Synergistically segmenting choroidal layer and vessel using deep learning for choroid structure analysis

Abstract

Objective. The choroid is the most vascularized structure in the human eye, whose layer structure and vessel distribution are both critical for the physiology of the retina, and disease pathogenesis of the eye. Although some works have used graph-based methods or convolutional neural networks to separate the choroid layer from the outer-choroid structure, few works focused on further distinguishing the inner-choroid structure, such as the choroid vessel and choroid stroma. Approach. Inspired by the multi-task learning strategy, in this paper, we propose a segmentation pipeline for choroid analysis which can separate the choroid layer from other structures and segment the choroid vessel synergistically. The key component of this pipeline is the proposed choroidal U-shape network (CUNet), which catches both correlation features and specific features between the choroid layer and the choroid vessel. Then pixel-wise classification is generated based on these two types of features to obtain choroid layer segmentation and vessel segmentation. Besides, the training process of CUNet is supervised by a proposed adaptive multi-task segmentation (AMS) loss which adds a regularization term that is used to balance the performance of the two tasks. Main results. Experiments show the high performance (4% higher dice score) and less computational complexity (18.85M lower size) of our proposed strategy. Significance. The high performance and generalization on both choroid layer and vessel segmentation indicate the clinical potential of our proposed pipeline.

Semi-supervised contrast learning-based segmentation of choroidal vessel in optical coherence tomography images

“…Moreover, From table 1, in AMD disease, CVI is significantly lower and the normal choroid appears slightly thicker than the AMD diseased choroid. These observations align with the findings reported in Agrawal et al (2016a), Zhu et al (2022). While there is no theoretical explanation for the decrease in CVI, we believe that the precise vessel segmentation obtained through our method can contribute to the field of clinical analysis and provide valuable insights.…”

Choroidal Vascularity Index and Choroidal Structural Changes in Children With Nephrotic Syndrome

“…In total, 45 children with nephrotic syndrome (32 males and 13 females; median age, 8 [3][4][5][6][7][8][9][10][11][12][13][14][15] years; median course of disease, 2.25 years [6 days-11 years]) and 40 normal controls (25 males and 15 females; median age, 9 [5][6][7][8][9][10][11][12][13][14][15][16] years) were included in the study. In patients with nephrotic syndrome, the median serum albumin was 20.90 (12.50-41.60) g/L, the median serum creatinine was 44.1 (24.10-152.2) μmol/L, and the median 24-hour urinary total protein was 2.45 (0.02-13.05) g/24 h. The clinical indicators of the children with nephrotic syndrome are shown in Table 1.…”

“…We used an automatic segmentation method based on deep learn-ing to segment the choroidal vessels and stroma instead of manual segmentation. 14,15 The manually labeled B-scan images are input into two deep convolutional neural networks (DCNNs) to segment the choroidal boundary and vessel; these networks include an encoder stage for extracting highlevel features and a decoder stage for reconstructing the image resolution. After training, the two DCNNs can accurately segment the choroidal boundary and vessels from each B-scan image and calculate the VV, CV, and CVI automatically (Fig.…”