Grading of gliomas using 3D CEST imaging with compressed sensing and sensitivity encoding.

PURPOSE: We evaluated the usefulness of three-dimensional (3D) chemical exchange saturation transfer (CEST) imaging with compressed sensing and sensitivity encoding (CS-SENSE) for differentiating low-grade gliomas (LGGs) from high-grade gliomas (HGGs). METHODS: We evaluated 28 patients (mean age 51.0 ± 13.9 years, 13 males, 15 females) including 12 with LGGs and 16 with HGGs, all acquired using a 3 T magnetic resonance (MR) scanner. Nine slices were acquired for 3D CEST imaging, and one slice was acquired for two-dimensional (2D) CEST imaging. Two radiological technologists each drew a region of interest (ROI) surrounding the high-signal-intensity area(s) on the fluid-attenuated inversion recovery image of each patient. We compared the magnetization transfer ratio asymmetry (MTRasym) at 3.5 ppm in the tumors among the (i) single-slice 2D CEST imaging ("2D"), (ii) all tumor slices of the 3D CEST imaging (3Dall), and (iii) a representative tumor slice of 3D CEST imaging (maximum signal intensity [3Dmax]). The relationship between the MTRasym at 3.5 ppm values measured by these three methods and the Ki-67 labeling index (LI) of the tumors was assessed. Diagnostic performance was evaluated with a receiver operating characteristic analysis. The Ki-67LI and MTRasym at 3.5 ppm values were compared between the LGGs and HGGs. RESULTS: A moderate positive correlation between the MTRasym at 3.5 ppm and the Ki-67LI was observed with all three methods. All methods proved a significantly larger MTRasym at 3.5 ppm for the HGGs compared to the LGGs. All methods showed equivalent diagnostic performance. The signal intensity varied depending on the slice position in each case. CONCLUSIONS: The 3D CEST imaging provided the MTRasym at 3.5 ppm for each slice cross-section; its diagnostic performance was also equivalent to that of 2D CEST imaging.

High-resolution systolic T1 mapping with compressed sensing for the evaluation of the right ventricle: a phantom and volunteer study.

To investigate the usefulness of high-resolution systolic T1 mapping using compressed sensing for right ventricular (RV) evaluation. Phantoms and normal volunteers were scanned at 3 T by using a high-resolution (HR) modified look-locker inversion recovery (MOLLI) pulse sequence and a conventional MOLLI pulse sequence. The T1 values of the left ventricular (LV) and RV myocardium and blood pool were measured for each sequence. T1 values of HR-MOLLI and MOLLI sequences were compared in the LV myocardium, blood pool, and RV myocardium. The T1 values of HR-MOLLI and MOLLI showed good agreement in both phantoms and the LV myocardium and blood pool of volunteers. However, there was a significant difference between HR-MOLLI and MOLLI in the RV myocardium (1258 ± 52 ms vs. 1327 ± 73 ms; P = 0.0005). No significant difference was observed between the T1 value of RV and that of LV (1217 ± 32 ms) in HR-MOLLI, whereas the T1 value of RV was significantly higher than that of LV in MOLLI (P < 0.0001). The interclass correlation coefficients of intraobserver variabilities from HR-MOLLI and MOLLI were 0.919 and 0.804, respectively, and the interobserver variabilities from HR-MOLLI and MOLLI were 0.838 and 0.848, respectively. Assessment of RV myocardium by using HR systolic T1 mapping was superior to the conventional MOLLI sequence in terms of accuracy and reproducibility.