Methods, apparatuses, and computer readable media for detecting abnormalities in a characteristic of an eye using eye-imaging methods are presented. A plurality of images of the eye are received over time. Each image includes a plurality of pixels, which can be partitioned into blocks of pixels with varying sizes, called pixel partitions. A value is determined for each pixel partition, e.g., an average of the pixel values. A pixel partition set may be identified, which includes a pixel partition from each image, corresponding to a common region of a patients eye. A regression model is computed for each pixel partition set using the values determined for each pixel partition. The regression model computes a rate of change of the retinal nerve fiber thickness at individual pixel partitions over time. An abnormality may be identified by comparing the rates of change of the model and the expected age-related rate of change.
