CorNet: Autonomous feature learning in raw Corvis ST data for keratoconus diagnosis via residual CNN approach

PeiPei Zhang; LanTing Yang; YiCheng Mao; XinYu Zhang; JiaXuan Cheng; YuanYuan Miao; FangJun Bao; ShiHao Chen; QinXiang Zheng; JunJie Wang

doi:10.1016/j.compbiomed.2024.108286

CorNet: Autonomous feature learning in raw Corvis ST data for keratoconus diagnosis via residual CNN approach

Comput Biol Med. 2024 Apr:172:108286. doi: 10.1016/j.compbiomed.2024.108286. Epub 2024 Mar 13.

Authors

Affiliations

¹ School of Ophthalmology and Optometry, Eye Hospital, Wenzhou Medical University, Wenzhou, 325027, China.
² School of Ophthalmology and Optometry, Eye Hospital, Wenzhou Medical University, Wenzhou, 325027, China; National Clinical Research Center for Ocular Diseases, Eye Hospital, Wenzhou Medical University, Wenzhou, 325027, China.
³ School of Ophthalmology and Optometry, Eye Hospital, Wenzhou Medical University, Wenzhou, 325027, China; National Clinical Research Center for Ocular Diseases, Eye Hospital, Wenzhou Medical University, Wenzhou, 325027, China. Electronic address: csh@eye.ac.cn.
⁴ School of Ophthalmology and Optometry, Eye Hospital, Wenzhou Medical University, Wenzhou, 325027, China; National Clinical Research Center for Ocular Diseases, Eye Hospital, Wenzhou Medical University, Wenzhou, 325027, China. Electronic address: qinxiangzheng@wmu.edu.cn.
⁵ School of Ophthalmology and Optometry, Eye Hospital, Wenzhou Medical University, Wenzhou, 325027, China; National Engineering Research Center of Ophthalmology and Optometry, Eye Hospital, Wenzhou Medical University, Wenzhou, 325027, China; Department of Ophthalmology, The Third Hospital of Mianyang, Sichuan Mental Health Center, Mianyang, 621054, China. Electronic address: junjie.wang@wmu.edu.cn.

PMID: 38493602
DOI: 10.1016/j.compbiomed.2024.108286

Abstract

Purpose: To ascertain whether the integration of raw Corvis ST data with an end-to-end CNN can enhance the diagnosis of keratoconus (KC).

Method: The Corvis ST is a non-contact device for in vivo measurement of corneal biomechanics. The CorNet was trained and validated on a dataset consisting of 1786 Corvis ST raw data from 1112 normal eyes and 674 KC eyes. Each raw data consists of the anterior and posterior corneal surface elevation during air-puff induced dynamic deformation. The architecture of CorNet utilizes four ResNet-inspired convolutional structures that employ 1 × 1 convolution in identity mapping. Gradient-weighted Class Activation Mapping (Grad-CAM) was adopted to visualize the attention allocation to diagnostic areas. Discriminative performance was assessed using metrics including the AUC of ROC curve, sensitivity, specificity, precision, accuracy, and F1 score.

Results: CorNet demonstrated outstanding performance in distinguishing KC from normal eyes, achieving an AUC of 0.971 (sensitivity: 92.49%, specificity: 91.54%) in the validation set, outperforming the best existing Corvis ST parameters, namely the Corvis Biomechanical Index (CBI) with an AUC of 0.947, and its updated version for Chinese populations (cCBI) with an AUC of 0.963. Though the ROC curve analysis showed no significant difference between CorNet and cCBI (p = 0.295), it indicated a notable difference between CorNet and CBI (p = 0.011). The Grad-CAM visualizations highlighted the significance of corneal deformation data during the loading phase rather than the unloading phase for KC diagnosis.

Conclusion: This study proposed an end-to-end CNN approach utilizing raw biomechanical data by Corvis ST for KC detection, showing effectiveness comparable to or surpassing existing parameters provided by Corvis ST. The CorNet, autonomously learning comprehensive temporal and spatial features, demonstrated a promising performance for advancing KC diagnosis in ophthalmology.

Keywords: Convolutional neural network; Corneal biomechanics; Corvis ST; Keratoconus.

MeSH terms

Biomechanical Phenomena
Cornea / diagnostic imaging
Corneal Topography
Humans
Keratoconus* / diagnosis
ROC Curve