3349
Super-Resolution Hybrid Diffusion Imaging (SR-HYDI)1Department of Radiology and Imaging Sciences, Indiana University School of Medicine, Indiana University, Indianapolis, IN, United States, 2Indiana Alzheimer Disease Center, Indianapolis, IN, United States, 3University of Bordeaux, LaBRI, UMR 5800, PICTURA, F-33400 Talence, France, 4CNRS, LaBRI, UMR 5800, PICTURA, F-33400 Talence, France
Synopsis
In this study, we present and validate an efficient pipeline for submillimeter super-resolution hybrid diffusion imaging (SR-HYDI). The pipeline employs a collaborative patch-based super-resolution interpolation approach, which uses self-similarity to drive the reconstruction of diffusion-weighted images. The FA and MD generated from the proposed pipeline are compared against the ground-truth for validation.
Introduction
High-resolution diffusion imaging such as cubic millimeter or less requires long scan times that is usually clinically impractical. To minimize scan time, fewer diffusion encoding directions are preferred limiting advanced diffusion approaches that require multiple diffusion weighting b-values or higher angular sampling resolution to resolve crossing fibers. And even with those long scans, the image quality can still suffer from low signal to noise ratio (SNR) and severe geometric distortion due to long echo spacing in echo-planar imaging sequence. In this abstract, we present a pipeline that shows the feasibility of sub-millimeter super-resolution hybrid diffusion imaging (SR-HYDI) using the well-established collaborative patch-based super-resolution method 1. This method is an adaptive optimization that is based on nonlocal means as well as mean consistency correction, and thus dubbed collaborative and locally adaptive super resolution (CLASR) 1.Methods
MR acquisition and reconstruction
HYDI 2,3 was performed on a healthy volunteer. The diffusion images were acquired on a Siemens Prisma scanner using a single-shot spin-echo EPI with a multiband factor of 3. TE=74.2 ms, and TR=4164 ms, 220 mm field of view, 114 slices and 10:51 (min:sec) acquisition time. A four-shell diffusion imaging with monopolar diffusion scheme was used with b-values 500, 800, 1600, 2600 s/mm2, 134 diffusion directions, and 8 non-diffusion-weighted volumes. Data was acquired in two sets with reversed phase-encode blips. Two of these sets had the resolution of 1.25 x1.25x1.25 mm3 (1.25-cubic-mm), and another two had the resolution of 2.5x2.5x2.5 mm3 (2.5-cubic-mm). We first validated the proposed pipeline by comparing an upsampled resolution of 1.25-cubic-mm (from 2.5-cubic-mm) with the acquisition resolution of 1.25-cubic-mm. The validation uses diffusion tensor imaging (DTI) metrics: fractional anisotropy (FA) and mean diffusivity (MD). Then, we demonstrated the upsampled submillimeter resolution of 0.625-cubic-mm from 1.25-cubic-mm resolution. We demonstrated DTI metrics (FA, MD, axial diffusivity (Da), and radial diffusivity (Dr)), neurite orientation dispersion and density imaging (NODDI) metrics (intracellular volume fraction (Vic) and orientation dispersion (OD)) 4, and a q-space metric (zero-displacement probability (P0)) 5.
Preprocessing
All the DW images were denoised from Rician noise using overcomplete local Principal Component Analysis as proposed in 6. And FSL-topup which is a part of the FSL package version 5.0.11 (FMRIB, Oxford, UK) was used to calculate and correct the susceptibility distortions, and FSL-eddy 7 was used to correct motion. We used the CLASR method to upsample the diffusion images with 2.5-mm-cubic resolution to 1.25-mm-cubic resolution for comparisons. In addition, we used the same method to upsample the 1.25-cubic-mm data to submillimeter 0.625-cubic-mm resolution.
Validation
FA and MD from the upsampled resolution of 1.25-cubic-mm and from the true acquisition resolution of 1.25-cubic-mm were transformed to the standard Montreal Neurological Institute space (MNI) using ANTs 8. Means and standard deviations of the FA and MD values were extracted from 48 cortical regions of interest (ROI) defined in the Harvard-Oxford atlas (Figure 1a) 9. Similarly, means and standard deviations of the FA and MD values were extracted from 48 white-matter ROIs defined in Johns-Hopkins white-matter atlas (Figure 1b) 10.
Results and Discussion
The means and the standard deviations of FA and MD in cortical and white-matter ROIs were almost identical (Figure 2). Figure 3 shows the comparison of the cerebrospinal fluid, gray matter, and white matter density maps 11 of the 2.5-cubic-mm versus that of the 1.25-cubic-mm as well as 0.625-cubic-mm. It demonstrates the ability of the SR-HYDI at the 0.625 -cubic-mm to recover details that exceed those of the original acquisition at 1.25 -cubic-mm. Figure 4 demonstrates the ability of the proposed SR-HYDI method to yield superior quality images at submillimeter (0.625-cubic-mm) with an affordable 20 minutes acquisition time.Conclusion
We presented a submillimeter SR-HYDI algorithm allowing superior quality diffusion maps with minimized partial volume effects.Acknowledgements
The work is supported by grant NIH NIA R01 AG053993.References
1. Coupé, P., Manjón, J., Chamberland, M., Descoteaux, M. & Hiba, B., Collaborative patch-based super-resolution for diffusion-weighted images. NeuroImage 83, 245-261 (2013).
2. Wu, Y.-C. & Alexander, A. L., Hybrid diffusion imaging. NeuroImage 36, 617-629 (2007).
3. Wu, Y.-C., Field, A. S. & Alexander, A. L., Computation of diffusion function measures in q-space using magnetic resonance hybrid diffusion imaging. IEEE Trans Med Imaging 27 (6), 858-865 (2008).
4. Lampinen, B. et al., Neurite density imaging versus imaging of microscopic anisotropy in diffusion MRI: A model comparison using spherical tensor encoding. NeuroImage 147, 517-531 (2017).
5. Kodiweera, C., Alexander, A. L., Harezlak, J., McAllister, T. W. & Wu, Y.-C., Age effects and sex differences in human brain white matter of young to middle-aged adults: A DTI, NODDI, and q-space study. Neuroimage 128, 180–192 (2016).
6. Manjón, J., Coupé, P., Concha, L., Buades, A. & Collins, D., Diffusion weighted image denoising using overcomplete local PCA. PLoS ONE 8 (9) (2013).
7. Andersson, J., Graham, M., Zsoldos, E. & Sotiropoulos, S., Incorporating outlier detection and replacement into a non-parametric framework for movement and distortion correction of diffusion MR images. NeuroImage 141, 556-572 (2016).
8. Avants, B. et al., A reproducible evaluation of ANTs similarity metric performance in brain image registration. Neuroimage 54, 2033-44 (2011).
9. Desikan, R. et al., An automated labeling system for subdividing the human cerebral cortex on MRI scans into gyral based regions of interest. NeuroImage 31, 968-980 (2006).
10. Mori, S., Wakana, S., van Zijl, P. & Nagae-Poetscher, L., MRI Atlas of Human White Matter (Elsevier, Amsterdam, The Netherlands, 2005).
11. Jeurissen, B., Tournier, J.-D., Dhollander, T., Connelly, A. & Sijbers, J., Multi-tissue constrained spherical deconvolution for improved analysis of multi-shell diffusion MRI data. NeuroImage 103, 411-426 (2014).
Figures
