Fig. 11
Experiment 2, Section 4.1.b and 4.2.b, investigated the effects of dMRI harmonization on ROI analysis and multi-shell dMRI measures, namely RTOP, RTPP, MSD, and RTAP, in eight randomly selected scanners and white matter ROIs. For this analysis, an unseen set of thirty-five (35) subjects who were not involved in creating the template, but were still matched in terms of age, sex, and IQ to the reference scanner (scanner 29), were selected. Using these subjects from each scanner, two unpaired t-tests were conducted to compare each multi-shell dMRI measure between (i) the original and reference scanners (orig. vs. ref.) and (ii) the harmonized and reference scanners (harmo. vs. ref.). To demonstrate the effectiveness of the harmonization process, we present representative plots for each dMRI measure. It is important to note that any significant differences observed prior to harmonization (p < 0.01) were statistically eliminated after the harmonization procedure (p > 0.1) in all measures derived from the 73 white matter tracts and 32 scanners. The resulting p-values, which confirm the successful harmonization, are displayed on the plots for a few randomly selected scanners and white matter tracts.
