Quantification of the spatio-temporal characteristics of walking trajectories of patients suffering from cerebellar disease

Cerebellar dysfunction can lead to disturbances of gait by causing disequilibrium, and by altering leg and trunk kinematics and interlimb coordination. Such disturbances are characterized by an instable walking path and a high variability of gait. Quantifying the type of impairments based on classical gait features show a significant change only in balance-related parameters like reduced step length, increased step width or reduced peak joint angles. No significant changes can be found for parameters related to impairment of inter-joint coordination or kinetics. For a quantitative description of kinetic impairments it seems desirable to take the full spatio-temporal characteristics of the movement into account.

To model spatio-temporal characteristics we applied a new computational method, Hierarchical Spatio-Temporal Morphable Models (HSTMMs). In order to get a precise description of the spatio-temporal patterns and its variability, a spatio-temporal correspondence between movement trajectories is established. The spatio-temporal correspondence is described by vectors of temporal and spatial displacements. By the linear combination of these spatial and temporal correspondence fields, new trajectories can be modeled.

We tested 14 patients suffering from degenerative cerebellar disease analyzing 8–12 gait cycles per patient.

Since we wanted to identify joint coordination impairments, we examined 16 feature sets consisting of different combinations of joint angles of lower limbs. In a first step we analyzed the intra-subject variability of joint angle coordination using spatio-temporal correspondence between walk cycles. The spatio-temporal variability of the inter-joint coordination shows a significant correlation (Spearman: rs=0.75, p<0.01) with the kinetic subscore of the ataxia scale ICARS. In a second step we identified systematic changes in inter-joint coordination patterns by modeling the trajectories through linear combination of prototypes. We found significant correlations for different feature sets to the posture and kinetic subscore. Based on this analysis we can identify subsets of joints which describe systematic changes of inter-joint coordination pattern correlated with balance and kinetic deficits.

In summary, exploiting HSTMMs we are able to quantify coordination and kinetic deficits in cerebellar ataxic gait.

Supported by the Deutsche Volkswagenstiftung and the Hermann and Lilly Schilling Foundation