Model-based fitting of in vivo 129Xe spectra in mice reveals five robust dissolved-phase peaks
Rohan S. Virgincar1, Scott H. Robertson2, Simone Degan3, Geoffry Schrank4, Mu He5, John Nouls4, and Bastiaan Driehuys4

1Biomedical Engineering, Duke University, Durham, NC, United States, 2Medical Physics Graduate Program, Duke University, Durham, NC, United States, 3Center for Molecular and Biomolecular Imaging, Duke University, Durham, NC, United States, 4Radiology, Duke University, Durham, NC, United States, 5Electrical and Computer Engineering, Duke University, Durham, NC, United States

Synopsis

Inhaled 129Xe exhibits chemical shifts which carry useful information about the underlying physiology. However, their resonant frequencies have been reported with a variability of 2-3 ppm likely attributable to using simplistic peak finding methods and inconsistent reference frequencies. In this work, we use robust non-linear curve fitting of the complex dissolved-phase spectrum in mice to identify resonances, and report shifts relative to an accurate reference frequency. At short 129Xe replenishment times curve fitting identified two peaks at 197.4±0.9 and 193.0±0.7 ppm, but as replenishment time was increased, five distinct peaks became apparent at 198.4±0.4, 195.5±0.4, 193.9±0.2, 191.3±0.2, and 190.7±0.3 ppm.

Purpose

129Xe readily interacts with biological media and exhibits chemical shifts that reflect the local tissue microenvironment. These shifts have been extensively studied in vitro, and to a limited extent in vivo in mice, rats and humans. Previously, in vivo 129Xe spectra have been processed by simply phasing the spectrum, identifying frequencies at which peaks occur, and calculating shifts relative to the readily available gas-phase resonance originating mostly from pulmonary airspaces. This practice has shortcomings. First, the rapid exchange of 129Xe between compartments may lead to broad and overlapping resonances that are often difficult to fully phase and isolate. Second, the in vivo 129Xe gas-phase resonance, used as a reference, also exhibits a shift1 due to bulk magnetic susceptibility (BMS) causing a bias in the reported dissolved-phase shifts. In this study, we identify individual resonances in the dissolved-phase 129Xe spectrum of the mouse lung by fitting the complex time domain signal to a series of exponentially decaying signals, and then report chemical shifts relative to an accurate gas-phase reference frequency adjusted for BMS shifts2. Additionally, we present spectra at increasingly long 129Xe replenishment times, and show that the spectra become progressively richer as 129Xe reaches downstream compartments within the thoracic cavity.

Methods

5 Balb/c mice were ventilated on an HP gas-compatible ventilator and underwent respiratory gated hyperpolarized 129Xe dissolved-phase spectroscopy in a quadrature 129Xe coil at 2T with the following parameters: frequency-selective 1.2-ms sinc pulse centered on the dissolved phase, TR/TE = 100-8000/1.2 ms, 2048 points, BW = 8.06 kHz, α = 90°, and 11 averages. Using an in-house MATLAB fitting tool, spectra were then interactively decomposed by least squares fitting the complex free induction decay signal to a sum of exponentially decaying signals. The software then reports the intensity, frequency, phase and decay rate of each individual peak in the spectrum.

Results and Discussion

At the shortest TR of 100 ms, 129Xe is limited to the gas-exchange region and exhibits two broad resonances at 197.4±0.9 and 193.0±0.7 ppm (n = 2). Increasing TR revealed additional peaks developing as 129Xe moved into downstream compartments. At the longest TR (8 s), five robust peaks were identified at 198.4±0.4, 195.5±0.4, 193.9±0.2, 191.3±0.2, and 190.7±0.3 ppm in all of the five mice. Based on the signal dynamics in our time series spectra, and comparing with the limited literature on in vivo 129Xe spectroscopy in mice, we can postulate the origins of these peaks. We suggest that the 198.4 ppm peak originates from blood1,3, while the 195.5 and 193.9 ppm peaks arise from aqueous media. They contain signal contributions from the pulmonary-capillary barrier tissue, plasma, and possibly also from the myocardium. The remaining upfield peaks (<192 ppm), only appear at long TRs, and likely arise from the slowly perfused and distal epicardial fat4. Using these improved fitting techniques and accurate frequency referencing should facilitate standardized comparison across sites, and aid in advancing the use of 129Xe spectroscopy and imaging to detect pathology.

Acknowledgements

R01-CA-142842, R01HL105643, and P41 EB015897

References

1. Wagshul ME et al. In vivo MR imaging and spectroscopy using hyperpolarized Xe-129. Magn. Reson. Med. 1996, 36(2):183-191

2. Virgincar RS et al. Establishing an accurate reference frequency for in vivo 129Xe spectroscopy. International Society for Magnetic Resonance in Medicine 24th Annual Meeting. 2016, Abstract ID 5540, submitted

3. Narazaki M et al. Origin of dissolved-phase hyperpolarized Xe-129 signal in the mouse chest based on experimental evidence from extensive magnetic resonance measurements. Mag. Reson. Med. Sci. 2011, 10(3):149-154

4. Swanson SD et al. Distribution and dynamics of laser-polarized 129Xe magnetization in vivo. Magn. Reson. Med. 1999, 42(6):1137-1145

Figures

Figure: 129Xe dissolved-phase spectroscopy at different 129Xe replenishment times. Fitting revealed 2 broad peaks at the shortest replenishment time (TR = 100 ms). The spectrum evolved with increasing TR, and at the longest TR of 8 s, 5 robust peaks we identified.



Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)
3983