GATSBY COMPUTATIONAL NEUROSCIENCE UNIT
UCL Logo
 

 

GATSBY COMPUTATIONAL NEUROSCIENCE UNIT
UCL Logo
 

David Greenberg

 

( MPI Tuebingen, Jason Kerrs Group )

 

Friday 4th November 2011

15.00pm

 

4th Floor Seminar Room,

Alexandra House, 17 Queen Square, London, WC1N 3AR

 

 

 

Oh, behave: problems and solutions in data analysis for two photon imaging in awake and freely moving animals

 

 

Two-photon imaging is uniquely able to detect single action potentials in populations of spatially resolved neurons of the mammalian CNS, but two data analysis problems arise when the restrictions of anesthesia and head fixation are removed. First, images from awake animals are distorted by brain motions faster than the laser-scanning acquisition of a single frame. These distortions can be corrected by a fully automated procedure that estimates the trajectory of brain displacement during each image frame. Second, during fiberscope imaging (Sawinski et al, 2009) of visual cortical neurons the animal controls its own visual input through movements that shift its vantage point and gaze orientation. We therefore developed an all-optical system for head tracking in rodents compatible with two-photon imaging and unrestrained movement. For both these problems we employed optimization over parameters representing physical quantities in a realistic model of image formation, which we found more robust than approaches with an intermediate step of image feature detection. As time allows, I will briefly outline a few insights into population activity produced from work in our lab using these advances in experimental methods and data analysis.