Movement analysis in the early detection of newborns at risk for developing spasticity due to infantile cerebral palsy.

In order to limit the consequences of infantile cerebral palsy (ICP), physiotherapy should start as early as possible. This requires that infants at risk are detected at the earliest age possible. Today, diagnosis is based on visual observation by physicians and as such is influenced by subjective impressions. Objective methods, quantifying the pathological deviation from normal spontaneous motor activity would be preferable as they, for example, allow an inter- and intra-individual comparison of movement. In this paper we have developed a methodology that allows the 3-dimensional acquisition of unconstrained movement in newborn babies, using a motion analysis system. From the recorded movement data we have extracted 53 quantitative parameters that describe the differences between healthy and affected participants. Considered individually, each of these parameters does not permit a conclusive statement to be made as to whether or not the patient is at risk. Cluster analysis based on Euclidian distances therefore has been used to find an optimal combination of eight parameters. The optimal combination has been subsequently applied to organize the participants' movement into preferably homogeneous classes labelled "healthy" or "at risk". Classification was performed utilising quadratic discriminant analysis. The methodology presented allows a reliable discrimination between healthy and affected participants. Overall detection rate reached 73%. This value is expected to rise with increasing patient and norm collective database size.