Evaluation of a new approach for semi-Automatic segmentation of the cerebellum in patients with multiple sclerosis

Cerebellar dysfunction is an important contributor to disability in patients with multiple sclerosis (MS), however, few in vivo studies focused on cerebellar volume loss so far. This relates to technical challenges regarding the segmentation of the cerebellum. In this study, we evaluated the semi-Automatic ECCET software for performing cerebellar volumetry using high-resolution 3D T1-MR scans in patients with MS and healthy volunteers. We performed test-retest as well as inter-observer reliability testing of cerebellar segmentation and compared the ECCET results with a fully automatic cerebellar segmentation using the FreeSurfer software pipeline in 15 MS patients. In a pilot matched-pair analysis with another data set from 15 relapsing-remitting MS patients and 15 ageand sex-matched healthy controls (HC), we assessed the feasibility of the ECCET approach to detect MS-related cerebellar volume differences. For total normalized cerebellar volume as well as grey and white matter volumes, intrarater (intraclass correlation coefficient (ICC) = 0.99, 95 % CI = 0.98-0.99) and interobserver agreement (ICC = 0.98, 95 % CI = 0.74-0.99) were strong. Comparison between ECCET and FreeSurfer results likewise yielded a good intraclass correlation (ICC = 0.86, 95 % CI = 0.58-0.95). Compared to HC, MS patients had significantly reduced normalized total brain, total cerebellar, and grey matter volumes (p B 0.05). ECCET is a suitable tool for cerebellar segmentation showing excellent test-retest and inter-observer reliability. Our matched-pair analysis between MS patients and healthy volunteers suggests that the method is sensitive and reliable in detecting cerebellar atrophy in MS.