Objective: Caudate atrophy rate measured from serial MRI is proposed as a biomarker of HD progression that may be of use in assessing putative disease-modifying agents. Manual measurement techniques are the most widely applied but are time-consuming. We describe and evaluate an automated technique based on a local registration and boundary shift integral (BSI) approach at the caudate-CSF and caudate-white matter boundaries; caudate boundary shift integral (CBSI).
Methods: Two-year caudate volume change was measured in controls, premanifest HD and early HD using the CBSI and compared with a detailed manual measure in terms of 1) raw caudate volume change, 2) group differentiation, 3) associations with clinical variables and 4) rater requirements. CBSI additivity was assessed by comparing measurements over a single scan pair (baseline-->2 years), with the sum of measurements from two scan pairs (baseline-->1 year-->2 years).
Results: Techniques produced comparable caudate volume change measurements, although CBSI under-reported by 0.04 ml relative to manual. Both techniques distinguished controls, premanifest and early HD with a stepwise increase in rates across groups. Higher rates (CBSI and manual) were associated with increased proximity to estimated disease onset but not clinical change scores. CBSI reduced rater requirements by 2/3 (2 h per subject) relative to manual for this three time-point investigation. CBSI measurements over one scan pair showed good agreement with the sum of measurements from two scan pairs.
Conclusions: CBSI results were comparable to a manual measure but with reduced rater requirements. CBSI may be of use in large-scale studies of HD.