0237

Robustness of PSIR segmentation and R1 mapping at 7T: a travelling head study

Olivier E. Mougin¹, William T Clarke², Ian Driver³, Catarina Rua⁴, Andrew T Morgan⁵, Susan Francis¹, Keith Muir⁵, Adrian Carpenter⁴, Chris Rodgers⁴, Richard Wise³, David Porter⁵, Stuart Clare², and Richard E. Bowtell¹

¹Sir Peter Mansfield Imaging Centre, School of Physics and Astronomy, University of Nottingham, Nottingham, United Kingdom, ²Nuffield Department of Clinical Neurosciences, University of Oxford, Wellcome Centre for Integrative Neuroimaging (FMRIB), Oxford, United Kingdom, ³School of Psychology, Cardiff University, Cardiff University Brain Research Imaging Centre, Cardiff, United Kingdom, ⁴Department of Clinical Neurosciences, University of Cambridge, Wolfson Brain Imaging Centre, Cambridge, United Kingdom, ⁵University of Glasgow, Imaging Centre of Excellence, Glasgow, United Kingdom

Synopsis

Ultra-high magnetic field (7T) MRI scanners can provide high spatial resolution images and excellent contrast for classifying brain tissue, but robustness of tissue segmentation and R₁ quantification across sites is key for the implementation of multi-site studies. Here, we present a subset of the main UK7T travelling-head study, focusing on harmonized T₁-weighted images (0.7mm³ isotropic resolution) acquired on six subjects across three 7T sites, with five repeats at one site. The aim of this study is to assess the harmonisation and robustness of the MP₂RAGE sequence and PSIR reconstruction across sites, by focusing on segmentation reproducibility and T₁ estimation.

Introduction

Ultra-high magnetic field (7T) MRI scanners can provide high spatial resolution images and excellent contrast for classifying brain tissue, but robustness of tissue segmentation and R₁ quantification across sites is key for the implementation of multi-site studies. Here, we present a subset of the main UK7T travelling-head study, focusing on harmonized T_1-weighted images (0.7mm³ isotropic resolution) acquired on six subjects across three 7T sites, with five repeats at one site. The aim of this study is to assess the harmonisation and robustness of the MP₂RAGE sequence and PSIR reconstruction across sites, by focusing on segmentation reproducibility and T₁ estimation.

Methods

Acquisition: The data used in this work was collected as part of the UK7T Network’s “travelling head study”. Six subjects (4 male, 74.0±9.1 kg, 33.8±6.2 years) underwent structural and functional neuroimaging scans at three of the five 7T-capable UK sites. Each subject was scanned at each site (N_away=3), and a further four times at their designated home site (N_home=4, N_total=7 scans per subject included here). A standardised 3D-MP₂RAGE acquisition was implemented on all sites using the following parameters: 0.7x0.7x0.7mm³ voxel size, FOV=224x224x157mm³, acceleration factor (A>>P) of 3 (GRAPPA/SENSE for Siemens/Philips systems), TR of 3500ms, echo spacing of 6.3ms[BW=300Hz], flip angles of 5/2˚ and inversion times (TI) of 725/2150ms for the first/second images. The acquisition time was 6min14s/7min51s (SENSE/GRAPPA). The same adiabatic inversion pulse⁽¹⁾ was successfully implemented on all sites.

Processing: Offline PSIR reconstruction⁽²⁾ was carried out on all MP₂RAGE data after uniform sensitivity Roemer coil combination⁽³⁾ (offline on Siemens scanners).The MP₂RAGE and PSIR reconstruction included a simple method to denoise non-tissue pixels⁽⁴⁾. Additionally, the MP₂RAGE data was segmented using Freesurfer⁽⁵⁾ after inhomogeneity correction using spm12⁽⁶⁾. After running the cross-sectional segmentation on individual visits independently, an unbiased within-subject template space was created from the three sites, before tissue classification was done via Freesurfer using the longitudinal stream⁽⁷⁾. A look-up table (LUT) was numerically calculated using Bloch-simulations incorporating the exact sequence parameters to allow T₁ estimation from the MP₂RAGE signals, considering the individual B₁ map and manual power calibration for each visit. The cortical thickness and cortical volume was compared using a 3-way analysis of variance (ANOVA), looking at the effect of sites, subjects and repeats on the individual subject segmentations. Additionally, Coefficients of Variation (CoV) and Intraclass Correlation Coefficients (ICC) were computed for each region outputted from the Desikan–Killiany–Tourville Atlas, and subsequently mapped onto an average surface for visualisation. Dice Similarity Coefficients (DSC) were also computed on the cortical ribbon segmentation, comparing the segmentation from the seven scans to the average subject space.

