Characterization of parotid gland tissue: a description of an MRI protocol set-up and results of in-vivo applications.

A reliable protocol for proton T2 mapping of the parotid region was set up for future characterization of parotid gland disease. A Carr-Purcell-Meiboom-Gill sequence, phase compensated, available on our 1.5 T imager, was selected and acquisition parameters were chosen on the basis of tests performed on phantoms (agarose-doped gels with T2 in the physiological range). Some experiments were carried out to evaluate the accuracy of T2 calculations for selective and nonselective refocussing pulses, for image uniformity corrections, and for different situations of slice shift and repetition times. The chosen protocol was then applied to in vivo evaluations to check the long-term precision by means of repeated measurements performed on the same subject over a 2-month period. Two or more reference gels were positioned both in the phantom and volunteer at the edge of the field-of-view (FOV). Image postprocessing consisted of an automatic procedure, written by the authors in Fortran 77, that selected the best fit for each pixel between mono- and biexponential decay models, and prepared four parametric images (T2 and Rho slow and fast contribution, Rho being a function of proton density and of T1) that may be used for future elaborations. The phantom experiment results showed an accuracy of 2.5% if a linear correction was performed using the reference gels at the edge of the FOV. No significant differences in accuracy were found between selective and nonselective refocussing pulse, and a homogeneity correction was not demonstrated necessary. The measurements performed on four volunteers showed that the best decaying model for healthy parotid tissue was monoexponential. Evaluated T2 resulted 80.18 +/- 6.11 ms (72.96 +/- 4.97 ms for uncorrected results). Long-term reproducibility of the group of measurements from one volunteer, summarizing all the measurement errors, ranged from 0.9 to 8.5%. The two-way ANOVA that was carried out considering the two classes of volunteers and of parotid positions (right or left) showed that differences found between the two parotids were not significant, while T2 differences among individuals are significant if a probability level higher than 1.1% is accepted. As in this case, the main source of error can be attributed to the biological variations among individuals. Future statistics collected on patients for the T2 evaluations of the pathologic tissue will clarify whether the T2 relaxation is a sufficient parameter for T2 discrimination of healthy and pathologic tissue.