Cell detection and joint shape tracking using local image filters

This chapter presents an overview of the application of local image filters for the problems of cell detection and tracking in microscopy images, and also extends their use to the joint tracking of motion and shape of cells in time-lapse videos. The use of cell tracking based on a detection-association approach has the advantage of simplicity but is limited by the initial detection. State modelling approaches that assume linear dynamics and Gaussian noise in the tracking estimation can make use of the Kalman filter. However, in real biological applications more complex models may be required, which may not be linear or Gaussian, invalidating the use of the Kalman filter. Particle filter-based tracking is applied when modelling nonlinear dynamics, as they are less restrictive in their assumptions. Cell morphology plays an important role on cell mobility more precisely in the directionality and randomness of the cell movement.