We study the sensorimotor adaptation in the cerebellum with the whiskers of the mouse as a model system. The whiskers are a sophisticated sensorimotor apparatus with which the mouse actively profiles its immediate 3D environment by swiftly whisking. To estimate distances, orientation and texture of its surroundings, the mouse uses a ‘maximum contact, minimum penetration’ scheme, with sensorimotor decisions taking place within as little as 10 ms. This scheme imposes the main experimental challenge to whisker-based studies: hair-like structures moving at high speeds need to be tracked accurately to ascribe whisker motion to cerebellar function. In order to capture the motion of the whiskers accurately, high-speed image acquisition at high frame rates (ca. 1000 fps) is required. On account of the paradigms developed to relate neural activity to whisker behavior, high-throughput and real-time whisker tracking are required. We have evaluated all presently available whisker-tracking tools that delivered the desired kinematics of whisker behavior. However, all have failed in terms of computational time and/or acquisition frequency, which rendered them unsuited both for offline use and for real-time applications. To address these limitations, we propose the commercialization of WhiskTrackGP, a high-performance image-acquisition and -processing platform for bleeding-edge neuroscientific experiments on behavior. WhiskTrackGP is a hardware-software platform that can recognize the kinematics of multiple whiskers simultaneously with minimal intervention from the experimenter. It is realized as powerful whisker-tracking software ported on a special hardware platform based on NVidia GP-GPUs and Maxeler Dataflow-Engines. We demonstrate accelerated offline video analysis in the order of 10000x (from an original processing time of 9.4 sec down to 0.9 msec). WhiskTrackGP comprises a “one-stop shop” solution for labs and companies active in the field of behavioral studies.
Call for proposal
See other projects for this call