1268
Multi-shell neonatal brain HARDI template1Division of Imaging Sciences & Biomedical Engineering, King's College London, London, United Kingdom, 2Department of Perinatal Imaging & Health, King's College London, London, United Kingdom, 3FMRIB Centre, University of Oxford, Oxford, United Kingdom, 4Nuffield Department of Clinical Neurosciences, University of Oxford, Oxford, United Kingdom
Synopsis
We describe a method for creating a group template of the developing brain using advanced multi-shell high angular resolution diffusion (HARDI) data. We decompose the signal into an anisotropic CSF-like and a white matter-like directional component and build an unbiased template of those tissue types from 27 healthy term control babies acquired as part of the Developing Human Connectome Project (gestational age: 40.2+-1.4 weeks). This template will facilitate the analysis of microstructural features at a group level and allow longitudinal investigations into healthy and pathological brain maturation.
INTRODUCTION
There is increasing interest in studying the developing brain using advanced multi-shell diffusion analysis methods, due to their potential to report on microstructural features not visible using other modalities. These analyses require group-wise non-linear registration of multi-shell High Angular Resolution Diffusion Imaging (HARDI) data over large numbers of subjects to a common group-average space (the template space). In this work, we describe a method for generating a high-quality neonatal multi-shell HARDI group template of the developing brain at term equivalent age, which will form the foundation for group and longitudinal analysis of brain development in normal and pathological cohorts.METHODS
The approach used in this study relies on the multi-tissue constrained spherical deconvolution technique (CSD) to decompose the diffusion signal into distinct, orientationally-resolved tissue types1. The resulting orientation density functions (ODFs) are then registered across subjects using a non-linear diffeomorphic algorithm tailored for ODFs, including appropriate reorientation of the ODFs to ensure preservation of the underlying topology2. This approach is then used to construct a group average template using an iterative registration and averaging approach, described below.
We use data from 27 healthy term control babies acquired as part of the Developing Human Connectome Project3,4 (gestational age at scan: 40.2+-1.4 weeks), consisting of four shells with b-values of 0, 400, 1000 and 2600s/mm2 using a dedicated neonatal coil12. The data were preprocessed by removal of motion-corrupted volumes, PCA-based denoising5, distortion correction and outlier replacement6, bias field correction7 and intensity normalisation across datasets.
Multi-tissue CSD requires the estimation of tissue-specific response functions. In this study, we focus on the white matter (WM) and cerebrospinal fluid (CSF) components, due to the ambiguities in differentiating between grey matter (GM) and WM in this age range.These responses were obtained per subject using CSF and WM/GM tissue probability maps generated using segmented co-registered T2-weighted images8. The WM responses were estimated from the single fibre voxel mask obtained using the technique described in [9], using the b=2600s/mm2 shell. The CSF responses were similarly estimated by selecting the 100 voxels with the highest signal attenuation between the averaged b=0 and b=2600s/mm2 shells within voxels with greater than 80% CSF probability.
These response functions were then averaged across subjects, and used in the multi-tissue CSD analysis to provide WM ODFs and CSF-like density maps for each subject1. These maps were then used to generate the atlas, by iteratively co-registering each subject to the group average template on a multi-resolution pyramid with scale factors ranging from 0.48 to 1.6 and with increasing angular resolution of the ODF compartment (maximum spherical harmonic order lmax=4). Affine and diffeomorphic registration with ODF reorientation2 was performed using a metric combining both tissue types simultaneously with equal contribution to the diffeomorphic update.
RESULTS
The approach proposed here provided good alignment across subjects on visual inspection, with clear definition of fine features such as the motor strip, brainstem, and anterior commissure (Figures 1-3).
We also observed clear differences between the neonatal and the adult cases. In the neonate, early maturing white matter in the cerebellum and cerebellar peduncle (Figure 1), corpus callosum and corticospinal tracts (Figure 2) show high WM and low CSF density. Conversely, parts of the periventricular deep white matter show low WM and high CSF density in the neonate template, in contrast to the adult case where high WM density is observed (Figure 3). Finally, we note the clear radial organisation of fibres in the cortex, as expected from the literature10 (Figure 4).
