Harmonization of the iBEAt Dixon MRI data

The iBEAt study on MRI in diabetic kidney disease includes 1328 pre- and post-contrast 3D Dixon MRI scans, each scan producing in-phase, opposed-phase, fat and water volumes, totalling 5,312 3D image series. Scans are acquired in 524 patients (twice in a subset of 108) and 30 controls (up to 4 repeat scans per control) in 6 imaging sites across Europe and centralised in DICOM format in miblab for processing.

Since data are acquired on different scanners and sites, there is significant variability in the data format. The standard approach within miblab is to harmonize those data first, and produce a clean dataset in DICOM format to serve as input for downstream image processing pipelines. 

This repository contains the pipeline for Dixon data harmonization and cleaning in iBEAt. The input data are the source data as they are exported from the scanners, and stored in miblab's central XNAT repository. Input- and output data are at this stage embargoed and only available to miblab members.