Use of Varying Constraints in Optimal 3-D Graph Search for Segmentation of Macular Optical Coherence Tomography Images

Abstract

Abstract. An optimal 3-D graph search approach designed for simultaneous multiple surface detection is extended to allow for varying smoothness and surface interaction constraints instead of the traditionally used constant constraints. We apply the method to the intraretinal layer segmentation of 24 3-D optical coherence tomography (OCT) images, learning the constraints from examples in a leave-one-subject-out fashion. Introducing the varying constraints decreased the mean unsigned border positioning errors (mean error of 7.3 ± 3.7 μm using varying constraints compared to 8.3 ± 4.9 μm using constant constraints and 8.2 ± 3.5 μm for the inter-observer variability).

Automated 3D segmentation of intraretinal layers from optic nerve head optical coherence tomography images

“…These constraints can be expressed as constant values for the two directions for each surface 6,7 or as a function that varies with respect to the (x, y) location. 8 For instance, the smoothness constraints in the x-direction when moving from location (x 1 , y 1 ) to (x 2 , y 2 ) can be defined as,…”

Automated retinal layers segmentation in SD-OCT images using dual-gradient and spatial correlation smoothness constraint

“…Graph search based segmentation algorithms [36][37][38][39][40][41][42][43][44][45] have been pursued. A 2-D graph segmentation method was introduced by Chiu et al [37,38] and patients with age-related macular degeneration (AMD), showing accurate results.…”

Simultaneous detection of multiple elastic surfaces with application to tumor segmentation in CT images

“…As originally presented by Li et al [8,9] and in the follow up extensions by Haeker et al [10,11], the cost functions presented are all image-derived, and are independent of the surface configurations. However, in some applications, it may also be desirable to penalize for certain surface configurations or shapes, especially if certain prior knowledge about these properties is available.…”

“…The core component of our method is a graph-theoretic algorithm that simultaneously detects multiple coupled surfaces under geometrical constraints. The algorithm is a non-trivial extension of the framework proposed by Li et al [8,9], Haeker et al [10,11], and Wu et al [12]. In the previous framework, surface segmentation is formulated as a cost minimization problem with geometric constraints.…”