RATIONALE

There exists a wide range of reported DTI metrics in the literature²², including reports that are potentially artefactual which have been previously highlighted²³. This will be influenced by the subject cohort e.g. health vs disease. Other potential sources of variation include differences in sites/operators/scanners, acquisition and post-processing methods. This variation needs to be better understood to support clinical validation of the technique.

Early work has evaluated reproducibility in a ten-site study in isotropic phantoms²⁴,

and in a two-site study in healthy volunteers²⁵. In this study, we propose a broad multi-centre collaborative effort to evaluate inter-site variation due to differences in (i) site / operator / scanner, (ii) acquisition and (iii) post-processing methods. The results will guide interpretation of the cardiac diffusion MRI literature, and help to refine strategies for standardisation and harmonisation of protocols. We propose in the first instance to study healthy volunteers to establish a baseline of variation due to technical (non-pathology related) factors.

THEORETICAL FRAMEWORK

In order to evaluate variation arising from different sources, we require a sampling of sites with a relevant range of characteristics in the NHS and non-NHS sites in the UK, and through collaboration with international sites. Key characteristics identified include scanner field strength, vendor, gradient hardware, pulse sequence, geographical location and cDTI experience. In addition, the variation arising from subjects is considered and can be isolated by comparing travelling and non-travelling volunteer cohorts. Statistical techniques such as Bland-Altman²⁵, t-tests, multivariate ANOVA, correlation and bootstrapping²⁴ will be used to evaluate differences in cDTI parameters such as MD and FA arising from different sources.

RESEARCH AIMS

1. To evaluate variation in cardiac diffusion tensor imaging (cDTI) data due to differences in (i) site/operator/scanner, (ii) subject, (iii) acquisition and (iv) post-processing methods.
2. To identify the greatest source(s) of variation, and to inform strategies for optimisation, standardisation and harmonisation of cDTI.
3. To inform power calculations and data interpretation in future larger studies, in particular, where different sites/methods are used.