Deflation-Inflation Gradient Vector Flow Snakes for Prescreening Glaucoma in Mobile Phone Retinal Images

Glaucoma Detection in Mobile Phone Retinal Images Based on ADI-GVF Segmentation with EM initialization

“…Besenczi et al [5] combined the works of Lalonde et al [11] and Sopharak et al [12] to create a hybrid OD localization process for their work of glaucoma assessment. To localize the OD, our previous method [1] applies the optimal thresholding technique [13] and adaptive histogram equalization. This method finds the regions with the maximum intensity value and merges the nearby regions to get an initial OD region.…”

“…Recently, a variant of GVF, Alternated Deflation-Inflation Gradient Vector Flow snakes (ADI-GVF), was proposed by Ruennak et al [1] to segment the boundaries of OD and OC for prescreening glaucoma. The method iteratively alternates the processes of deflating and inflating the GVF until the region converges.…”

“…Previously, there have been large efforts to analyze the reliability and quality issues of mobile phone reti-nal images for computer-aided diagnosis systems in ophthalmology. Diabetic retinopathy pre-screening [4] and glaucoma assessment [1] [5] are performed by detecting abnormalities in the images acquired by a mobile phone. This work is based on our previously proposed method, Alternated Deflation-Inflation Gradient Vector Flow (ADI-GVF) snakes [1] with a more efficient algorithm for localizing the OD.…”

“…Diabetic retinopathy pre-screening [4] and glaucoma assessment [1] [5] are performed by detecting abnormalities in the images acquired by a mobile phone. This work is based on our previously proposed method, Alternated Deflation-Inflation Gradient Vector Flow (ADI-GVF) snakes [1] with a more efficient algorithm for localizing the OD. We estimate the likelihood of having glaucoma for the patient using the segmented OD and OC areas in the retinal images obtained by our improved method.…”

“…Existing image processing algorithms have a problem with handling retinal images with narrow elds of view and poor-quality taken by a mobile phone. This paper provides a detailed method to enhance the accuracy of our recently proposed scheme, Alternated Deation-Ination Gradient Vector Flow model (ADI-GVF), to improve the segmentation of the optic disk (OD) and the optic cup (OC) for glaucoma detection [1]. We integrated the exclu-sion method (EM), a precise algorithm for localizing the OD, with the ADI-GVF algorithm.…”