A comparative study of quantitative metrics in chemical exchange saturation transfer imaging for grading gliomas in adults

Ruibin Liu,Xianlong Wang,Zhiyong Zhao,Qingqing Wen,Tingting Liu,Dan Wu,Zhibo Wen,Yi Zhang
DOI: https://doi.org/10.1016/j.mri.2022.11.008
IF: 3.13
2022-11-19
Magnetic Resonance Imaging
Abstract:Purpose To evaluate the performance of different chemical exchange saturation transfer (CEST) metrics for grading gliomas with semiautomatically defined regions of interest (ROIs). Methods Thirty-eight adult subjects were included, including 23 high-grade gliomas and 15 low-grade gliomas confirmed by histopathology. The B 0 -corrected CEST z-spectra were first calculated with magnetization transfer ratio asymmetry (MTR asym ) analysis at frequency offsets of 3.5, 3, 2.5, 2, 1.5, and 1 ppm to obtain the fit-free metrics and subsequently fitted with three Lorentzian functions denoting direct water saturation (DS), amide proton transfer (APT), and combined semisolid magnetization transfer and nuclear Overhauser enhancement (MT & NOE) effects to derive the fit-based metrics. Wilcoxon rank-sum test was performed to determine if a statistically significant difference was present in the CEST metrics between low- and high-grade gliomas. Receiver operating characteristic (ROC) curves were used to evaluate the differentiation of CEST metrics between low- and high-grade gliomas. Pearson correlation coefficients were employed to evaluate the correlations of CEST metrics. Results For the fit-free metrics, the highest areas under the curve (AUCs) of 0.85, 0.88, and 0.88, corresponding to MTR asym , MTR normref (normalization by the reference scan), and MTR Rex (subtraction of inverse z-spectra), respectively, were obtained at 3 ppm across various frequency offsets. In addition, the AUCs generated from the fit-based metrics (0.88–0.90) were higher than those generated from the fit-free metrics at 3 ppm. Conclusion The results of this preliminary study indicate that fit-free CEST metrics at 3 ppm are superior to the other frequency offsets for grading human brain gliomas. The fit-based metrics manifested improved differentiation between low- and high-grade gliomas compared to the fit-free CEST metrics.
radiology, nuclear medicine & medical imaging
What problem does this paper attempt to address?