EPI based distortion- and resolution-matched T1-Like anatomy for submillimeter-resolution fMRI at 7T

Poster No:

W393 

Submission Type:

Abstract Submission 

Authors:

Adnan Shah1, Guoxiang Liu1, Takashi Ueguchi1

Institutions:

1CiNet, NICT, Suita City, Osaka, Japan

E-Poster

Introduction:

Submillimeter-resolution isotropic fMRI (1) at 7T or higher requires the use of anatomy that avoids any spatial resampling of the functional EPI data during co-registration to allow discriminating BOLD responses at the level of cortical columns and layers. We developed an anatomical imaging technique called DAIREPI (2) that generates distortion- and resolution- matched anatomy aligned to functional imaging. In this study, we examined echo-planar imaging (EPI) for the generation of T1-Like anatomy based on a post-processing procedure involving segmentation, EPI inversion, image fusion and denoising performed on the acquired brain anatomy at submillimeter-resolution. The resulting image is a distortion- and resolution- matched T1-Like anatomy we named DRAEPI that bypasses the need for any spatial resampling during co-registration and allow 3D anatomical and surface-based reconstruction for submillimeter-resolution fMRI analysis.

Methods:

An adult human brain was scanned using 2D multi-shot EPI sequence (#shots = 16) on a Siemens MAGNETOM 7T scanner with a 32-channel phased array head coil (Nova Medical, MA, USA) to obtain 60 volumes at a spatial resolution of 0.6 × 0.6 × 0.6 mm3 followed by the same resolution two structural scans covering the upper- and lower halves of the human brain. The same pulse sequence was used for acquiring both functional and structural scans with aligned slice acquisitions except differing in parameters unrelated to distortions i.e. TR, FA and TE. Acquisition parameters for functional / 2 x structural scans were as follows: TR = 1000/6600.1 ms, TE = 24/22 ms, Flip Angle (FA) = 50°/80°, # Slices = 18/128, Averages = 1/1, Acquisition Matrix = 320 x 320. A standard block design task of checkerboard visual stimulation was performed in the experiment with ON/OFF stimulation. The following post-processing was applied to 2D EPI anatomical data: Stitching the upper- and lower halves anatomical EPI volumes offline in Matlab (3) to generate whole-brain EPI anatomy followed by tissue segmentation in SPM12 (4) and T1-Like image reconstruction (2, 5) as a first step. In the 2nd step, the whole-brain EPI anatomy was inverted, scaled up and added to the image obtained from step 1 in FSL (5). Finally, the resulting image is denoised (6, 7) generating a high-contrast EPI based distortion- and resolution-matched T1-Like anatomy (DRAEPI). This DRAEPI anatomical image at 0.6 mm isotropic resolution is AC-PC transformed in BrainVoyager 21.0 and processed for generating high-resolution cortical surfaces for both hemispheres utilizing the advanced segmentation tools (8). fMRI raw data were reconstructed from k-space in Matlab prior to analysis. fMRI pre-statistics processing includes MCFLIRT motion correction and high-pass filtering (5). No spatial smoothing was performed. Furthermore, the functional data were denoised using ICA (9). Generalized linear model (GLM) analysis was applied with stimulation ON/OFF as a binary regression variable.

Results:

Figure 1 shows an axial slice from the acquired (A) task-based functional EPI, (B) structural EPI, (C) the inverted EPI, and (D) the resulting T1-Like DRAEPI anatomical image. Figure 2 shows the task-based BOLD activity overlaid on this anatomical image and the resulting sphere obtained by inflating the cortical surface.
Supporting Image: abs-figure1_new.jpg
Supporting Image: abs-figure2_new.jpg
 

Conclusions:

The proposed novel distortion- and resolution-matched T1-Like DRAEPI anatomy alleviates the problem of distortion-mismatch between function and anatomy. DRAEPI bypasses the need for spatial resampling associated with conventional co-registration and ascertains the reliability of activation maps. It offers substantial benefits for voxel-wise submillimeter-resolution fMRI. Moreover, the proposed anatomy can be used for layer-specific fMRI analysis and interpretations.

Acknowledgements
This study was supported in part by Japan Society for the Promotion of Science (JSPS) Grants-in-Aid for Scientific Research "KAKENHI" (Grant Numbers JP26282223 and JP26350471).

Imaging Methods:

BOLD fMRI 1
Imaging Methods Other 2

Keywords:

Other - Ultra-high-field MRI, 7T, Submillimeter-resolution fMRI

1|2Indicates the priority used for review

My abstract is being submitted as a Software Demonstration.

No

Please indicate below if your study was a "resting state" or "task-activation” study.

Task-activation

Healthy subjects only or patients (note that patient studies may also involve healthy subjects):

Healthy subjects

Was any human subjects research approved by the relevant Institutional Review Board or ethics panel? NOTE: Any human subjects studies without IRB approval will be automatically rejected.

Yes

Was any animal research approved by the relevant IACUC or other animal research panel? NOTE: Any animal studies without IACUC approval will be automatically rejected.

No

Please indicate which methods were used in your research:

Functional MRI

For human MRI, what field strength scanner do you use?

7T

Which processing packages did you use for your study?

SPM
Brain Voyager
FSL

Provide references using author date format

[1] Liu G et al (2018), ‘Block-Interleaved Segmented EPI for voxel-wise high-resolution fMRI studies at 7T’, Proceedings of the International Society for Magnetic Resonance in Medicine, pp. 5450. [2] Shah A et al (2018), ‘Distortion-Matched Anatomical Imaging using Inversion Recovery-Prepared EPI for high-resolution fMRI’, OHBM Proceedings, pp. 1727. [3] MATLAB (R2016b): The MathWorks, Inc., Natick, Massachusetts, USA. [4] SPM12: http://www.fil.ion.ucl.ac.uk/spm/, Wellcome Trust Center for Neuroimaging, London, England. [5] FSL 5.0.7, Analysis Group, FMRIB, Oxford, UK. [6] Coupe P et al (2012), ‘Adaptive Multiresolution Non-Local Means Filter for three-dimensional magnetic resonance image denoising’, IET Image Processing, vol. 6, no. 5, pp. 558-568. [7] Coupe P et al (2010), ‘Robust Rician Noise Estimation for MR Images’, Medical Image Analysis, vol. 14, no. 4, pp. 483-493. [8] Brain Innovation B.V., Maastricht, the Netherlands. [9] Beckmann CF et al (2004), ‘Probabilistic Independent Component Analysis for Functional Magnetic Resonance Imaging’, IEEE Transactions on Medical Imaging, vol. 23, no. 2, pp. 137-152.