Precision, accuracy, and image plane uniformity in NMR relaxation time imaging on a 1.5 T whole-body MR imaging system.

In order to evaluate the accuracy, precision and image plane uniformity a phantom study was undertaken on a 1.5 T whole-body MR imaging system (Philips Gyroscan S15 HP). The manufacturers algorithm for obtaining relaxation time images was used. This method uses Ratios and Least Squares (RLSQ-algorithm) on data obtained from a combination of a multi-echo (CPMG) and an inversion recovery pulse sequence, yielding a calculated T1-, T2- and a proton density image simultaneously. T1 and T2 accuracy was measured in phantoms consisting of CuSO4 and MnCl2 in aqueous solution in different concentrations. The results were compared to a reference T1 and T2 obtained by spectroscopy. The accuracy error over the entire relaxation time range was expressed as the root mean square of the modulus of the difference between the reference and the image measurement and was 2.3% for T1 and 4.1% for T2. Precision or reproducibility was measured by 11 double estimations on 11 samples. The precision error was 0.35% for both T1 and T2. Image plane uniformity, i.e., the homogeneity of pixel values throughout the image plane, was measured by 3 consecutive image readings from a phantom of 19 cm in diam. in 8 directions 45 degrees apart and 1, 3, 5, 7, 9 cm from the center. Errors in the imaging plane were expressed as the deviation from the center value. For both T1 and T2 the maximum deviation was less than 5% over the entire image in the transverse plane. It is concluded that when the pulse sequence timings are carefully optimized, the mixed imaging sequence in combination with the RLSQ algorithm used in this MRI system is a reliable and precise means of obtaining relaxation time data.