Automated MRI-derived measurements of human brain volumes from anatomical scans provide novel insights into normal and abnormal neuroanatomy, but few studies have probed the effects of sequence-dependent parameters on these measurements.¹ The multi-echo MPRAGE (MEMPRAGE) sequence reduces signal distortion by using a higher bandwidth and averaging multiple echoes to recover SNR while using variable T₂* decays to enhance contrast, and hence, cortical segmentation.² For maximal efficiency, a rapid 2-minute MEMPRAGE protocol has been implemented for large multi-center studies.³ This yields quantitatively repeatable morphometric information on sub-cortical/white-matter structures in comparison to scans employing a higher-resolution MEMPRAGE protocol with lower image acceleration.⁴ However, small but significant differences in cortical thickness and gray matter volume were observed between the two protocols.⁵ Here, we investigate the impact on morphometric results for the cortex by systematically varying all the parameters modified between the rapid 2-minute and the conventional 6-minute MEMPRAGE scans.All measurements were performed using a 3.0 T MRI scanner (Siemens Tim Trio). 10 subjects (mean: 28.9 years, 5 male) were scanned using the 12-channel coil. Each session included a standard MEMPRAGE (6:03 min, TR/TI = 2530/1200 ms, matrix 256×256×176, resolution = 1 mm iso (no partial fourier), parallel imaging (PI) acceleration = 2 (p2)); and a rapid MEMPRAGE (2:12 min, TR/TI = 2200/1100 ms, matrix 230×230×144, resolution = 1.2 mm iso (6/8 partial fourier), PI = 4 (p4)). TR/TI; matrix/resolution/partial fourier; and PI acceleration were systematically varied between standard and rapid scans, resulting in a total of eight MEMPRAGE scans for each subject (See Fig. 1). Images were analyzed using FreeSurfer,⁶ after the eight scans from each subject were aligned using the FreeSurfer robust registration tool.⁷ An automated parcellation of the cortex, subcortical and white matter structures was performed. The 33 cortical regions of the Desikan-Killiany atlas were combined into five principal cortical lobes.⁸ Correlation and Bland-Altman difference plots were made for the thickness and volume of each principal cortical lobe determined from each scan.Three difference-analyses were performed: all 1.0mm scans for all subjects vs. all corresponding 1.2mm scans for all subjects; all p2 scans for all subjects vs. all p4 scans; and all TR2530 vs. all TR2200 scans. Example Bland-Altman plots are shown in Fig. 2. The difference analyses are summarized in Fig. 3. The resolution change resulted in significant differences in cortical thickness and volume, similar to those observed previously.⁵ TR/TI variation had no effect, while PI acceleration had no significant effect. Bland-Altman slopes were ~ -0.2 – 0.2, indicating no bias with measurement value. Correlation coefficients are summarized in Fig. 4.Small but significant differences in cortical thickness and gray matter volume were previously seen between 6-min and 2-min MEPRAGE scans.⁵ This could result from either slight blurring of the GM/WM boundary in the rapid scan from the use of partial fourier acquisition, or from the resampling of the 1.2mm data to 1.0mm by the standard Freesurfer processing stream. Surprisingly, the reduced TR/TI from the Freesurfer recommended values had no effect on cortical data, while PI acceleration had no significant effect. It has recently been shown that motion by subjects can result in reduced estimates of cortical thickness.⁹ While longer scans are more likely to be impacted by subject motion, it is the shorter scans here which show a reduced cortical thickness, indicating motion is likely not responsible for the effects seen in this study. Rather, blurring of the GM/WM boundary must happen at the expense of pure GM, resulting in cortical thinning in the case of the faster scans. For a rapid MEMPRAGE with no impact on cortical thickness compared to conventional scans, a 1.0 mm spatial resolution scan employing the other techniques to speed the acquisition is optimal. Such a scan would take 3 mins 13 sec.