Date Published: March 7, 2018
Publisher: Public Library of Science
Author(s): Sushmitha Rao Uppugunduri, Mohammed Abdul Rasheed, Ashutosh Richhariya, Soumya Jana, Jay Chhablani, Kiran Kumar Vupparaboina, Gianni Virgili.
To develop an algorithm for automated quantification of Haller’s layer in choroid using swept-source optical coherence tomography (OCT).
So far, to understand the association of various diseases with structural changes of choroid, only gross indicators such as thickness, volume and vascularity index have been examined. However, certain diseases affect specific sublayers of the choroid. Accordingly, a need for targeted quantitation arises. In particular, there is significant interest in understanding Haller’s layer, a choroidal sublayer comprising relatively large blood vessels. Unfortunately, its intricate vasculature makes, manual quantitation difficult, tedious, and error-prone. To surmount this difficulty, it is imperative to develop an algorithmic method.
The primary contribution of this work consists in developing an approach for detecting the boundary between Haller’s and Sattler’s layers, the latter comprising medium-sized vessels. The proposed algorithm estimates vessel cross-sections using exponentiation-based binarization, and labels a vessel large if its cross-section exceeds certain statistically determined threshold. Finally, the desired boundary is obtained as a smooth curve joining the innermost points of such large vessels. On 50 OCT B-scans (of 50 healthy eyes), our algorithm was validated both qualitatively and quantitatively, by comparing with intra-observer variability. Extensive statistical analysis was performed using metrics including Dice coefficient (DC), correlation coefficient (CC) and absolute difference (AD).
The proposed algorithm achieved a mean DC of 89.48% (SD:5.03%) in close agreement with the intra-observer repeatability of 89.12% (SD:5.68%). Corresponding mean AD and mean CC were of 17.54 μm (SD:16.45μm) and 98.10% (SD:1.60%) which too approximate the respective intra-observer repeatability values 19.19 μm (SD:17.69 μm) and 98.58% (SD:1.12%).
High correlation between algorithmic and manual delineations indicates suitability of our algorithm for clinically analyzing choroid in greater finer details, especially, in diseased eyes.
The choroid consists of three sub-structures: choriocapillaris, Sattler’s layer and Haller’s layer . Choriocapillaris, the innermost structure adjoining retinal outer boundary defined by Bruch’s membrane, comprise the tiniest choroidal vessels. In contrast, Haller’s layer, comprising the largest vessels is the outermost and adjoins the choroid-sclera interface (CSI). Sandwiched in between lies Sattler’s layer comprising medium-sized vessels. Structural changes in choroid potentially indicate various diseased conditions including choroid neovascularization (CNV), high myopia, chorioretinal inflammatory diseases, tumors, central serous chorioretinopathy (CSC), age-related macular degeneration (AMD) and diabetic retinopathy (DR) [2, 3, 4, 5, 6, 7, 8, 9]. Against this backdrop, understanding the structural changes in choroid and their relevance to various diseases as well as disease management has been of long-standing interest to clinicians. Indeed, ophthalmologists now envision diving deeper and examining at sub-structural levels .
We now proceed to evaluate the accuracy of the proposed algorithm. In particular, on 50 B-scans of 50 eyes from unique subjects, our methodology was tested. The mean age of the subjects was 50.7 years (standard deviation: 18.5 years). Further, 25 subjects were women, and 25 subjects were men. The mean spherical equivalent was 0.1 ± 0.9. Fig 4 depicts close agreement between the algorithmic segmentation and manual markings on nine representative B-scans. Next, we turn to evaluate the proposed methodology quantitatively.
In this work, we proposed an automated methodology to detect the boundary separating Haller’s and Sattler’s layers of the choroid with an aim of quantifying Haller’s layer. Specifically, we described an intuitive heuristic to identify the large blood vessels after binarizing the choroid layer. Delineation results of our algorithm exhibit close agreement with those obtained manually, specifically achieving a Dice coefficient of 89.48% vis-à-vis an intra-observer Dice coefficient of 89.12%. In fact, to the best of our knowledge, ours remains the first attempt at quantitative performance measurement comparison for the problem at hand. An earlier algorithm in similar circumstances was evaluated only via subjective quality comparison with 2D ICGA images . Here, note that retinal sublayers, with boundaries marked by significant intensity gradient, are relatively straightforward to delineate [36, 37, 38]. In contrast, segmentation of choroidal substructures poses a tougher challenge. Unlike in retinal sublayers, choroidal sublayers are not defined by sharp intensity transition and potentially possess structural similarity. Consequently, traditional gradient-based segmentation techniques, employed for quantifying retinal layers, are not appropriate. The present method, employing extended-minima transform alongside watershed segmentation, has been developed to meet such a requirement.