Andreia Vasconcellos Faria1, He Zhu1, Richard Edden1, Peter van Zijl1, Michael Miller2, Susumu Mori1, Peter Barker1
1Radiology, Johns Hopkins University, Baltimore, MD, United States; 2Biomedical Engineering, Johns Hopkins University, Baltimore, MD, United States
We present an automated, quantitative atlas-based analysis of high-resolution, multi-slice magnetic resonance spectroscopic images (MRSI). Using an accurate mapping algorithm, the brains are mapped to a common spatial reference. A 3D parcelation is then applied in each individual to obtain the average spectrum of dozens of structures anatomically defined. The atlas-based MRSI analysis yields results in good agreement with prior literature values obtained using manual analysis techniques. The signal/noise ratio of average parcel based spectra is higher than that of individual voxels. A future application of this method is, for example, the comparison of regional metabolism between patients and controls.