Glaucoma progression detection using nonlocal Markov random field prior

Abstract

Abstract. Glaucoma is neurodegenerative disease characterized by distinctive changes in the optic nerve head and visual field. Without treatment, glaucoma can lead to permanent blindness. Therefore, monitoring glaucoma progression is important to detect uncontrolled disease and the possible need for therapy advancement. In this context, three-dimensional (3-D) spectral domain optical coherence tomography (SD-OCT) has been commonly used in the diagnosis and management of glaucoma patients. We present a new framework for detection of glaucoma progression using 3-D SD-OCT images. In contrast to previous works that use the retinal nerve fiber layer thickness measurement provided by commercially available instruments, we consider the whole 3-D volume for change detection. To account for the spatial voxel dependency, we propose the use of the Markov random field (MRF) model as a prior for the change detection map. In order to improve the robustness of the proposed approach, a nonlocal strategy was adopted to define the MRF energy function. To accommodate the presence of false-positive detection, we used a fuzzy logic approach to classify a 3-D SD-OCT image into a "non-progressing" or "progressing" glaucoma class. We compared the diagnostic performance of the proposed framework to the existing methods of progression detection.

Structural Change Can Be Detected in Advanced-Glaucoma Eyes

“…This is important because there is no generally accepted definition or “gold standard” for detection of progression in advanced glaucoma that can be used to identify progressing eyes for use in training the classifier. It is important to note that we have previously demonstrated the utility of the BKDS for detection of ONH progression in early- to moderate-glaucoma eyes using VF progression as the “gold standard.” 18 , 19 Finally, BKDS can be tailored to specific progression detection tasks. For example, in our previous publications, BKDS was implemented without any retinal layer segmentation.…”

“…The Bayesian-kernel detection scheme is a Bayesian-based approach that uses change in the 3D ONH volume scans to classify an eye as “nonprogressing” or “progressing.” Details of the BKDS have been described previously. 18 , 19 In brief, raw 3D SD-OCT ONH cube scans were exported to a numerical computing language (MATLAB, MathWorks). The Bayesian-kernel detection scheme was used to estimate glaucoma progression from ONH cube scans.…”

“…We utilized a support vector data description (SVDD) classifier. 19 The SVDD enables us to distinguish between targets (nonprogressing eyes) and outliers (progressing eyes) by defining a closed boundary around the target data. As done previously, 29 we included two features as input for the SVDD based on a glaucoma-related region of interest of the baseline image that generally consists of the RNFL, neuoretinal rim, and prelamina tissue of the ONH identified using SALSA.…”

“…Recently, we proposed a new glaucoma change detection method (Bayesian-kernel detection scheme [BKDS]) that utilizes the whole three-dimensional (3D) SD-OCT ONH and does not require extensive retinal layer segmentation. 18 , 19 In early- to moderate-glaucoma eyes with progressing VF damage, while maintaining high specificity the BKDS method had higher sensitivity for detecting progression than cpRNFL. 18 , 19 This method also has the advantage that it detects 3D change in a region of interest, which can include deeper layers such as the lamina cribrosa.…”

“… 18 , 19 In early- to moderate-glaucoma eyes with progressing VF damage, while maintaining high specificity the BKDS method had higher sensitivity for detecting progression than cpRNFL. 18 , 19 This method also has the advantage that it detects 3D change in a region of interest, which can include deeper layers such as the lamina cribrosa. Reports suggest that in eyes with advanced glaucoma, the lamina cribrosa changes its shape by developing a W-shaped configuration.…”