Optimization for automatic whole brain volumetry An MPRAGE image contrast study of a 12- and 32 channel head coil at 3T

Abstract: Introduction: Regional and global brain volumes are of great interest in many different research projects, not least in Alzheimer’s research. The purpose of this project is to optimize the MPRAGE sequence for automatic and repeatable segmentation of human brain tissue. The volume is calculated for the whole brain, white matter (WM), grey matter (GM) and cerebrospinal fluid (CSF). Materials and methods: MPRAGE (Magnetization Prepared Rapid Acquisition by Gradient Echo) is a T1 weighted 3D sequence with high resolution showing fine anatomical details of the brain. The high resolution and high T1 contrast makes it a good choice for tissue specific segmentation of the brain. Parameters in the sequence that are altered to manipulate the contrast relationships between WM/GM and GM/CSF are flip angle (FA) and inversion time (TI). FA of 8-11° are used for TI varying between 700-1000 ms. The contrast relationship between different types of tissue is an important factor to achieve a successful segmentation. The project includes a volunteer study with 9 healthy subjects aged 22-41 (5 men and 4 women) and is carried out on a 3T Magnetom Trio (Siemens, Medical Solution, Erlangen, Germany). The study also contains a comparison between a 12 channel and a 32 channel Matrix head coil and a test of repeatability of the optimized sequence. Brain Map Statistics is a locally developed software used for fully automatic brain tissue segmentation. The image histograms are analyzed for quantification and classification of brain tissue. The images are also evaluated by an experienced neuro-radiologist in terms of tissue contrast for clinical diagnosis. Results: Contrast differences between WM-GM up to 56% could be found when the contrast parameters of the sequence were changed. It is shown that small FA and short TI results in higher WM-GM contrast. The best contrast parameters were found to be TI = 850 ms and FA = 8° from the optimization. All images acquired with FA and TI in the range of 8-10° and 800-900 ms respectively was approved for clinical use. The repeatability study resulted in a coefficient of variation ≤ 0.7% for the total brain volume and BPF with both coils. Conclusions: No other studies about optimization of the MPRAGE sequence, also containing a comparison between two different multi channel coils has been found. The comparison between the 12- and the 32 channel coils show that the contrast variation as a function of FA and TI is larger with the 32 channel coil. This indicates that the 32 channel coil is in a bigger need of optimization. The anatomical details were slightly better with the 32 channel coil when small FA and short TI were used. The relative contrast between WM-GM and GM-CSF was higher for the 12 channel coil. The repeatability result needs further investigation