Abstract
This paper proposes a teleophthalmology support system in which we use algorithms of object detection and semantic segmentation, such as faster region-based CNN (FR-CNN) and SegNet, based on several CNN architectures such as: Vgg16, MobileNet, AlexNet, etc. These are used to segment and analyze the principal anatomical elements, such as optic disc (OD), region of interest (ROI) composed by the macular region, real retinal region, and vessels. Unlike the conventional retinal image quality assessment system, the proposed system provides some possible reasons about the low-quality image to support the operator of an ophthalmoscope and patient to acquire and transmit a better-quality image to central eye hospital for its diagnosis. The proposed system consists of four steps: OD detection, OD quality analysis, obstruction detection of the region of interest (ROI), and vessel segmentation. For the OD detection, artefacts and vessel segmentation, the FR-CNN and SegNet are used, while for the OD quality analysis, we use transfer learning. The proposed system provides accuracies of 0.93 for the OD detection, 0.86 for OD image quality, 1.0 for artefact detection, and 0.98 for vessel segmentation. As the global performance metric, the kappa-based agreement score between ophthalmologist and the proposed system is calculated, which is higher than the score between ophthalmologist and general practitioner.
Highlights
Eye diseases, such as diabetic retinopathy (DR), aged-related macular degeneration (AMD), glaucoma, retinopathy of prematurity (ROP), and so on, can be diagnosed by retinal image screening.the patients with some diseases detected by the screening can receive an adequate treatment to avoid serious vision loss
We propose a support system for teleophthalmology framework based on non-mydriatic fundus cameras, basically on the scanning laser ophthalmoscopes (SLOs)
The optic disc (OD) shows as a high intensity spot, Taking into account the above-mentioned analysis about the retinal image quality, we propose it can be distinguish from other high intensity spots caused by some white lesions, because a teleophthalmology support system, which is composed of mainly four steps: OD region the OD shows some particular features due to the presence of vessels inside of it. (2) The blurriness of segmentation, blurriness analysis of the OD region, region of interest (ROI) segmentation and computation of the grade the OD reflects directly the blurriness of the whole retinal image, if the region of the OD is blurred, of obstruction by the artefacts, as well as vessels segmentation in the ROI
Summary
Eye diseases, such as diabetic retinopathy (DR), aged-related macular degeneration (AMD), glaucoma, retinopathy of prematurity (ROP), and so on, can be diagnosed by retinal image screening. The retinal images with some specific eye disease may be determined to have inadequate quality by the proposed system because important anatomical structures, such as OD, are not present in the image Considering this situation, we decide that the maximum repeat number of captures is set to five, following an advice of ophthalmologists. If an acquired image cannot pass some of the firsts three steps after the maximum number of repetitions (five times), the system transmits the best-quality image among all five acquired images, together with a generated report, to the central eye hospital In this manner, the proposed support system helps the operator and the patient to acquire a better-quality retinal image, and provides several useful information to the ophthalmologist in the central eye hospital.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