DISCUSSION
The atlas provided by our approach matches the expected anatomy and composition of the developing brain in the neonatal period, with high water content overall and higher anisotropy in the cortex. We chose to use a simple model with two tissue types given the difficulty in distinguishing between WM and GM, and other ongoing developmental processes (e.g. WM neurogenesis and pruning, membrane proliferation) occurring during this period. This will be further explored in future work.CONCLUSION
We present a methodology to produce a high-quality multi-shell HARDI template of the human brain at the age of birth, in the form of the white matter ODF and CSF-like density. This framework forms the foundation for advanced longitudinal and/or group-wise investigations into brain maturation in a number of important pathologies.Acknowledgements
ERC funded dHCP project (Grant Agreement no. 319456) and MRC strategic funds (MR/K006355/1) and the GSTT BRC.References
[1] Jeurissen, B., Tournier, J.D., Dhollander, T., Connelly, A., Sijbers, J., 2014. Multi-tissue constrained spherical deconvolution for improved analysis of multi-shell diffusion MRI data. Neuroimage 103, 411–426.
[2] Raffelt, D., Tournier, J.-D., Fripp, J., Crozier, S., Connelly, A., Salvado, O., 2011. Symmetric diffeomorphic registration of fibre orientation distributions. Neuroimage 56, 1171–1180.
[3] J.-D. Tournier, E. Hughes, N. Tusor, S.N. Sotiropoulos, S. Jbabdi, J. Andersson, D. Rueckert, A.D. Edwards, J.V Hajnal, 2015. Data-driven optimisation of multi-shell HARDI, ISMRM 23rd Annual Meeting & Exhibition
[4] Jana Hutter, Jacques-Donald Tournier, Emer J. Hughes, Anthony N. Price, Lucilio Cordero-Grande, Rita G. Nunes, Rui Pedro A. G. Teixeira, Serena J. Counsell, Jesper L. R. Andersson, Daniel Rueckert, A. David Edwards, and Jo V. Hajnal 2015. Optimized multi-shell HARDI acquisition with alternating phase encoding directions for neonatal dMRI, ISMRM 23rd Annual Meeting & Exhibition
[5] Veraart, J., Novikov, D.S., Christiaens, D., Adesaron, B., Sijbers, J., Fieremans, E., 2016. Denoising of diffusion MRI using random matrix theory. Neuroimage S1053-8119(16)30394-9.
[6] Andersson, J.L.R., Graham, M.S., Zsoldos, E., Sotiropoulos, S.N., 2016. Incorporating outlier detection and replacement into a non-parametric framework for movement and distortion correction of diffusion MR images. Neuroimage 141, 556–572
[7] Tustison NJ, Avants BB, Cook PA, Zheng Y, Egan A, Yushkevich PA, Gee JC, 2010 N4ITK: improved N3 bias correction. IEEE Trans Med Imaging. 2010 Jun;29(6):1310-20
[8] Makropoulos A, Gousias IS, Ledig C, Aljabar P, Serag A, Hajnal JV, Edwards AD, Counsell SJ, Rueckert D, 2014. Automatic whole brain MRI segmentation of the developing neonatal brain, IEEE Trans Med Imaging. 2014 Sep;33(9):1818-31.
[9] Tournier, J.-D.; Calamante, F. & Connelly, A. Determination of the appropriate b value and number of gradient directions for high-angular-resolution diffusion-weighted imaging. NMR Biomedicine, 2013, 26, 1775-1786
[10] McKinstry, R.C., Mathur, A., Miller, J.H., Ozcan, A., Snyder, A.Z., Schefft, G.L., Almli, C.R., Shiran, S.I., Conturo, T.E., Neil, J.J., 2002. Radial organization of developing preterm human cerebral cortex revealed by non-invasive water diffusion anisotropy MRI. Cereb. Cortex 12, 1237–43.
[11] Judaš, M., Radoš, M., Jovanov-Miloševic, N., Hrabac, P., Štern-Padovan, R., Kostovic, I., 2005. Structural, immunocytochemical, and MR imaging properties of periventricular crossroads of growing cortical pathways in preterm infants. Am. J. Neuroradiol. 26, 2671–2684.
[12] Hughes, E.J., Winchman, T., Padormo, F., Teixeira, R., Wurie, J., Sharma, M., Fox, M., Hutter, J., Cordero-Grande, L., Price, A.N., Allsop, J., Bueno-Conde, J., Tusor, N., Arichi, T., Edwards, A.D., Rutherford, M.A., Counsell, S.J., Hajnal, J. V., 2016. A dedicated neonatal brain imaging system. Magn. Reson. Med. doi:10.1002/mrm.26462
Figures
