Non-invasive MR thermometry by 2D spectroscopic imaging of the Pr[MOE-DO3A] complex.

Future progress in regional hyperthermia requires a practical method for non-invasive thermometry. In magnetic resonance tomography, spin density, T1 relaxation time, diffusion coefficient and proton resonance frequency are candidates to measure temperature distributions. When used clinically in the pelvic region, all these methods are compromized by artifacts arising from different tissues, tissue alterations under hyperthermia, physiological and random movements, inhomogeneities, drift phenomena, and field instabilities. In this study a paramagnetic complex was evaluated, Pr[MOE-DO3A], with praseodymium as central atom, similar to common gadolinium containing MRI contrast media. The temperature dependence of its methoxy side group approximately -24 ppm downfield from the water resonance at 25 degrees C was employed to determine 2-D temperature distributions in a cylindrical agar phantom containing 9.5 mM of Pr[MOE-DO3A]. The phantom was heated externally through a water jacket creating a stationary temperature distribution throughout the phantom. At first, the correlation between temperature and the chemical shift of the methyl group of the lanthanide complex Pr[MOE-DO3A] was determined. Calibration curves obtained exhibited a linear relationship of 0.12 +/- 0.01 ppm/degree C, nearly independent from the surrounding medium. Local temperature distributions were determined employing the spatially resolved method of spectroscopic imaging (SI). 2-D spectroscopic images for three orthogonal slices were obtained by narrow-band excitation and 16 phase encoding steps in two dimensions. The FOV was 180 mm and the slice thickness in all cases was 20 mm for maximal spatial temperature resolution (11.2 x 11.2 mm2). The results indicate a measurement time of about 5s per acquisition under the following conditions: An estimated concentration of 1 mmol/l, a reduced matrix size of 8 x 8, and a reduced repetition time of 3 x T1 (TR approximately 85 ms). Those SI measurements produced a SNR of approximately 4 per acquisition, a measurements duration of 10-20 s, equivalent to two to four acquisitions per spectrum, seem sufficient for online temperature monitoring during hyperthermia. The in vitro data suggest the spectroscopic temperature measurement utilizing a temperature-sensitive Pr[MOE-DO3A] complex with a therapeutically realistic concentration of 1 mmol/l to be suitable for clinical use. Compared to the methods tested so far (rho, T1, diffusion, proton resonance), the method presented has the unique advantage of being less susceptible to artifacts. The competing methods of non-invasive thermometry employing magnetic resonance imaging are currently being investigated using the same experimental setup.