
Deep learning has shown great potential to automate abdominal organ segmentation and quantification. However, most existing algorithms rely on expert annotations and do not have comprehensive evaluations in real-world multinational settings. To address these limitations, we organised the FLARE 2022 challenge to benchmark fast, low-resource, and accurate abdominal organ segmentation algorithms. We first constructed an intercontinental abdomen CT dataset from more than 50 clinical research groups. We then independently validated that deep learning algorithms achieved a median dice similarity coefficient (DSC) of 90·0% (IQR 87·4-91·3%) by use of 50 labelled images and 2000 unlabelled images, which can substantially reduce manual annotation costs. The best-performing algorithms successfully generalised to holdout external validation sets, achieving a median DSC of 89·4% (85·2-91·3%), 90·0% (84·3-93·0%), and 88·5% (80·9-91·9%) on North American, European, and Asian cohorts, respectively. These algorithms show the potential to use unlabelled data to boost performance and alleviate annotation shortages for modern artificial intelligence models.