Background In pre-manifest Huntington’s disease there is an increase in symptom severity that is thought to be accompanied by an acceleration of cortical grey matter (CGM) volume loss in the period immediately prior to diagnosis that has not been well characterised. Before examining CGM change, the tools used to calculate CGM volume need evaluation as it is unclear which software most accurately and sensitively measures between-group differences and within-group change in CGM in HD.

Aims Conduct a validation of software in order to determine the most sensitive and accurate tools for quantifying disease-related between-group differences and within-group change in CGM in HD.

Methods 20 controls, 40 premanifest HD (20 preHD-A, >10.8 yrs from predicted onset; 20 preHD-B, <10.8 yrs from predicted onset) and 40 early HD (20 stage 1 (HD1); 20 stage 2 (HD2)) participants from Track-HD were included. 3 T T1-weighted scans were acquired from four scanners. Participants had two 2008 baseline scans and one follow-up scan at 2011. Software used for comparison was FSL FAST version 5.0.9, SPM 8 Unified Segment and New Segment, SPM 12 Segment, FreeSurfer version 5.3.0, ANTs version 2.1.0, and MALPEM version 1.2. Segmented volumes underwent visual quality control (QC). Intra-class correlations were calculated for 2008 back-to-back scans, and between-group differences and within-group change were examined via generalised least squares regression. Total GM, CGM and lobular GM were investigated.

Results Visual QC highlighted that some techniques over- and under-estimate GM, especially in occipital and temporal regions. Most methods showed high reliability. Results for between-group and within-group analysis varied depending on the software used.

Conclusions Whilst reliability was high, between and within-group results and detailed QC suggest that accuracy of the software is variable. This study provides recommendations for the measurement of total, cortical and lobular GM in HD.