Рақамли технологияларнинг назарий ва амалий масалалари Ҷилди 9 № 3 (2026) · Саҳифаҳои 82-89
Retinal layer segmentation in pathological OKT B-scans based on anatomical surfaces
Юсупов, О.Р., Шамсиева, Х.Г.
Аннотатсия
Pathological changes in optical coherence tomography (OCT) B-scan images complicate the automatic detection of retinal layer boundaries. This study proposes a sequential anatomical surface detection algorithm for extracting the complete retinal layer from pathological OCT images. After speckle noise reduction, the internal limiting membrane (ILM), retinal pigment epithelium (RPE), and lower RPE/BM boundary are sequentially localized according to their anatomical and reflectivity characteristics. The detected surfaces are refined using a local minimum-cost path, and the retinal mask is then generated by fully preserving the region between the ILM and BM boundaries. This approach reduces the risk of losing pathological structures during segmentation, including intraretinal cysts, CNV-related abnormalities, and drusen. The proposed algorithm does not require supervised training and enables the formation of a retinal region of interest for subsequent automated analysis.
оптическая когерентная томографияпатологические B-сканысегментация слоя сетчаткиопределение анатомических поверхностейподавление спекл-шумадинамическое программированиеретинальная область интересаoptical coherence tomographypathological B-scansretinal layer segmentation
Манбаи метамаълумот: бойгонии OAI-PMH-и маҷалла · Sindex матни пурраро нигоҳ намедорад, ба манбаъ пайванд медиҳад.