Assisting doctors on assessing movements in infants using motion tracking

In this work, we consider the possibilities of having an automatic computer-based system for tracking the movements of infants. An existing motion tracking system is used to process recorded video sequences containing both color and spatial information of the infant's body pose and movements. The system uses these sequences of data to estimate the underlying skeleton of the infant and parametrize the movements.

Post-processing of these parameters can yield objective measurements of an infant's movement patterns. This could e.g. be quantification of (a)symmetry and recognition of certain gestures/actions such as kicking, crying, roll over and bringing hands together. Clinicians could benefit from such a system, as it would ease the task of diagnosing infants with motor disorders, and in some cases, this could initiate an intervention earlier than usual.

It should be clear, that we do not seek to substitute the task of the clinicians, but the system should instead be considered as a tool for easy extraction of objective measurements describing the movements and as well as a screening tool for highlighting certain patterns in the movements.