GeometryPhantom:3TBRIC
Geometry Phantom scanning and evaluation from the 3T BRIC scanner on neonate and baby images
This is a discussion about phantom scans for baby sibs scans on the 3T BRIC scanner. Kathy Wilber scanned the geometric phantom. Rachel mentioned, that the shape looks rather funky compared to the scans they get from Duke. Not sure if this due to the increased distortion from the 3T, the sequence used to scan it, how it is supposed to be shaped, or who knows what. The zip file of dicom slices is located here: /Autism/autism/babysibs/CAN_study/TID_03/TID_03-1/archive/series-2_mprage_2mm_ax_phantom.zip
An example of the geometric phantom scans we get from Duke is located here: /Autism/fragilex/unc/7050-003/7050-003-01/phantom_3Dgeometric/7050-003-01_phant3Dgeom_dcm.zip [or 7050-003-01_phant3Dgeom.gipl.gz]
Martin looked at the phantom and this looks like the non-cylindrical shape is due to geometric distortions in the RF coil. Thus the "widening" towards the top and bottom are not the shape of the phantom (physically) but likely the distortion to the image when an mp-rage is taken with the same coil as the ones used for the babys. Now that does not mean that the image is not suitable, as only in the part of the image that is important needs to be free of distortions. To illustrate that set of snapshots using rview to overlap the phantom scan with the baby's mp-rage scan can be seen below. All the snapshots have the same zoom, so you can directly compare them. When you look at the overlay visualizations, you can see that for the area where the baby brain is we have an image quite free of distortions (this view crops the phantom scan to the size of the baby scan). When you look at the uncropped view of the phantom scan, you can see that most of the distortion is either outside of the scanning area of the baby image, or it is really far above the head or in the neck area. Probably we should acquire a phantom scan that uses exactly the same protocol as the baby scan (currently the phantom scan has 2mm slice thickness and the baby scan has 1 mm slice thickness).
These images are a helpful verification that the area of distortion does not seem to affect the scans of 1 year old subjects. As Martin suggests, we should also scan the geometric phantom using the same sequence parameters as the subject images. It would be most informative to have scans of the phantom with all three brain sequences: mp-rage, 3D T2, and the PD+T2 FSE.
Another issue to consider is the longitudinal nature of the baby sibs study -- as the subjects get older, their brains will grow. Will the larger brain size at age 2 mean we might have a problem with field distortions at that age? It would be helpful to look at some 2 year old scans, which would be the oldest age in the baby sibs study.
