Magnetic resonance support vector machine discriminates between Parkinson disease and progressive supranuclear palsy

ABSTRACT Background The aim of the current study was to distinguish patients with Parkinson disease (PD) from those with progressive supranuclear palsy (PSP) at the individual level using pattern recognition of magnetic resonance imaging data. Methods We combined diffusion tensor imaging and voxel‐based morphometry in a support vector machine algorithm to evaluate 21 patients with PSP and 57 patients with PD. Results The automated algorithm correctly distinguished patients who had PD from those who had PSP with 100% accuracy. This accuracy value was obtained when white matter atrophy was considered. Diffusion parameters combined with gray matter atrophy exhibited 90% sensitivity and 96% specificity. Conclusions Our findings demonstrate that automated pattern recognition can help distinguish patients with PSP from those with PD on an individual basis. © 2013 International Parkinson and Movement Disorder Society