Structural brain differences in healthy LRRK2-G2019s mutation carriers who are at risk of developing Parkinson's disease (PD) were studied using radiomics analysis of DTI and T1-weighted images. 83 subjects were included: 43 healthy-carriers (HC) and 40 healthy-non-carriers (HNC). 18 statistical parameters were extracted for each modality in 14 subcortical brain regions. Various machine-learning classifiers were tested. The best classification results were obtained using RUSBoosted classifier, with average accuracy 73%, sensitivity 68% and specificity 79%. Radiomics analysis revealed brain differences in HC in comparison to HNC. These results together with the preliminary results among converters support the hypothesis of utilization of structural compensatory mechanisms in this "at risk" cohort.