GABNet: global attention block for retinal OCT disease classification

Xuan Huang; Zhuang Ai; Hui Wang; Chongyang She; Jing Feng; Qihao Wei; Baohai Hao; Yong Tao; Yaping Lu; Fanxin Zeng

doi:10.3389/fnins.2023.1143422

GABNet: global attention block for retinal OCT disease classification

Front Neurosci. 2023 Jun 2:17:1143422. doi: 10.3389/fnins.2023.1143422. eCollection 2023.

Authors

Xuan Huang^#^{1

2}, Zhuang Ai^#³, Hui Wang^#¹, Chongyang She¹, Jing Feng¹, Qihao Wei³, Baohai Hao⁴, Yong Tao¹, Yaping Lu³, Fanxin Zeng^{5

6}

Affiliations

¹ Department of Ophthalmology, Beijing Chaoyang Hospital, Capital Medical University, Beijing, China.
² Medical Research Center, Beijing Chaoyang Hospital, Capital Medical University, Beijing, China.
³ Sinopharm Genomics Technology Co., Ltd., Changzhou, Jiangsu, China.
⁴ AI-Farm (Nanjing) Big Data Services Co., Ltd., Nanjing, China.
⁵ Department of Clinical Research Center, Dazhou Central Hospital, Dazhou, Sichuan, China.
⁶ Department of Big Data and Biomedical AI, College of Future Technology, Peking University, Beijing, China.

^# Contributed equally.

Abstract

Introduction: The retina represents a critical ocular structure. Of the various ophthalmic afflictions, retinal pathologies have garnered considerable scientific interest, owing to their elevated prevalence and propensity to induce blindness. Among clinical evaluation techniques employed in ophthalmology, optical coherence tomography (OCT) is the most commonly utilized, as it permits non-invasive, rapid acquisition of high-resolution, cross-sectional images of the retina. Timely detection and intervention can significantly abate the risk of blindness and effectively mitigate the national incidence rate of visual impairments.

Methods: This study introduces a novel, efficient global attention block (GAB) for feed forward convolutional neural networks (CNNs). The GAB generates an attention map along three dimensions (height, width, and channel) for any intermediate feature map, which it then uses to compute adaptive feature weights by multiplying it with the input feature map. This GAB is a versatile module that can seamlessly integrate with any CNN, significantly improving its classification performance. Based on the GAB, we propose a lightweight classification network model, GABNet, which we develop on a UCSD general retinal OCT dataset comprising 108,312 OCT images from 4686 patients, including choroidal neovascularization (CNV), diabetic macular edema (DME), drusen, and normal cases.

Results: Notably, our approach improves the classification accuracy by 3.7% over the EfficientNetV2B3 network model. We further employ gradient-weighted class activation mapping (Grad-CAM) to highlight regions of interest on retinal OCT images for each class, enabling doctors to easily interpret model predictions and improve their efficiency in evaluating relevant models.

Discussion: With the increasing use and application of OCT technology in the clinical diagnosis of retinal images, our approach offers an additional diagnostic tool to enhance the diagnostic efficiency of clinical OCT retinal images.

Keywords: GABNet; attention mechanism; model visualization; retinal OCT; retinal disease classification.

Grants and funding

This work was supported in parts by a grant from the National Natural Science Foundation of China to FZ (81902861), a grant from the National Natural Science Foundation of China (32000485) and the Beijing Chaoyang Hospital Science and Technology Innovation Fund (22kcjjyb-11), both awarded to XH, a grant from the National Natural Science Foundation of China to HW (62006161), and the Sinopharm Genomics Technology Co., Ltd. The funder Sinopharm Genomics Technology Co., Ltd. had the following involvement with the study: design, collection, analysis, interpretation of data, the writing of this article, and the decision to submit it for publication.