Enhanced detection of cell paths in spatiotemporal plots for noninvasive microscopy of the human retina

Abstract

Spatiotemporal (ST) plots have been used for studying cell motion in the microcirculation. However, ST plots are typically applied to invasive imaging methods. Noninvasive video microscopy of the human retinal capillaries can be performed using an Adaptive Optics Scanning Laser Ophthalmoscope, but direct implementation of ST plots is difficult due to low contrast and high noise. We introduce motion contrast enhancement to enhance detection of cell paths, and enable ST plot analysis. Using features generated by the motion contrast enhancement, a method for automatic extraction of cell paths was developed. Our results show that motion contrast is an important precursor step to ST plot analysis for videos with low signal to noise ratios.