Evaluation of ovarian tumors by proton magnetic resonance spectroscopy at three Tesla

AIM The purpose of this study was to determine the feasibility of acquiring in vivo proton magnetic resonance spectroscopy of ovarian lesions at a magnetic field strength of 3 Tesla (T). The goal was to provide potentially diagnostic biochemical information that may aid in the characterization of ovarian neoplasms detected during clinical magnetic resonance imaging scanning. METHODS: Fourteen patients referred to 2 gynecologic oncologic surgeons were examined in a whole-body 3.0 T clinical scanner using an 8-element phased-array surface coil. Single voxel spectroscopy (SVS) was undertaken after identification of lesions on T1-weighted and T2-weighted imaging. SVS was performed using the point resolved spectroscopy (PRESS) localization technique using a echo time (TE) of 135 milliseconds and repetition time (TR) of 2000 milliseconds and with 192 signal averages. Resonance integrals for the prominent signals from choline-containing compounds and creatine (Cr) were studied and presence of other prominent spectroscopic signals reported. Each SVS acquisition was performed in less than 8 minutes. Magnetic resonance spectral findings were correlated with the detailed pathology reports obtained after resection of each tumor. RESULTS Pathology revealed 7 patients with malignant surface epithelial-stromal tumors, 3 patients with germ cell tumors, 3 patients with benign serous cystadenomas, and 1 patient with a non-neoplastic endometrioma. Spectroscopic data were acquired from 16 voxels in 14 patients. Resonances attributable to choline-containing compounds and Cr were recorded in all malignant tumors and some of the benign tumors. When detected, a choline/Cr integral ratio of greater than 3 was found to indicate that a tumor was malignant in nature, whereas a choline/Cr integral ratio less than 1.5 was found to indicate that a tumor was benign in nature. There was 1 exception, a 13-cm serous cystadenofibroma, where the choline/Cr integral ratio was 3.13. Several other prominent metabolites were recorded including lactate, lipid, and an as yet unassigned resonance (possibly N-acetylaspartate or sialic acid) at 2.07 ppm. CONCLUSIONS Spectroscopy of ovarian masses can be recorded at 3.0 T with acceptable spectral quality and good signal-to-noise ratio. There are stringent technical considerations to be considered in obtaining good spectral quality. Further experience with a larger and more biologically variable range of tumors needs to be undertaken to determine the final clinical utility of this technique, but initial results from this small cohort are promising