obtaining a series of ultra-widefield angiography images of a retina of a patient;

providing each of the series of ultra-widefield angiography images to a neural network trained on a set of labeled images to generate a quality parameter for each of the series of ultra-widefield angiography images representing a quality of the image, each of the set of labeled images being assigned to one of a plurality of classes representing image quality, the plurality of classes comprising a first class, representing a highest image quality in which each image has a field of view above a first threshold percentage, a second class, representing an acceptable level of image quality in which each image has a field of view above a second threshold percentage, a third class, representing a poor level of image quality in which each image has a field of view above a third threshold percentage, and a fourth class, representing images that are unacceptable for use in assessing the state of the retina of the patient; and

providing an instruction to a user to obtain a new series of ultra-widefield angiography images of the retina of the patient if no image of the series of ultra- widefield angiography images has a quality parameter that meets a threshold value.