Asynchronous feature regularization and cross-modal distillation for OCT based glaucoma diagnosis. (December 2022)

• Record Type: Journal Article
• Authors: Song, Diping; Li, Fei; Li, Cheng; Xiong, Jian; He, Junjun; Zhang, Xiulan; Qiao, Yu
• Is Part Of: Computers in biology and medicine. Volume 151:Part B(2022)
• Publication Date: 2022-12
• Subjects: Deep learning -- Glaucoma diagnosis -- Cross-modal distillation -- Convolutional neural networks -- Optical Coherence Tomography; Medicine -- Data processing -- Periodicals; Biology -- Data processing -- Periodicals; 610.285
• DOI: 10.1016/j.compbiomed.2022.106283
• Languages: English
• ISSNs: 0010-4825

Abstract

Abstract: Glaucoma has become a major cause of vision loss. Early-stage diagnosis of glaucoma is critical for treatment planning to avoid irreversible vision damage. Meanwhile, interpreting the rapidly accumulated medical data from ophthalmic exams is cumbersome and resource-intensive. Therefore, automated methods are highly desired to assist ophthalmologists in achieving fast and accurate glaucoma diagnosis. Deep learning has achieved great successes in diagnosing glaucoma by analyzing data from different kinds of tests, such as peripapillary optical coherence tomography (OCT) and visual field (VF) testing. Nevertheless, applying these developed models to clinical practice is still challenging because of various limiting factors. OCT models present worse glaucoma diagnosis performances compared to those achieved by OCT&VF based models, whereas VF is time-consuming and highly variable, which can restrict the wide employment of OCT&VF models. To this end, we develop a novel deep learning framework that leverages the OCT&VF model to enhance the performance of the OCT model. To transfer the complementary knowledge from the structural and functional assessments to the OCT model, a cross-modal knowledge transfer method is designed by integrating a designed distillation loss and a proposed asynchronous feature regularization (AFR) module. We demonstrate the effectiveness of the proposed method for glaucoma diagnosis by utilizing a public OCT&VF dataset and evaluating it on an external OCT dataset. Our final model with only OCT inputs achieves the accuracy of 87.4% (3.1% absolute improvement) and AUC of 92.3%, which are on par with the OCT&VF joint model. Moreover, results on the external dataset sufficiently indicate the effectiveness and generalization capability of our model. Highlights: A cross-modal knowledge distillation based approach is proposed to benefit fast glaucoma screening in the real world by removing the barrier of multimodal data requirements. A novel asynchronous feature regularization (AFR) module is proposed, which helps to improve the efficiency of knowledge distillation and mitigate overfitting. The developed OCT single-modal model achieves significantly better performance than the regular OCT model and performs on par with the OCT&VF multimodal model. Results on the external dataset indicate the effectiveness and generalization capability of the developed model.