Reconstruction of trajectories with data loss by Kalman filtering: an application to Morris water maze tests

Abstract

The Kalman filter was applied in the reconstruction of trajectories that present data loss. The trajectories were obtained by tracking a rat’s swim from video sequences in the Morris water maze tests. These video sequences are being used in neuroscience studies in spatial memory tests. In this work, the Kalman filter showed to be a good alternative to estimate position and velocity when some measurement data have been lost. The reconstruction was successfully accomplished for long and short data losses. The algorithm can be applied in other behavioral tests for different kind of mazes. For example, in the guinea pig maze and elevated Y maze, where is not unusual to have obstruction of the observation path.