Objective QC for diffusion MRI data: artefact detection using normative modelling

Diffusion MRI is a neuroimaging modality used to evaluate brain structure at a microscopic level and can be exploited to map white matter fiber bundles in the brain. One common issue is the presence of artefacts such as acquisition artefacts (e.g. ringing), physiological artefacts (cardiac pulsation, breathing, motion), distortions (EPI, eddy currents) or post-processing (e.g. blurring, registration errors). These may lead to problems in downstream processing and can bias subsequent analyses. In this work we use normative modeling to create a semi-automated pipeline for detecting diffusion imaging artefacts and errors by modeling 24 image derived phenotypes using the UK Biobank dataset. We used several parameter types with different modelling complexities (DTI model: FA and MD; NODDI model: ICVF and ISOVF). Likewise, we selected several tracts of various sizes and modelling difficulties (corpus callosum, corticospinal tract, uncinate fasciculus and fornix). Our method is compared to two traditional quality control modalities: a visual quality control protocol performed on 500 subjects and quantitative quality control using other automated pipelines. The normative modelling framework proves to be comprehensive and efficient in detecting diffusion imaging artefacts arising from various sources as well as outliers resulting from inaccurate post-processing (i.e., erroneous registrations). This is an important contribution by virtue of this methods’ ability to identify the two problem sources (i) image artefacts and (ii) post-processing errors, which subsequently allows for a better understanding of our data and informs on inclusion/exclusion criteria of participants.