Cognitive impairment in Parkinson's disease (PD) results in significant morbidity and mortality being early diagnosis essential. Identification of patients who are at higher risk of developing cognitive impairment based only on clinical data is not sufficient. To this end, magnetic resonance imaging (MRI) with automatic segmentation, such as FreeSurfer, could be a useful tool with high accuracy because it has histological validation. ObjectivesThe objective of this study was to evaluate clinical, neuropsychological and FreeSurfer variables that may be related to worse cognitive outcomes over 18 months in PD patients compared with controls. MethodsPD patients were recruited according to established inclusion and exclusion criteria as well individuals without any neurological or psychiatric diagnosis and were submitted to the same protocol: neurological, neuropsychological and neuroimaging evaluations. After 18 months, the study subjects were reassessed by neurological and neuropsychological evaluations. ResultsOf 171 individuals selected for first evaluation, 96 concluded the study during 18-month follow-up. The PD group presented worse performance in the neuropsychological assessment during both the initial and final evaluations. The results obtained by FreeSurfer revealed a significant reduction (unilateral or bilateral) in the volume of thalamus, caudate nucleus, putamen, hippocampus, amygdala, accumbens, corpus callosum and cerebral gray matter in the PD group. A worse cognitive outcome was more prevalent in the PD group. ConclusionsWorse cognitive performance documented by neuropsychological assessment in the PD group was correlated with reduced volume of several structures by FreeSurfer analysis and may be a biomarker of cognitive decline.