A phantom consisting of tubes with known water concentrations was scanned using 3 T MRI. T2 relaxation data was collected with both gradient echo spin-echo (GRASE) and spin echo sequences, while an inversion recovery experiment provided T1 relaxation data. Voxel-wise T2 distributions were calculated by fitting the T2 relaxation data with a non-negative least squares algorithm that incorporated a correction for errors in flip angle due to B1+ inhomogeneity. TWC was calculated as the sum of the signal in the T2 distribution, corrected for T1 relaxation, relative to that of a tube containing 100% water. TWC from GRASE was compared to that of spin echo in order to test if the accuracy of the TWC measurement was impacted by using additional gradient echoes to fill k-space. Simulations were performed to determine theoretical errors in TWC.
Measured TWC strongly correlated to actual TWC (R = 0.997, p = 9 × 10− 8, mean discrepancy = 1.8%). Accuracy of GRASE and spin echo TWC measurements did not significantly differ. Simulations indicated a mean systematic TWC error of 0.07% and random error of 0.8%, and revealed that the technique performs well in the presence of B1+ inhomogeneity.
This work demonstrates that, using the T2 relaxation decay curve, TWC can be measured to within 3% accuracy at 3 T. Given that T2 relaxation can provide accurate estimates of both TWC and myelin water fraction, multi-echo T2 measurement should be considered a multifaceted approach for assessing pathology and evaluating therapy of central nervous system diseases.