EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Efficient Retinal Boundary Detection Method Based on an Improved Canny Operator on OCT Images

Samir Boulkaibet, Meriem Hacini, Ayoub Abderrazak Maarouf, Fella Hachouf, Nadhira Khezami, Bilel Neji, Sofia Mezaache

IEEE Access · 2026 · 14 · 111010-111028

DOI: 10.1109/access.2026.3714563 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Access, 2026. Verfasst von Samir Boulkaibet und 6 weiteren Autor:innen. Die Arbeit nennt 29 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Optical Coherence Tomography (OCT) is an essential imaging tool for retinal visualization and ocular disease diagnosis. Accurate detection of retinal layers is crucial for quantitative analysis and informed clinical decision making. However, existing commercial software may offer limited adaptability to varying imaging conditions and clinical requirements. This paper presents, a novel approach for retinal layers detection in OCT images based on an enhanced version of the classical Canny operator. Initially CLAHE Algorithm is applied to OCT images to enhance their contrast. Subsequently, The Gaussian filter in Canny operator is replaced with a fractional variational filter. In the next stage, fractional order gradients are employed instead of the traditional Sobel operator, providing improved denoising while preserving edge information. Finally, Fuzzy C-Means (FCM) clustering is used to refine boundary delineation and improve segmentation accuracy, thereby reducing false detections. The proposed approach was validated on both healthy and pathological OCT cases, enabled the identification of up to eleven retinal boundaries. Its performance was assessed through both quantitative and qualitative analyses, demonstrating strong agreement with expert annotations. Furthermore, qualitative assessments confirmed the method’s ability to preserve fine structural details while reducing false detections. The results the method’s ability to preserve for accurate and reliable retinal layer detection in clinical settings. In particular, the method demonstrated robust performance in challenging pathological cases such as diabetic macular oedema (DME), age related macular degeneration (AMD), and choroidal neovascularization (CNV), supporting its application in automated ophthalmic diagnostics.

Zitieren

APA

Boulkaibet, S., Hacini, M., Maarouf, A. A., Hachouf, F., Khezami, N., Neji, B., & Mezaache, S. (2026). Efficient Retinal Boundary Detection Method Based on an Improved Canny Operator on OCT Images. IEEE Access, 14, 111010-111028. https://doi.org/10.1109/access.2026.3714563

Vancouver

Boulkaibet S, Hacini M, Maarouf AA, Hachouf F, Khezami N, Neji B, et al. Efficient Retinal Boundary Detection Method Based on an Improved Canny Operator on OCT Images. IEEE Access. 2026;14:111010-111028. doi:10.1109/access.2026.3714563

BibTeX
@article{Boulkaibet2026,
  author = {Boulkaibet, Samir and Hacini, Meriem and Maarouf, Ayoub Abderrazak and Hachouf, Fella and Khezami, Nadhira and Neji, Bilel and Mezaache, Sofia},
  title = {Efficient Retinal Boundary Detection Method Based on an Improved Canny Operator on OCT Images},
  journal = {IEEE Access},
  year = {2026},
  volume = {14},
  pages = {111010--111028},
  doi = {10.1109/access.2026.3714563}
}
Kostenlos starten →