Results

Images with similar quality and high contrast to noise ratio were produced at all sites and successful segmentation using Freesurfer was obtained on all data. Reproduceable T₁ maps could also be produced (Figure 1) using the LUT (CoV_home=1.28, CoV_away=6.02). The cortical ribbon segmentation from Freesurfer was reproducible across sites, as no significant difference in brain cortical volume was present either between sites (p=0.62) or repeats (p=0.74), testing the null hypothesis on the sites, patients and repeats using a 3-way analysis of variance (ANOVA). This was also true when testing either the left cortical hemisphere volume (p=0.81 and p=0.51 respectively) or the right cortical hemisphere volume (p=0.19 and p=0.30 respectively). CoV and ICC were also mapped on an average subject for volume (Figure 2) and cortical thickness (Figure 3), showing good reproducibility when comparing all, away or home scans only. Apart from slightly higher variability in the cortical thickness and volume of the pre-central gyrus cortical area in particular, the reproducibility was very good, with little difference between home and away data, also visible with the DICE similarity coefficient (Figure 4).

Discussion

Segmentation of MP₂RAGE data using Freesurfer produced robust quantitative cortical thickness measurements across sites. Small cortical thickness variation is visible on Figure 3, which might be improved with the use of a better MP₂RAGE brain extraction step. Measured T₁ values were comparable across sites, but small residual variations due to B₁ are still present. A LUT including B₁ correction was implemented⁽⁹⁾ to reduce variability in the measured T₁, using B₁ maps together with cross-site power normalization. Although DSC is comparable within sites (CoV_home=0.54) and across sites (CoV_away=1.51), part of the segmentation variability might still be explained by intrinsic registration errors present within the longitudinal Freesurfer stream.

Conclusion

Good agreement in segmentation and cortical thickness, as well as in quantitative T₁ measurements is reported for this travelling head study, as no significant difference was seen across sites, with expected higher reproducibility obtained within-site. Future work will include processing of the full dataset, including 10 subjects scanned over the 5 UK7T sites.

Acknowledgements

We acknowledge the UK Medical Research Council for funding support (MR/N008537/1). CTR is funded by a Sir Henry Dale Fellowship from the Welcome Trust and the Royal Society [098436/Z/12/B].

References

(1): Tailored RF pulse for magnetization inversion at ultrahigh field. Hurley AC, Al-Radaideh A, Bai L, Aickelin U, Coxon R, Glover P, Gowland PA. Magn Reson Med (2010) 63: 51–58.

(2): Imaging gray matter with concomitant null point imaging from the phase sensitive inversion recovery sequence. Mougin O, Abdel-Fahim R, Dineen R, Pitiot A, Evangelou N, Gowland PA. Magn Reson Med (2016) 76: 1512-1516.

(3): The NMR Phased Array. Roemer PB, Edelstein WA, Hayes CE, Souza SP, Mueller OM. Magn Reson Med (1990) 16: 192–225

(4): Robust T1-Weighted Structural Brain Imaging and Morphometry at 7T Using MP2RAGE. PLoS One (2014) 9(6): e99676

(5): Cortical surface-based analysis. I. Segmentation and surface reconstruction. Dale AM, Fischl B, Sereno MI, Neuroimage (1999) 9, 179-194. (http://surfer.nmr.mgh.harvard.edu/)

(6): SPM 12: http://www.fil.ion.ucl.ac.uk/spm/software/spm12/

(7): Within-Subject Template Estimation for Unbiased Longitudinal Image Analysis. Reuter, M., Schmansky, N.J., Rosas, H.D., Fischl, B. Neuroimage (2012) 61 (4), 1402-1418.

(8): New Developments and Applications of the MP2RAGE Sequence - Focusing the Contrast and High Spatial Resolution R1 Mapping. Marques J, Gruetter R. PLoS ONE8(7): e69294.

Figures

Figure 1: T₁ map data obtained on one subject across three sites (a total of 4 home repeats at site A, plus away site visit at A, B and C) from the LUT T1 estimation based on the PSIR reconstructed image and B₁ correction.

Figure 2: Coefficient of Variation (left) and Intraclass Correlation Coefficient (right) obtained on the volumetric measure of each cortical area as classified by the Desikan–Killiany–Tourville Atlas obtained via Freesurfer. All data from all sites are gathered together on the top row, and only away sites data are represented on the middle row, with the home sites scans represented on the bottom row.

Figure 3: Coefficient of Variation (left) and Intraclass correlation Coefficient (right) obtained on the cortical thickness measure of each cortical area as classified by the Desikan–Killiany–Tourville (DKT) Atlas obtained via Freesurfer. All data from all sites are gather together on the top row, and only away sites data are represented on the middle row, with home sites scan only represented on the bottom row. Please note colour scale between 0 and 1 for the ICC, showing lower correlation in thickness compared to volume (Figure 2).

Figure 4: Dice Similarity coefficient (DSC) of the segmentation computed for each visit compared to the base anatomical (average of the seven visits) considered as the ground truth. Note the higher CoV across sites compared to the CoV of the repeats.

Proc. Intl. Soc. Mag. Reson. Med. 27 (2019)

0237