According to the latest guidance by UCR IRB, all human studies in CAN are suspended as of Monday (March 16, 2020) for COVID-19 precautions until further notice .
NEW!!! We are looking for postdocs !
QSM-based Dentate Nucleus Atlas
We used QSM maps to create dentate nucleus atlases for use as seed points in functional magnetic resonance imaging (fMRI) and diffusion tensor imaging (DTI) connectivity studies. In addition, the atlases may be applied to examine iron load by defining a region of interest in R2* map. Follow the links below for access to the atlases. To download the atlases, right click on the links and select the "Save Link As... option"
1. Dentate nucleus atlas with 1mm x 1mm x 1mm resolution in MNI152 space.
2. Dentate nucleus atlas with 2mm x 2 mm x 2mm resolution in MNI152 space.
Please cite the following publication if this atlas is used in your study:
He, N., et al. "Improved neuroimaging atlas of the dentate nucleus." The Cerebellum. In press
Background.
QSM provides better contrast in delineation of dentate nucleus than standard T2-weighted images or susceptibility weighted images (SWI). This is illustrated in the figure below where the dentate nucleus is outlined in red and overlaid on T2-weighted, SWI, and QSM images.
The details for constructing the atlas are described in the following paragraphs.
Data acquisition parameters.
Thirty-eight healthy elderly subjects (15 male and 23 female; age range: 45-75) were used to create our dentate nucleus atlas. The mean age and standard deviation of the cohort is 64.1 years and 7.5 years, respectively. Data were acquired on a 3.0 T GE Signa HDxt scanner equipped with an eight-channel phase array head coil.
A three dimensional multi-echo gradient echo (GRE) sequence was used to obtain data suitable for quantitative susceptibility mapping (QSM) analysis. Imaging parameters for the multi-echo GRE sequence were: repetition time: 59.3 ms; 16 echoes with the first echo time of 2.7 ms and an echo spacing of 2.9 ms; pixel bandwidth: 488.3 Hz/pixel; flip angle: 12 degrees; resolution: 0.86 mm x 0.86 mm x 1 mm; matrix size 256 x 256 x 136. Slices were oriented axially.
Whole brain images were also acquired with a T1-weighted fast spoiled gradient echo (FSPGR) sequence. These images were used for common space registration and imaging parameters for this sequence were: repetition time: 5.5 ms; echo time: 1.7 ms; inversion time: 450 ms, flip angle: 12 degrees; voxel size: 1 mm x 1 mm x 1 mm. Slices for this sequence were oriented axially.
The image below shows a T1-weighted image, T2-weighted image at an echo time of 20 ms, and QSM image from our data:
Image processing.
QSM images were reconstructed from the multi-echo GRE data using STI Suite and the dentate nucleus was segmented using a threshold of 0.025 ppm. A QSN image of the segmented dentated nucleus highlighted in red is shown below:
Dentate nuclei of subjects were transformed to Montreal Neurological Institute (MNI) 152 space using the FMRIB's Software library (FSL) software package using the MNI registration procedure as follows. First, the FMRIB's Linear Registration Tool (FLIRT) was used to perform a rigid body registration (degrees of freedom = 6) between individual T2-weighted images (T2-weighted images share the same "space" as QSM images) with individual T1-weighted images, Next, dentate nucleus masks in QSM space were transformed to their individual T1-weighted images using this transformation. FMRIB's Nonlinear Image Registration Tool (FNIRT) was used to construct a transformation between T1-weighted images and MNI T1-weighted images. To construct this transformation, brain extracted individual T1-weighted images were aligned to the brain extracted MNI T1-weighted image using an affine transformation (degrees of freedom = 12) in FLIRT. Then FNIRT was used to derive non-linear transformation between individual T1-weighted images and MNI T1-weighted images. T1-weighted dentate nucleus masks were transformed to MNI T1-common space with the non-linear transformation using FNIRT. Finally, the dentate nucleus atlas was created by adding the 38 dentate nuclei in MNI space together and normalizing the resultant image. This process is described schematically below:
The resultant dentate nucleus atlas is illustrated in the image below:
