Introduction

Chemical exchange transfer (CEST) imaging has been reported as clinically useful for brain tumor analysis. Amide proton transfer (APT) imaging has been applied to histological grading, assessment of treatment effectiveness and/or monitoring of brain tumors at 3 Tesla. On the other hand, the nuclear Overhauser effect (NOE) signal is generally observed as an oppositional offset frequency to the CEST signal, of which has been clearly observed with low B₁ power at ultra-high-field MRI.¹ Moreover, various CEST pool models have been developed and demonstrated including APT and NOE. On a 3 Tesla scanner, because a high B₁ setting is used for acquiring adequate APT contrast, the NOE signal which is on the opposite side of APT would be challenging of distinguish with MTR asymmetry (MTR_asym) analysis. Detecting a NOE signal in a brain tumor may be useful for histological classification when compared to APT signals. The purpose in this study was to demonstrate CEST imaging with multi-pool Bloch equation modeling for different types of brain tumors.

Materials and Methods

This study was approved by the institutional review board and performed using five patients with brain tumors (two glioblastomas, one lymphoma, and two radiation necrosis). On a 3.0 Tesla MR system (GE Healthcare), CEST imaging was performed with a single-shot fast spin-echo (SSFSE) sequence and phase cycle radio frequency (RF) preparation; the frequency offset range was from -7 to +7 ppm at intervals of 0.5 ppm. The mean B₁ values of the MT pulses with 1.5 sec RF duration were set at 0.5, 1.0 µT; then, 0.5 µT B₁ was used for B₀ correction. The other imaging parameters were echo time, 27.1 ms; repetition time, 9588 ms; bandwidth, 488 Hz/pixel; field of view, 22 cm; matrix size, 128 × 128; slice thickness, 8 mm. Slice positions were set at center of tumor for a maximum of three slices. Acquired imaging data had motion correction applied to each pixel. Post processing of CEST imaging was performed using the multi-pool model Bloch equation; T₂/T₁ as bulk water, magnetization transfer (MT), amide proton transfer (APT), and nuclear Overhauser effect (NOE). Each parameter estimation of the multi-pool model Bloch equation was as follows;
$$\frac{dm}{dt}=A\cdot{m}+R_{1}{m_{0}}=0\left(Eq.1\right) $$
$$m=-A^{-1}R_{1}m_{0}\left(Eq.2\right)$$
where, A is the matrix representation including parameters of MT, APT, and NOE, m₀ is thermal equilibrium including bulk water. m is a steady state signal, which includes measured Z-spectrum. R₁ is relaxation rate, which was fixed in this fitting. Each parameter was calculated from Z and CEST peak extraction (CPE) spectrum fitting,² this minimized the error between measured and calculated Z-spectrum. Then APT and NOE concentration map were generated. Regions of interest (ROIs) within the tumors and contralateral white matter (WM) were manually delineated in CEST images to be used as a comparison of each of the tumors.

Results and Discussions

Table 1 shows the histological diagnoses of brain tumors and estimated parameters derived from CEST for each patient. Figure 1 shows the relationship of each estimated parameter of each tumor and bilateral normal WM. Mean values for MTR_asym at 3-4 ppm, APT_{3.5 ppm}, NOE_{3.5 ppm}, T₂/T₁, and MT of brain tumors were 0.008 ± 0.017, 7.91 ± 5.42 (×10^-4), 5.24 ± 3.32 (×10^-4), 0.08 ± 0.43, and 0.19 ± 0.85, respectively. Mean values for each parameters of normal WM tissue were -0.009 ± 0.018, 7.00 ± 4.44 (×10^-4), 0.12 ± 3.27 (×10^-4), 0.06 ± 0.11, and 0.53 ± 0.12, respectively. Figure 2 shows the relationship between mean MTR_asym at 3-4 ppm and estimated CEST parameters derived from the multi-pool Bloch equation model of each brain tumor. Figures 3 and 4 show each image and Z-spectrum of each patient with brain tumors. In the glioblastomas case (patient No. 2), we observed high MTR_{asym_3-4 ppm} and APT_{3.5 ppm}, and low NOE_{3.5 ppm} signal areas (Fig. 3). In the other glioblastomas case (patient No. 4), we observed negative value of MTR_{asym_3-4 ppm} and a higher NOE_{3.5 ppm} signal than that found in the other tumors (Fig. 2b&4). Because NOE signal was significantly different between tumors and normal white matter, MTR_asym was affected by NOE. In the future, we must research what the biomechanism is that causes the NOE signal in a brain tumor or normal brain tissue.

Conclusion

Distinguishing APT to NOE signal may lead to more detailed characteristics of metabolism in brain tumors than that of MTR_asym analysis.

Acknowledgements

No acknowledgement found.