Optical Brain Imaging Computational Dewarping and Neuron Selection

Current technology allows optical recording of brain activity in awake behaving mice through the use of miniature head-mounted microscopes. These images require processing, as they are unevenly illuminated, contain framewide shifts, and do not readily reveal individual neurons. After statistical methods yield a set of images containing putative individual neuron shapes and locations, a human must classify images as containing a single neuron or some kind of noise or clutter. This task is confounded by the fact that imaging optics can create aberrations in the image features. We present two algorithms: one for removing optical aberrations for presentation purposes, and another for automatically rejecting images that contain noise or non-neuron features. Keywords—calcium imaging; GRIN lens; optical aberrations