Volumetric abnormalities of subcortical structures, and in particular the striatum, have been robustly observed in childhood Attention Deficit Hyperactivity Disorder (ADHD).¹ Whole-brain analyses in adults have, however, been less consistent.^1-6 Although it has been suggested that this may reflect a reduced severity of these abnormalities in adulthood, this may also reflect reduced statistical power in typically smaller adult samples. Indeed, ROI analyses have revealed striatal abnormalities in adults not detected using whole brain methods,^3,4 though such approaches preclude assumption-free whole brain mapping of abnormalities. In lieu of large scale databases dedicated to morphometric assessment of adult ADHD, enhancing accuracy of whole brain morphometric imaging techniques is therefore essential to address these issues. Magnetisation transfer (MT) saturation maps have enhanced subcortical contrast compared to T1 images typically used in morphometric analyses, yielding more accurate automated segmentation of subcortical structures.⁶ To investigate whether these advancements translated into enhanced sensitivity to morphometric abnormalities in ADHD, we compare results from typical VBM analyses using MT saturation maps and T1 weighted volumes in the same group of subjects.

Thirty patients with ADHD, and 30 age, IQ, gender and handedness matched controls were scanned using a 1.5T Siemens Avanto scanner.The scanner’s body coil is used for RF transmission, and a 32-channel head coil for NMR signal reception. MT maps were calculated from a multi-parameter protocol adapted from a 3D multi-echo fast low angle shot (FLASH) sequence.⁶ Three co-localised 3D multi-echo flash sequences were acquired in the sagittal plane: Proton density (TR: 24, TE: 2.51 to 21.9ms (eight equidistant bipolar echoes collected), flip angle (α): 6°, 1.25mm3 resolution (FOV: 240x217.5mm², 144 partitions), T1 (TR: 19, TE: 2.51 to 10.82ms (four equidistant bipolar echoes collected), α: 20°, 1.25mm³ resolution) and MT weighted volumes (TR: 30, TE: 2.51 to 10.82ms, α: 12°, MT contrast on, 1.25x1x1mm³ resolution). T1 weighted images were acquired in the axial plane with an isotropic voxel resolution of 1mm3 (FOV: 256x240mm², 192 partitions), with a TR of 27.3 ms, TE of 3.57ms and a flip angle of 7°.

Both datasets were processed identically: T1 volumes and MT saturation maps were segmented using the VBM8 toolbox in SPM8, before smoothing with a 5mm³ Gaussian kernel. Group comparisons were performed according to the General Linear Model (GLM), controlling for age, total intracranial volumes (derived separately for each modality), and Beck’s Depression Inventory and State and Trait Anxiety Index scores which significantly differed between the groups. Analyses were thresholded at p < 0.001 uncorrected, with surviving clusters thresholded at p < 0.05 corrected for False Discovery Rate (FDR).

MT saturation maps revealed clear volumetric reductions in the right inferior parietal lobule and ventral striatum in ADHD. Contrastingly, whilst T1 VBM was able to detect the same parietal differences, differences in the ventral striatum were not detected even at more relaxed statistical thresholds. T1 VBM results did however indicate thalamic volumetric reductions in the ADHD group that were not detectable using MT VBM.

Whilst T1 and MT saturation maps reveal similar parietal volumetric reductions, MT saturation maps are sensitive to striatal abnormalities that are not detectable in T1 weighted volumes in the same subjects. These results are in line with findings from childhood ADHD and therefore suggest an improved sensitivity of MT saturation maps to subcortical changes. The findings of reduced thalamic volume using T1-weighted images without a corresponding result in MT saturation maps is more difficult to interpret, as previous assessments of MT saturation maps have reported reduced GM probability in the thalamic subregions we observe, suggesting that this likely reflects partial volume effects with the internal medullary lamina.⁶ Further work needed to determine whether these reflect genuine grey or white matter differences between the two groups. These findings urge interpretative caution in the absence of evidence using traditional VBM analyses, and make a case for the inclusion of multi-parameter maps in research protocols, particularly in disorders with known or hypothesised subcortical components.