Vision-based, Real-time Retinal Image Quality Assessment

Real-time medical image quality is a critical requirement in a number of healthcare environments, including ophthalmology where studies suffer loss of data due to unusable (ungradeable) retinal images. Several published reports indicate that from 10% to 15% of images are rejected from studies due to image quality. With the transition of retinal photography to lesser trained individuals in clinics, image quality win suffer unless there is a means to assess the quality of an image in real-time and give the photographer recommendations for correcting technical errors in the acquisition of the photograph. The purpose of this research was to develop and test a methodology for evaluating a digital image from a fundus camera in real-time and giving the operator feedback as to the quality of the image. By providing real-time feedback to the photographer, corrective actions can be taken and loss of data or inconvenience to the patient eliminated. The methodology was tested against image quality as perceived by the ophthalmologist. We successfully applied our methodology on over 2,000 images from four different cameras acquired through dilated and undilated imaging conditions. We showed that the technique was equally effective on uncompressed and compressed (JPEG) images. We achieved a 100 percent sensitivity and 96 percent specificity in identifying "rejected" images.