Introduction

The ¹⁷O MRSI/MRI at UHF is promising for direct measurement of oxygen metabolism and perfusion in both animal and human brains using non-radioactive tracer, although it is challenging to achieve high resolution due to short field-independent quadrupolar relaxation times and low SNR.^1,2 In connection with the low ¹⁷O gyromagnetic ratio, the thermal tumbling of H₂¹⁷O resonance makes its linewidth/T₂* dependent on the local chemical environment (such as temperature), more so than macro field inhomogeneities. This motivates this exploration of a robust yet simple method using the relaxation of H₂¹⁷O resonance to measure of relative temperature changes in phantoms and tissues as an alternative imaging tool to other existing approaches.³

Methods

Phantom validation on a human 7T scanner
Phantom measurements were performed at 7T on a human clinical system (Terra, Siemens, Erlangen, Germany) from 2L 50mM NaCl solution with natural abundance H 2 17 O glass sphere with a dual tuned 1 H/ 17 O quadrature coil (loop diameter 15cm). CSIs with FSW technique were acquired at TR=70ms, 45° flip angle, 20cm 3 FOV, of 19x19x15 matrix size, spectral bandwidth of 30 kHz/1024pts, 4 averages in a total acquisition time of 4min per 3D CSI volume. The phantom was immersed in a warm water bath(+42°C), as measured with a thermocouple sensor. The temperature was then measured at further timepoints between the acquisitions at 39°C(t=15min) 33.5°C (t=86min) and 31.9°C (t=113min) and between points interpolated for the actual times of the measurements by a mono-exponential function (40-exp -0.0018 ).
Acquisitions on a preclinical 16.4T system
Acquisitions were performed using FID and CSI at 16.4T scanner (Paravision, Bruker MRI Neo, Ettlingen, Germany) from a 77mM NaPiO₄ cylindric water phantom of natural abundance H₂¹⁷O. RF signal detected with a custom-built 1 cm diameter dual-turn, double tuned ¹H/¹⁷O surface coil. FIDs were acquired with a TR=70ms, 256 BW 4kHz/512 points, after FA45 degree excitation by a 0.1ms hard pulse. 256 averages were accumulated within a single set in 17s with a series of 100 repetitions were measured in 29 minutes each. Several of measurement series were used to observe changes in equilibration to room temperature (e.g. 39-27°C). The temperature inside the cylindric phantom was measured by a shielded thermocouple sensor on the opposite of the surface coil.
In-vivo and post-mortem rodent validation
A male Sprague-Dawley rat (233g) was anesthetized with Isoflurane and artificially ventilated maintained at 38°C by a water bath and then euthanized by KCl injection. Procedures were in accordance with the protocol approved by UMN IACUC. A temperature probe was inserted into the right ear canal to approximate brain temperature. Several CSI acquisitions of 5-6min each volume were performed on the rat head while the body cooled down over the course of approx. 120min and the brain temperature was read at regular intervals until 26.3°Cbetween weighted CSI volumes (TR=21ms, TE 1ms, RF 0.2ms/FA 20°/9W, FOV 25x25x25mm³/Matrix 8x8x8, BW 30kHz/512 points in 17.4ms spectral duration, nt=256 averages within a total time of 5min 53s).

Results

Figure 1 illustrates the 7T phantom results showing representative CSI of H₂¹⁷O spectra, resonance linewidth maps measured at 38°C and 32°C, and their ratio map. The results indicate a high sensitivity of the linewidth to the sample temperature changes. Preclinically acquired FIDs from a phantom at 16.4T were phase corrected manually and the linewidths determined at various timepoints and plotted against temperature over 30-minute duration (Figure 2). Global linewidths overall ranged from 45.4 Hz at 39.3°C up to 91.2 Hz at 8°C. The rat brain CSI linewidths were in the range of 120-200 Hz depending on location and temperature and they were fitted to the linear function between linewidth and temperature changes to plot slopes in two representative voxels (Figure 3), indicating a slow slope in the central brain area as compared to the peripheral areas corresponding to faster cooling.

Discussion and Conclusions

The sensitivity to measure FHWM linewidth and/or indirectly by fitting the slope of T₂* ² needs a moderate sampling rate. Expectedly the comparison between different field strengths confirmed very similar linewidth, despite small but significant differences in the phantom shape, solvents, and acquisition RF coil/sampling schemes.⁴ This temperature mapping method should be beneficial to application in brain imaging since the temperature in the brain is extremely narrowly auto-regulated both from a temperature standpoint, but also from several other physiological parameters (pH and oxygenation being the most important ones), resulting significant spatial temperature variation, especially in the brain diseases such as stroke and tumor. The application of ¹⁷O temperature mapping could provide an alternative situation, where susceptibility issues hinder the standard phase-measurement approach and enable the investigation of highly temperature dependent pathophysiological changes in the brain. Finally, the imaging sensitivity could be improved by many times if one apply an infusion of high ¹⁷O-enriched H₂¹⁷O which is safe.

Acknowledgements

NIH grants: R01 MH111413, R01 NS118330, R01 CA240953, U01 EB026978, S10OD028712 and P41 EB015894.