AUTHOR=Fan Zhenliang , Yang Qiaorui , Xia Hong , Zhang Peipei , Sun Ke , Yang Mengfan , Yin Riping , Zhao Dongxue , Ma Hongzhen , Shen Yiwei , Fan Junfen 

TITLE=Artificial intelligence can accurately distinguish IgA nephropathy from diabetic nephropathy under Masson staining and becomes an important assistant for renal pathologists

JOURNAL=Frontiers in Medicine

VOLUME=Volume 10 - 2023

YEAR=2023

URL=https://www.frontiersin.org/journals/medicine/articles/10.3389/fmed.2023.1066125

DOI=10.3389/fmed.2023.1066125

ISSN=2296-858X

ABSTRACT=IgA nephropathy (IgAN) and diabetic nephropathy (DN) are the most prevalent primary and secondary glomerular diseases, respectively. Both two diseases can be manifested as mesangial hyperplasia in pathology, which leads to difficulties in distinguishing them, especially when clinical data are lacking. Artificial intelligence (AI) has shown its advantages in pathological diagnosis, and in this study, we also try to use AI to distinguish IgAN and DN. We selected patients diagnosed with IgAN or DN in the First Affiliated Hospital of Zhejiang Chinese Medicine University from September 1, 2020 to April 30, 2022 as the training set. Patients diagnosed between May 1 and June 30, 2022 were used as the test set. We selected glomeruli at 200x magnification under Masson staining, all in 1,000×1,000 pixels. The Yolov5 6.1 algorithm were used to train the model. We tested the model with test set and compared it with renal pathologists. The accuracy of AI reached 73.24%, the accuracy of IgAN reached 77.27% and the accuracy of DN reached 69.59%. Compared with pathologists, AI can distinguish IgAN from DN faster and more accurately, making it the best assistant for renal pathologists.