Background Validating neuroimaging biomarkers that linearly track clinical progression over the lifetime of a HD patient is essential for clinical trials to test drug efficacy and ultimately develop a cure. Different image processing techniques can introduce variability and standardisation is required for cross-study comparison.

Aim This study aims to investigate image analysis across three large observational datasets, TRACK-HD, TrackOn-HD and HD-YAS, in order to standardise a processing pipeline for grey and white matter segmentation. In this way, the progression of HD pathology across the entire lifespan can be investigated.

Methods Neuroimaging data from 497 participants from the TRACK-HD, TrackOn-HD and HD-YAS datasets were combined. Original processing with SPM5, SPM8 and standardised processing using the CAT12 tool in SPM12 were compared. Grey and white matter volumes were plotted to observe the development of atrophy in HD. Statistical parametric maps correlating cognitive and motor outcomes with regional atrophy were also assessed.

Results Comparison of SPM5 and SPM8 with SPM12 standardised brain segmentations showed significant differences. The CAT12 pre-processing tool was validated as an accurate method for standardised brain segmentation across both data sets. In the premanifest cohort, caudate volume was the earliest biomarker detected, followed by grey and white matter volumes. Examination of the cognitive tests highlighted a need for more sensitive measures for early premanifest individuals.

Conclusions We demonstrate a systematic difference between different SPM software versions, which was removed when the CAT12 tool was used. This emphasises the need for standardisation of image analysis processing when combining multiple datasets.