D021 Characterising blood neurofilament light dynamics with multi-scanner harmonised voxel-based morphometry in Huntington's disease

<h3>Background</h3> Examining blood neurofilament light (NfL) in conjunction with structural brain changes in Huntington’s disease (HD) offers a multi-dimensional perspective crucial for advancing our understanding of NfL. Combing multiple retrospective cohorts with matched blood NfL and MRI data increases the statistical power to investigate NfL across all HD stages. However, analysing imaging data from multiple studies can introduce biases due to variations in scanners and acquisition protocols, potentially compromising the reliability and interpretability of the study outcomes. Various complex statistical methods have been proposed to address neuroimaging harmonisation problems at the feature level including ComBat, however directly harmonizing images at the voxel-level has been less explored. <h3>Aims</h3> This project aims to explore harmonisation protocols for voxel-based morphometric (VBM) measurements across multiple retrospectively collected HD datasets (HD-YAS, HD-CSF, TRACK-HD, TrackOn-HD, PREDICT-HD). <h3>Methods</h3> Using retrospectively collected T1W images from HD cohorts, we plan to investigate the harmonisation performance of ComBat at the voxel-level (voxel-ComBat), and going beyond voxels, a new approach which operates on three-dimensional superpixels (SP-ComBat). <h3>Results</h3> We will present the methodological pipelines and planned study design for this ongoing work. This will include the plan for evaluation and selection of the neuroimaging harmonisation method to take forward and characterise blood NfL in an unbiased VBM analysis across HD-ISS stages and CAG repeat lengths. <h3>Conclusions</h3> Establishing a harmonized VBM protocol across scanners is imperative to effectively utilize and unify data from multiple HD cohorts to enrich the mechanistic insights of NfL across multiple HD-ISS stages.