Skip to main content
Fig. 2 | BMC Medicine

Fig. 2

From: Deep learning-based six-type classifier for lung cancer and mimics from histopathological whole slide images: a retrospective study

Fig. 2

High AUCs achieved across multiple cohorts. AUC was utilized to measure the performance of the model on different testing cohorts, including a the subset of the initial cohort SYSU1, b an independent internal cohort SYSU2, c an external cohort from Shenzhen People’s Hospital (SZPH) that contained 4 types of lung tissues, and d a public cohort from ‘TCGA’ which was actually a subset consisting of slides randomly selected from the TCGA-LUAD and TCGA-LUSC projects. Blind tests were conducted on all the cohorts by four pathologists of three levels (Pathologist1 is senior attending, Pathologist2 and Pathologist3 are junior attending, and Pathologist4 is junior); performance of each pathologist on each cohort was depicted as a star in a–d, respectively

Back to article page