You are here

  • Home
  • Online First
  • Deep learning-based image quality assessment for optical coherence tomography macular scans: a multicentre study

Article Text

Article menu
Download PDFPDF
Lens and cataract
Deep learning-based image quality assessment for optical coherence tomography macular scans: a multicentre study
  1. Ziqi Tang1,
  2. Xi Wang2,3,
  3. An Ran Ran1,
  4. Dawei Yang1,
  5. Anni Ling1,
  6. Jason C Yam1,4,
  7. Xiujuan Zhang1,
  8. Simon K H Szeto1,4,
  9. Jason Chan1,4,
  10. Cherie Y K Wong1,4,
  11. Vivian W K Hui1,4,
  12. Carmen K M Chan1,4,
  13. Tien Yin Wong5,6,
  14. Ching-Yu Cheng7,8,
  15. Charumathi Sabanayagam7,9,
  16. Yih Chung Tham7,8,
  17. Gerald Liew10,
  18. Giridhar Anantharaman11,
  19. Rajiv Raman12,
  20. Yu Cai13,
  21. Haoxuan Che14,
  22. Luyang Luo3,
  23. Quande Liu3,
  24. Yiu Lun Wong1,
  25. Amanda K Y Ngai1,
  26. Vincent L Yuen1,
  27. Nelson Kei15,
  28. Timothy Y Y Lai1,
  29. Hao Chen14,16,
  30. Clement C Tham1,4,
  31. Pheng-Ann Heng3,17,
  32. Carol Y Cheung1
  1. 1Department of Ophthalmology and Visual Sciences, The Chinese University of Hong Kong, Hong Kong SAR, China
  2. 2Zhejiang Lab, Hangzhou, Zhejiang, China
  3. 3Department of Computer Science and Engineering, The Chinese University of Hong Kong, Hong Kong SAR, China
  4. 4Hong Kong Eye Hospital, Hong Kong SAR, China
  5. 5Tsinghua Medicine, Tsinghua University, Beijing, China
  6. 6School of Clinical Medicine, Beijing Tsinghua Changgung Hospital, Beijing, China
  7. 7Singapore Eye Research Institute, Singapore National Eye Centre, Singapore
  8. 8Centre for Innovation and Precision Eye Health, Department of Ophthalmology, Yong Loo Lin School of Medicine, National University of Singapore, Singapore
  9. 9Ophthalmology and Visual Sciences Academic Clinical Program, Duke-NUS Medical School, Singapore
  10. 10Department of Ophthalmology, Westmead Institute for Medical Research, University of Sydney, Sydney, New South Wales, Australia
  11. 11Giridhar Eye Institute, Cochin, India
  12. 12Shri Bhagwan Mahavir Vitreoretinal Services, Sankara Nethralaya, Chennai, India
  13. 13Department of Electronic and Computer Engineering, The Hong Kong University of Science and Technology, Hong Kong SAR, China
  14. 14Department of Computer Science and Engineering, The Hong Kong University of Science and Technology, Hong Kong SAR, China
  15. 15School of Life Sciences, The Chinese University of Hong Kong, Hong Kong SAR, China
  16. 16Department of Chemical and Biological Engineering, The Hong Kong University of Science and Technology, Hong Kong SAR, China
  17. 17Institute of Medical Intelligence and XR, The Chinese University of Hong Kong, Hong Kong SAR, China
  1. Correspondence to Professor Pheng-Ann Heng, Department of Computer Science and Engineering, The Chinese University of Hong Kong, Hong Kong SAR, China; pheng{at}cse.cuhk.edu.hk; Dr Carol Y Cheung, Department of Ophthalmology and Visual Sciences, The Chinese University of Hong Kong, Hong Kong SAR, China; carolcheung{at}cuhk.edu.hk

Abstract

Aims To develop and externally test deep learning (DL) models for assessing the image quality of three-dimensional (3D) macular scans from Cirrus and Spectralis optical coherence tomography devices.

Methods We retrospectively collected two data sets including 2277 Cirrus 3D scans and 1557 Spectralis 3D scans, respectively, for training (70%), fine-tuning (10%) and internal validation (20%) from electronic medical and research records at The Chinese University of Hong Kong Eye Centre and the Hong Kong Eye Hospital. Scans with various eye diseases (eg, diabetic macular oedema, age-related macular degeneration, polypoidal choroidal vasculopathy and pathological myopia), and scans of normal eyes from adults and children were included. Two graders labelled each 3D scan as gradable or ungradable, according to standardised criteria. We used a 3D version of the residual network (ResNet)-18 for Cirrus 3D scans and a multiple-instance learning pipline with ResNet-18 for Spectralis 3D scans. Two deep learning (DL) models were further tested via three unseen Cirrus data sets from Singapore and five unseen Spectralis data sets from India, Australia and Hong Kong, respectively.

Results In the internal validation, the models achieved the area under curves (AUCs) of 0.930 (0.885–0.976) and 0.906 (0.863–0.948) for assessing the Cirrus 3D scans and Spectralis 3D scans, respectively. In the external testing, the models showed robust performance with AUCs ranging from 0.832 (0.730–0.934) to 0.930 (0.906–0.953) and 0.891 (0.836–0.945) to 0.962 (0.918–1.000), respectively.

Conclusions Our models could be used for filtering out ungradable 3D scans and further incorporated with a disease-detection DL model, allowing a fully automated eye disease detection workflow.

  • Retina
  • Imaging
  • Macula

Data availability statement

No data are available.

https://doi.org/10.1136/bjo-2023-323871

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    Data availability statement

    No data are available.

    View Full Text

    Footnotes

    • P-AH and CYC are joint senior authors.

    • X @KikiTang13, @JasonYam7, @Yih Chung Tham

    • ZT and XW contributed equally.

    • Contributors ZT, ARR and DY supervised by CYC conceived and designed the study. JCY, XZ, SKHS, JC, CYKW, CKMC, TYW, C-YC, CS, YCT, GL, GA and RR provided data for model development, validation and testing. ZT, AL, AKYN, VLY and NK organised all the data. ZT, YLW, AKYN, VLY and NK contributed to the imaging labelling. XW developed and validated the deep learning system supervised by P-AH with clinical input from CYC, ZT, ARR, DY, SKHS, TYYL and CCT. YC and HChe contributed to the deep learning method discussion, deep learning method development and coding. LL and QL contributed to method discussion and data analysis. HChen provided guidance and supervision on model development. ZT did the statistical analysis and interpreted the results. ZT and XW wrote the initial draft. P-AH and CYC contributed equally to the work as senior authors. CYC is guarantor.

    • Funding Bright Focus Foundation (Reference No. A2018093S). Innovation and Technology Fund, Hong Kong SAR, China (MRP/056/20X). Research Grants Council of the Hong Kong Special Administrative Region, China (Project Reference Number: T45-401/22-N)

    • Competing interests None declared.

    • Provenance and peer review Not commissioned; externally peer reviewed.

    • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.