John DeWitt Port1, Ileana Hancu2,
Heidi Alyssa Edmonson1, Zhonghao Bao3, Mark A. Frye4
1Radiology, Mayo Clinic, Rochester, MN,
United States; 2GE Global Research Center, Niskayuna, NY, United
States; 3Information Services, Mayo Clinic, Rochester, MN, United
States; 4Psychiatry, Mayo Clinic, Rochester, MN, United States
Various
methods have been used to correct for the amount of CSF within spectroscopic
voxels. However, it remains unclear
which method is best. We performed CSF
correction on an MRS dataset comparing depressed psychiatric patients to
normal controls, using the ratio to creatine as well as two
anatomically-based CSF correction methods.
All three CSF correction methods yielded significant results for most
statistical comparisons; ROC analysis demonstrated no single CSF correction
technique to be better than the others.
If the metabolite value used in the denominator is stable, ratios may
actually improve statistical sensitivity relative to anatomically-based CSF
correction methods.
Keywords