AUTHOR=Wu Xing , Xu Di , Ma Tong , Li Zhao Hui , Ye Zi , Wang Fei , Gao Xiang Yang , Wang Bin , Chen Yu Zhong , Wang Zhao Hui , Chen Ji Li , Hu Yun Tao , Ge Zong Yuan , Wang Da Jiang , Zeng Qiang 

TITLE=Artificial Intelligence Model for Antiinterference Cataract Automatic Diagnosis: A Diagnostic Accuracy Study

JOURNAL=Frontiers in Cell and Developmental Biology

VOLUME=Volume 10 - 2022

YEAR=2022

URL=https://www.frontiersin.org/journals/cell-and-developmental-biology/articles/10.3389/fcell.2022.906042

DOI=10.3389/fcell.2022.906042

ISSN=2296-634X

ABSTRACT=Background: Cataract is the leading cause of blindness worldwide. In order to achieve large-scale cataract screening, several studies have applied artificial intelligence (AI) to cataract detection based on fundus images and achieve remarkable performance. However, fundus images they used are original from normal optical circumstances, which is less impractical due to the existence of poor-quality fundus images for inappropriate optical conditions in actual scenarios. Furthermore, these poor-quality images are easily mistaken as cataract because both appear the fuzzy imaging characteristics, which may decline the performance of cataract detection. We therefore aimed to develop and validate an anti-interference AI model for the rapid and efficient diagnosis based on fundus images.
Materials and Methods: The datasets (including both cataract and non-cataract labels) were derived from the Chinese PLA general hospital. The anti-interference AI model consisted of two AI sub-modules, a quality recognition model for cataract labeling and a CNN-based model for cataract classification. The quality recognition model was performed to distinguish poor-quality images from normal-quality images and further generate the pseudo labels related to image quality for non-cataract. Through this, original binary-class label (cataract and non-cataract) was adjusted to three categories (cataract, non-cataract with normal-quality images, and non-cataract with poor-quality images), which could be used to guide the model to distinguish cataract from suspected cataract fundus images. In the cataract classification stage, the convolutional-neural-network-based model was proposed to classify cataract based on the label of the previous stage. The performance of the model was internally validated and externally tested in real-world setting and the evaluation indicators included area under the receiver operating curve (AUC), accuracy (ACC), sensitivity (SEN) and specificity (SPE).
Results: In the internal and external validation, the anti-interference AI model showed robust performance in cataract diagnosis (three classifications with AUCs >91%, ACCs>84%, SENs>71% and SPEs>89%). Compared with the model that was trained on the binary-class label, the anti-interference cataract model improved the performance by 10 percent. 
Conclusion: We proposed an efficient anti-interference AI model for cataract diagnosis, which could achieve accurate cataract screening even with the interference of poor-quality images and help the government formulate a more accurate aid policy.