Optimal Multiple Surface Segmentation With Shape and Context Priors

Abstract

Segmentation of multiple surfaces in medical images is a challenging problem, further complicated by the frequent presence of weak boundary evidence, large object deformations, and mutual influence between adjacent objects. This paper reports a novel approach to multi-object segmentation that incorporates both shape and context prior knowledge in a 3-D graph-theoretic framework to help overcome the stated challenges. We employ an arc-based graph representation to incorporate a wide spectrum of prior information through pair-wise energy terms. In particular, a shape-prior term is used to penalize local shape changes and a context-prior term is used to penalize local surface-distance changes from a model of the expected shape and surface distances, respectively. The globally optimal solution for multiple surfaces is obtained by computing a maximum flow in a low-order polynomial time. The proposed method was validated on intraretinal layer segmentation of optical coherence tomography images and demonstrated statistically significant improvement of segmentation accuracy compared to our earlier graph-search method that was not utilizing shape and context priors. The mean unsigned surface positioning errors obtained by the conventional graph-search approach (6.30 ± 1.58 μm) was improved to 5.14 ± 0.99 μm when employing our new method with shape and context priors.

Automated Bruch’s Membrane Opening Segmentation in Cases of Optic Disc Swelling in Combined 2D and 3D SD-OCT Images Using Shape-Prior and Texture Information

“…In Wang et al's work [2], the center for the unwrapped enface image was computed by averaging the six manually-placed BMO points, and the unwrapped cost image was then locally modified by significantly decreasing the cost values at the regions located at these manual landmarks. With the knowledge that the BMO shape is usually elliptical, a graph-search algorithm with a soft constraint to limit the target surface deviating from the shape-prior [7] was used to segment the BMO unwrapped contour. In this work, the unwrapped center was defined as the middle point of the estimated two BMOs from Section 2.2, and the unwrapped en-face cost image was computed using the same method as Wang et al's work [2] without the local modifications.…”

An Automatical Segmentation Method for Brain MR Images

“…The active contour models are several desirable than classical image segmentation methods, such as edge detection, thresholding, and region grow. The active contour models can be easily formulated into an energy minimization framework, which enable the models allow incorporation of various prior knowledge, such as shape and intensity distribution, for robust image segmentation [6,7]. Furthermore, the active contour models can provide the segmentation results as smooth and closed contours, which can be readily used for further applications, such as shape analysis and recognition.…”

Surface Detection Using Round Cut

“…The cost function, originally defined only in terms of onsurface costs, has also been extended to incorporate regional information [15], [16], [7]. In addition, the roughness has been allowed to vary across the surface [17], and soft shape constraints have been introduced [18].…”