Translational vision science & technology · 2026
Time Series Segmentation in Adaptive Optics Ophthalmoscopy With Application to Flicker-Induced Vasodilation.
СтатьяНазвание и аннотация: Перевод пока не подготовлен
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
Vard Alireza, Aissa Abir, Rossant Florence, Bloch Isabelle, Castro-Farias Daniela, Chabriat Hugues, Paques Michel
Аннотация
PURPOSE: To develop an automated image processing pipeline to segment retinal arterial walls in a series of adaptive optics ophthalmoscopy (AOO) images, overcoming difficulties related to various quality levels within the series, to accurately evaluate morphometric differences.
METHODS: We first assess the quality of AOO images by deep learning in order to select the best quality image in the sequence. Then, we segment this image by combining deep learning and dedicated active contour models. Finally, we segment the other images of the sequence using the best image segmentation as a regularization exploiting the temporal continuity to compensate for lower quality.
RESULTS: We applied the proposed method to flicker series, a challenging use-case given the image quality heterogeneity. The quality assessment method achieves a Spearman correlation coefficient of 0.86, demonstrating very good agreement with the quality ranking established by the medical expert. The automatic segmentation method results in accurate segmentations with a mean root mean square error of 0.5 pixel and 1.8 pixels for the inner and outer borders of the arterial wall, respectively. This leads to accurate and unbiased estimates of the wall to lumen ratio.
CONCLUSIONS: Our algorithm can be applied as a reliable and useful computer-aided tool to process large time series databases in AOO, and can help experts in longitudinal monitoring applications.
TRANSLATIONAL RELEVANCE: For the first time, AOO image series can be processed with low supervision to accurately monitor morphometric changes in retinal vascularization.
METHODS: We first assess the quality of AOO images by deep learning in order to select the best quality image in the sequence. Then, we segment this image by combining deep learning and dedicated active contour models. Finally, we segment the other images of the sequence using the best image segmentation as a regularization exploiting the temporal continuity to compensate for lower quality.
RESULTS: We applied the proposed method to flicker series, a challenging use-case given the image quality heterogeneity. The quality assessment method achieves a Spearman correlation coefficient of 0.86, demonstrating very good agreement with the quality ranking established by the medical expert. The automatic segmentation method results in accurate segmentations with a mean root mean square error of 0.5 pixel and 1.8 pixels for the inner and outer borders of the arterial wall, respectively. This leads to accurate and unbiased estimates of the wall to lumen ratio.
CONCLUSIONS: Our algorithm can be applied as a reliable and useful computer-aided tool to process large time series databases in AOO, and can help experts in longitudinal monitoring applications.
TRANSLATIONAL RELEVANCE: For the first time, AOO image series can be processed with low supervision to accurately monitor morphometric changes in retinal vascularization.