Lei Feng1,2, Hang Li1,3, Kenichi Oishi4, Virendra Mishra5, Minhui Ouyang1, Tina Jeon1, Yun Peng3, Shuwei Liu2, and Hao Huang1,6
1Department of Radiology, Children’s Hospital of Philadelphia, Philadelphia, PA, United States, 2Research Center for Sectional and Imaging Anatomy, Shandong University School of Medicine, Jinan, China, People's Republic of, 3Department of Radiology, Beijing Children’s Hospital Affiliated to Capital Medical University, Beijing, China, People's Republic of, 4Department of Radiology and Radiological Science, Johns Hopkins University, Baltimore, MD, United States, 5Advanced Imaging Research Center, University of Texas Southwestern Medical Center, Dallas, TX, United States, 6Department of Radiology, Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA, United States
Synopsis
The large brain morphological differences of the preterm brain at 33 or
36 postmenstrual week (PMW) to that at 40 PMW makes it necessary to establish
age-specific atlases for preterm brains. In this study, with diffusion MRI (dMRI) data acquisition of 82 preterm
and term normal neonates, we aimed to establish a comprehensive digital atlas
including labeling of gray and white matter for preterm brains at 33 and 36 PMW.
We demonstrated these atlases and showed the differences of the major neural
structures including ganglionic eminence and uncinate fasciculus by comparison to
JHU-neonate-SS atlas for brains at around 40PMW.Target audience
Pediatric neuroradiologist and
pediatric neurologist using preterm brain atlas.
Purpose
During
3
rd trimester of brain development, dramatic structural changes take
place and underlie the brain circuit formation at birth. Preterm births have increased significantly [1]. The large
brain morphological differences of the preterm brain at 33 or 36 postmenstrual
week (PMW) to that at 40PMW makes it necessary to establish age-specific
atlases for preterm brains. In this study, with diffusion MRI (dMRI)
data acquisition of 82 preterm and term normal neonates, we aimed to establish
a digital atlas including comprehensive labeling of all major gray and white
matter structures for preterm brains at 33 and 36 PMW. We demonstrated these
atlases and showed the morphological changes of the major neural structures,
specifically ganglionic eminence (GE) and uncinate fasciculus (unc), from 33PMW to 40PMW with JHU-neonate-SS atlas [2] as the
reference.
Methods
Subjects and data
acquisition: 82 preterm and term normal neonates (age range 31 to 42 PMW) were
recruited and divided into three groups 33, 36 and 39 PMW according to their
ages at scan time. Diffusion weighted image (DWI) was acquired from 3T clinical
scanner (Philips, Best, The Netherland) using a single-shot EPI sequence (SENSE
factor = 2.5) without sedation. The dMRI imaging parameters were: TE=78ms,
TR=6850ms, in-plane field of view= 168x168mm
2, in-plane imaging
resolution=1.5x1.5mm
2, slice thickness=1.6mm, slice number=60, 30
independent diffusion encoding directions, b-value = 1000 sec/mm
2,
repetition=2.
Single-subject templates at
33, 36 and 39 PMW:
Single-subject brain templates at 33, 36 and 39 PMW were established using Diffeomap
(mristudio.org), following the same steps elaborated in the literature [2].
Computation of Jacobian determinants:
Test data of 15 subjects with 5 at 33PMW, 5 at 36PMW and 5 at 39PMW were
registered to the established single-subject templates using large deformation
diffeomorphic metric mapping (LDDMM) registration [3]. Logarithms
of the averaged whole brain Jacobian determinant were computed with LDDMM
transformation matrices.
Establishment
of the DTI atlas at 33 and 36 PMW: The annotation and labeling of all
major gray matter structures were conducted with contrasts from mean
diffusivity (MD) maps and averaged DWI images using guidance of available
atlases [2,4,5]. All major white matter tracts were manually delineated with contrasts
from orientation-encoded DTI color maps and the anatomical information from
tractography. Manual delineation was conducted on axial images with ROIEditor (mristudio.org), followed by adjustment in coronal and
sagittal images.
Identification of
major structural differences based on atlases of 33PMW and 36PMW as well as the
published term brain neonatal atlas: The morphological differences
among major structures were examined by visual inspection from an experienced
neuroanatomist, based on the established atlases of 33PMW and 36PMW in this
study and term brain neonatal atlas (JHU-neonate-SS) [2].
Results
Jacobian determinants: Jacobian determinants,
shown in Fig 1, demonstrate that dramatic brain morphological changes from 33
to 39 PMW. Smallest Jacobian determinants at the diagonal entries in Fig 1 were
found with age of the test data subject same as that of the single-subject
template. Larger absolute Jacobian determinants were found with bigger age
difference between age of the test data subject and that of the single-subject
template.
Gray and white matter DTI
atlas at 33 and 36 PMW: The comprehensive labels of single-subject
template at 33 and 36 PMW are shown in Fig 2 (Cau: caudate nucleus; STG:
superior temporal gyrus; EC: entorhinal cortex; Amyg: amygdala) on the underlying
high resolution MD maps and orientation-encoded DTI color maps, respectively.
The atlases cover as many as 22 brainstem and cerebellar structures, 39
cerebral white matter structures, 12 deep gray matter structures and 52
cerebral cortical structures.
Dramatic
structural changes from 33PMW to 40PMW: Dramatic morphological changes
were found in most gray and white matter structures. As shown in Fig 2a-2c, the
size decreases of GE from 33PMW to 36PMW and its final disappearance at around
40PMW are most striking. Dramatic size increase of
unc is also highlighted in Fig 2d-2f.
Discussion and
conclusion
The results of the Jacobian determinant showed it is critical to
establish a brain atlas on every gestational week. Moreover, the dramatic
morphological changes of GE and unc shown in Fig 2 suggest that major neural
structures still undergo rapid growth from 33PMW to 40PMW. The established
age-specific comprehensive gray and white matter atlases of 33PMW and 36PMW
brains may be used for not only understanding this highly ordered development
processes but also serving as clinical anatomical references to detect neural abnormalities.
Acknowledgements
This study is funded by NIH MH092535 and MH092535-S1. References
[1] Woodward et al. (2006) N Engl J Med 355:685. [2] Oishi et al.
(2011) Neuroimage 56: 8. [3] Miller et al. (2012) Annu Rev Biomed Eng 4:375. [4] Bayer and Altman (2004) CRC Press. [5] Huang et al. (2009) J Neurosci 29: 4263.