To investigate the sensitivity gain of ¹⁷O MRS imaging at ultrahigh field of the world's first 10.5T whole-body human MRI scanner as compared to 7T.In vivo ¹⁷O MRS imaging provides a valuable tool for quantitatively imaging the cerebral rate of oxygen metabolism, cerebral blood flow and oxygen extraction fraction from a brief inhalation of ¹⁷O-isotope labeled oxygen gas. Due to the unique NMR properties of short ¹⁷O quadrupolar relaxation time and field independence, the NMR sensitivity of the ¹⁷O water signal was found to increase quadratically with magnetic field strength (B₀) in the phantom test and animal brain ^{1, 2}. In this study, we conducted a pilot test to examine the ¹⁷O sensitivity using ahuman head size water phantom in the world's first 10.5T whole-body human scanner at CMRR and compared it with that at 7T. Chemical shift imaging (CSI) based on the Fourier-series window technique was performed on two human MR-scanners (Siemens Healthcare, Erlangen, Germany) at the field strengths of 7T and 10.5T, respectively. A 2 liter spherical glass phantom (~16 cm diameter similar to human head size) containing natural abundance ¹⁷O saline solution (with 50 mM NaCl and 25 mM inorganic phosphate) was scanned at fully relaxed conditions (TR 100ms, TR>>5*T₁ of ¹⁷O water) with the following sequence parameters: 0.5ms TE, 0.5ms hard pulse width, 18x18x15cm³ FOV, 9x9x7 matrix size, 23ms acquisition time and 30kHz spectral width. Signal was acquired using a home-built quadrature surface-coil of two loops (diameter 16cm) tuned and matched to the resonance frequencies of 40.292MHz and 60.611 MHz at 7T and 10.5, respectively. The reference RF voltage was optimized to obtain maximal signals in the region of interest at two fields. Two low noise high-gain pre-amplifiers and T/R-switches with similar circuitry but different operation frequencies from the same manufacturer (Stark Contrast, Erlangen, Germany) were used at both field-strengths. Signal-to-noise ratio (SNR) was measured by dividing the peak intensity by the peak-to-peak noise level. In addition, we also measured and compared the extra magnet room noise by comparing the difference of the noise levels measured under two conditions: T/R switch was connected with a 50 ohm terminator versus the RF quadrature coil as a room noise pickup coil. The phantom shape was clearly resembled in the transversal slices of 3D ¹⁷O water CSI data as shown in Fig. 1, and the phantom position was confirmed to be in the same position in both scanners with the iso-center as the center CSI voxel. Highest signal intensity occurred in the central voxels and tilted towards to the quadrature coil (Fig 1). For the voxels, the averaged SNR was 545 at 10.5T and 383 at 7T, resulting in an SNR ratio of 1.42 between the two fields or a SNR gain of 42% at 10.5T. This gain was less than the estimated SNR ratio of near 2.0 based on the field dependence relationship ^1-2 of SNR µ B₀^7/4. Interestingly, we also observed a higher level of approximately 40% in the 10.5T magnet room noise based on the comparison of noise measurements between 50 ohm terminator and RF coil. After considering and correcting the extra noise observed in the 10.5T magnet room, it resulted in an SNR ratio of gain of near 2 that was close to the prediction. From this first study and preliminary results, we found a significant SNR gain of ¹⁷O MRS imaging at 10.5T as compared to 7T. We also found a relatively large magnet room noise for the new 10.5T magnet room and it was under investigation nor for fixing. Nevertheless, after the correction of the extra noise, the net SNR gain was consistent with the prediction. In addition, we also found a larger RF power is needed at 10.5T for reaching the same RF pulse flip angle ^3,4,5,6.The preliminary results suggest a possibility to double sensitivity for in vivo ¹⁷O MRS imaging at 10.5T as compared to 7T. This sensitivity gain will advance the technology and utility of in vivo ¹⁷O MRS imaging for studying brain metabolism and perfusion.