Deep learning-driven automated quality assessment of ultra-widefield optical coherence tomography angiography images for diabetic retinopathy

被引:2
作者
Jin, Yixiao [1 ,2 ]
Gui, Fu [3 ]
Chen, Minghao [1 ,2 ]
Chen, Xiang [4 ]
Li, Haoxuan [5 ]
Zhang, Jingfa [1 ,2 ]
机构
[1] Shanghai Jiao Tong Univ, Sch Med, Shanghai Gen Hosp, Dept Ophthalmol, 100 Hai Ning Rd, Shanghai 200080, Peoples R China
[2] Shanghai Engn Ctr Precise Diag & Treatment Eye Dis, Natl Clin Res Ctr Eye Dis, Shanghai Key Lab Ocular Fundus Dis,Shanghai Key Cl, Shanghai Engn Ctr Visual Sci & Photomed,Shanghai C, Shanghai, Peoples R China
[3] Nanchang Univ, Dept Ophthalmol, Affiliated Hosp 2, Nanchang, Jiangxi, Peoples R China
[4] Shanghai Jiao Tong Univ, Dept Comp Sci & Engn, Shanghai, Peoples R China
[5] Shanghai Univ Sport, Sch Exercise & Hlth, Sports Engn, 650 Qingyuan Ring Rd, Shanghai 200438, Peoples R China
关键词
Deep learning; Image quality assessment (IQA); Transfer learning; Diabetic retinopathy; Optical coherence tomography angiography (OCTA); Ultra-widefield optical coherence tomography angiography (UW-OCTA); ARTIFICIAL-INTELLIGENCE; VESSEL SEGMENTATION; MOTION CORRECTION;
D O I
10.1007/s00371-024-03383-6
中图分类号
TP31 [计算机软件];
学科分类号
081202 ; 0835 ;
摘要
Image quality assessment (IQA) of fundus images constitutes a foundational step in automated disease analysis. This process is pivotal in supporting the automation of screening, diagnosis, follow-up, and related academic research for diabetic retinopathy (DR). This study introduced a deep learning-based approach for IQA of ultra-widefield optical coherence tomography angiography (UW-OCTA) images of patients with DR. Given the novelty of ultra-widefield technology, its limited prevalence, the high costs associated with equipment and operational training, and concerns regarding ethics and patient privacy, UW-OCTA datasets are notably scarce. To address this, we initially pre-train a vision transformer (ViT) model on a dataset comprising 6 mm x 6 mm OCTA images, enabling the model to acquire a fundamental understanding of OCTA image characteristics and quality indicators. Subsequent fine-tuning on 12 mm x 12 mm UW-OCTA images aims to enhance accuracy in quality assessment. This transfer learning strategy leverages the generic features learned during pre-training and adjusts the model to evaluate UW-OCTA image quality effectively. Experimental results demonstrate that our proposed method achieves superior performance compared to ResNet18, ResNet34, and ResNet50, with an AUC of 0.9026 and a Kappa value of 0.7310. Additionally, ablation studies, including the omission of pre-training on 6 mm x 6 mm OCTA images and the substitution of the backbone network with the ViT base version, resulted in varying degrees of decline in AUC and Kappa values, confirming the efficacy of each module within our methodology.
引用
收藏
页码:1049 / 1059
页数:11
相关论文
共 45 条
[1]   Automatic detection of microaneurysms in optical coherence tomography images of retina using convolutional neural networks and transfer learning [J].
Almasi, Ramin ;
Vafaei, Abbas ;
Kazeminasab, Elahe ;
Rabbani, Hossein .
SCIENTIFIC REPORTS, 2022, 12 (01)
[2]   Estimating Rates of Progression and Predicting Future Visual Fields in Glaucoma Using a Deep Variational Autoencoder [J].
Berchuck, Samuel I. ;
Mukherjee, Sayan ;
Medeiros, Felipe A. .
