Hybrid method of image analysis based on artificial intelligence technologies and fuzzy sets

Cover Page

Cite item

Full Text

Open Access Open Access
Restricted Access Access granted
Restricted Access Subscription or Fee Access

Abstract

The paper deals with the development of a prototype of a hybrid intelligent system for image analysis on the example of the task of diagnosis and staging of diabetic retinopathy – a complication of diabetes mellitus, characterized by damage to the retinal vessels. As a result of chronically elevated blood glucose levels, microcirculation is impaired, leading to the development of microaneurysms, exudation, hemorrhage and, in severe cases, neovascularization. This can lead to visual impairment and, ultimately, to blindness in the absence of timely treatment. Detection and staging of the disease are based on the analysis of photographic images of the ocular fundus (fundus images). An overview of the research topic is given, the basis for the advantages of hybrid intelligent systems in comparison with solutions based on the application of a single technology is presented. The steps of creating a system that combines the joint use of classical methods of computer vision, artificial neural networks, elements of fuzzy logic theory and methods of explainable artificial intelligence are described. With the help of combined architecture of the software solution it was possible to achieve flexibility in the issues of applicability of criteria of disease staging, which indicates the broad prospects of such a solution in the diagnosis of other diseases with logically formalizable criteria.

Full Text

Restricted Access

About the authors

A. N. Averkin

Plekhanov Russian University of Economics

Author for correspondence.
Email: averkin2003@inbox.ru
Russian Federation, Moscow

E. N. Volkov

Plekhanov Russian University of Economics

Email: averkin2003@inbox.ru
Russian Federation, Moscow

S. A. Yarushev

Plekhanov Russian University of Economics

Email: averkin2003@inbox.ru
Russian Federation, Moscow

References

  1. Volkov E.N., Averkin A.N. Explainable Artificial Intelligence in Medical Image Analysis: State of the Art and Prospects // XXVI Intern. Conf. on Soft Computing and Measurements (SCM). IEEE, 2023. P. 134–137. https://doi.org/10.1109/SCM58628.2023.10159033
  2. Averkin A.N., Volkov E.N., Yarushev S.A. Possibilities of application of neuro-fuzzy networks for ophthalmologic image classification // Pattern Recognition Image Analysis. 2024. V. 34. № 3. P. 610–616. https://doi.org/10.1134/S1054661824700421
  3. Averkin A.N., Volkov E.N., Yarushev S.A. Explainable artificial intelligence in deep learning neural nets-based digital images analysis //J. Comp. Systems Sci. Int. 2024. V. 63. № 1. P. 175–203. https://doi.org/10.1134/S1064230724700138
  4. Рыжов А.П. О качестве классификации объектов на основе нечетких правил // Интеллектуальные системы. 2005. Т. 9. С. 253–264.
  5. Krzywicki T., Brona P., Zbrzezny A.B. et al. A global review of publicly available datasets containing fundus images: characteristics, barriers to access, usability, and generalizability //J. Clin. Med. 2023. V. 12. № 10. P. 3587. https://doi.org/10.3390/jcm12103587
  6. Jha D., Smedsrud P.H., Riegler M.A. et al. Resunet++: an advanced architecture for medical image segmentation // IEEE Intern. Sympos. Multimedia (ISM). 2019. P. 225–2255.
  7. Van der Velden B.H.M., Kuijf B.H., Gilhuijs H.J. et al. Explainable artificial intelligence (XAI) in deep learning-based medical image analysis // Med. Image Analysis. 2022. V. 79. P. 102470. https://doi.org/10.1016/j.media.2022.102470
  8. Qian J., Li H., Wang J. et al. Recent advances in explainable artificial intelligence for magnetic resonance imaging // Diagnostics. 2023. V. 13. № 9. P. 1571. https://doi.org/10.3390/diagnostics13091571
  9. Volkov E.N., Averkin A.N. Possibilities of explainable artificial intelligence for glaucoma detection using the LIME method as an example // XXVI Intern. Conf. on Soft Computing and Measurements (SCM). IEEE: Saint-Petersburg, 2023. P. 130–133. https://doi.org/10.1109/SCM58628.2023.10159038
  10. Saeed W., Omlin C. Explainable Ai (Xai): a systematic meta-survey of current challenges and future opportunities // Knowledge-Based Systems. 2023. V. 263. P. 110273. https://doi.org/10.1016/j.knosys.2023.110273
  11. Clement T., Kemmerzell N., Abdelaal M. et al. XAIR: a systematic metareview of explainable AI (XAI) aligned to the software development process // Mach. Lear. Knowledge Extraction. 2023. V. 5. № 1. P. 78–108. https://doi.org/10.3390/make5010006
  12. Selvaraju R.R., Cogswell M., Das A. et al. Grad-cam: visual explanations from deep networks via gradient-based localization // Proc. IEEE Intern. Conf. on Computer Vision. Venice, 2017. P. 618–626.
  13. Zhou B., Khosla A., Lapedriza A. et al. Learning deep features for discriminative localization // Proc. IEEE Conf. on Computer Vision and Pattern Recognition. Las Vegas, 2016. P. 2921–2929.
  14. Cheng B., Girshick R., Dollar P. et al. Boundary IoU: improving object-centric image segmentation evaluation // Proc. IEEE/CVF Conf. on Computer Vision and Pattern Recognition. Nashville, USA. 2021. P. 15334–15342.
  15. Zhao R., Qian B., Zhang X. et al. Rethinking dice loss for medical image segmentation // IEEE Intern. Conf. on Data Mining (ICDM). Sorrento, Italy. IEEE, 2020. P. 851–860. https://doi.org/10.1109/ICDM50108.2020.00094
  16. Hehn T., Kooij J., Gavrila D. Fast and compact image segmentation using instance stixels // IEEE Transactions on Intelligent Vehicles. 2021. V. 7. № 1. P. 45–56. https://doi.org/10.1109/TIV.2021.3067223

Supplementary files

Supplementary Files
Action
1. JATS XML
Download
2. Fig. 1. Signs of diabetic retinopathy on the fundus image.

Download (241KB)
Indexing metadata
3. Fig. 2. Structure of the hybrid intelligent system for diagnosing diabetic retinopathy.

Download (387KB)
Indexing metadata
4. Fig. 3. Fundus images in RGB color channels (own data).

Download (95KB)
Indexing metadata
5. Fig. 4. Results of applying the CLAHE filter (own data).

Download (101KB)
Indexing metadata
6. Fig. 5. Examples of images of the APTOS2019 dataset.

Download (103KB)
Indexing metadata
7. Fig. 6. Examples of images of datasets for segmentation: the first row is FDGAR, the second row is IDRID, columns are class masks.

Download (168KB)
Indexing metadata
8. Fig. 7. EfficentNetB0 architecture.

Download (245KB)
Indexing metadata
9. Fig. 8. CenterNet architecture.

Download (202KB)
Indexing metadata
10. Fig. 9. ResUNet++ architecture [3].

Download (673KB)
Indexing metadata
11. Fig. 10. Diagnostic decision surface for the stage of diabetic retinopathy – DDS (own data).

Download (253KB)
Indexing metadata
12. Fig. 11. Segmentation results using ResUnet++ (own data).

Download (114KB)
Indexing metadata

Copyright (c) 2025 Russian Academy of Sciences