SCIENTIFIC REPORTS, 2019, 9 (1)
[3]  
Brock Andrew, 2018, arXiv
[4]   Evaluation of artifact reduction in optical coherence tomography angiography with real-time tracking and motion correction technology [J].
Camino, Acner ;
Zhang, Miao ;
Gao, Simon S. ;
Hwang, Thomas S. ;
Sharma, Utkarsh ;
Wilson, David J. ;
Huang, David ;
Jia, Yali .
BIOMEDICAL OPTICS EXPRESS, 2016, 7 (10) :3905-3915
[5]   Transformer Interpretability Beyond Attention Visualization [J].
Chefer, Hila ;
Gur, Shir ;
Wolf, Lior .
2021 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION, CVPR 2021, 2021, :782-791
[6]   Diabetic retinopathy [J].
Cheung, Ning ;
Mitchell, Paul ;
Wong, Tien Yin .
LANCET, 2010, 376 (9735) :124-136
[7]   Comparison of widefield swept-source optical coherence tomography angiography with ultra-widefield colour fundus photography and fluorescein angiography for detection of lesions in diabetic retinopathy [J].
Cui, Ying ;
Zhu, Ying ;
Wang, Jay C. ;
Lu, Yifan ;
Zeng, Rebecca ;
Katz, Raviv ;
Vingopoulos, Filippos ;
Le, Rongrong ;
Lains, Ines ;
Wu, David M. ;
Eliott, Dean ;
Vavvas, Demetrios G. ;
Husain, Deeba ;
Miller, Joan W. ;
Kim, Leo A. ;
Miller, John B. .
BRITISH JOURNAL OF OPHTHALMOLOGY, 2021, 105 (04) :577-581
[8]   A deep learning system for detecting diabetic retinopathy across the disease spectrum [J].
Dai, Ling ;
Wu, Liang ;
Li, Huating ;
Cai, Chun ;
Wu, Qiang ;
Kong, Hongyu ;
Liu, Ruhan ;
Wang, Xiangning ;
Hou, Xuhong ;
Liu, Yuexing ;
Long, Xiaoxue ;
Wen, Yang ;
Lu, Lina ;
Shen, Yaxin ;
Chen, Yan ;
Shen, Dinggang ;
Yang, Xiaokang ;
Zou, Haidong ;
Sheng, Bin ;
Jia, Weiping .
NATURE COMMUNICATIONS, 2021, 12 (01)
[9]   Clinically applicable deep learning for diagnosis and referral in retinal disease [J].
De Fauw, Jeffrey ;
Ledsam, Joseph R. ;
Romera-Paredes, Bernardino ;
Nikolov, Stanislav ;
Tomasev, Nenad ;
Blackwell, Sam ;
Askham, Harry ;
Glorot, Xavier ;
O'Donoghue, Brendan ;
Visentin, Daniel ;
van den Driessche, George ;
Lakshminarayanan, Balaji ;
Meyer, Clemens ;
Mackinder, Faith ;
Bouton, Simon ;
Ayoub, Kareem ;
Chopra, Reena ;
King, Dominic ;
Karthikesalingam, Alan ;
Hughes, Cian O. ;
Raine, Rosalind ;
Hughes, Julian ;
Sim, Dawn A. ;
Egan, Catherine ;
Tufail, Adnan ;
Montgomery, Hugh ;
Hassabis, Demis ;
Rees, Geraint ;
Back, Trevor ;
Khaw, Peng T. ;
Suleyman, Mustafa ;
Cornebise, Julien ;
Keane, Pearse A. ;
Ronneberger, Olaf .
NATURE MEDICINE, 2018, 24 (09) :1342-+
[10]  
Devlin J, 2019, 2019 CONFERENCE OF THE NORTH AMERICAN CHAPTER OF THE ASSOCIATION FOR COMPUTATIONAL LINGUISTICS: HUMAN LANGUAGE TECHNOLOGIES (NAACL HLT 2019), VOL. 1, P4171