INTRODUCTION

T₂-based MRI oximetry (T2O) measures T₂ of venous blood water protons, followed by a conversion to SvO₂ via a calibration curve ^{[1, 2]}. To quantify metabolic rate of O₂ (MRO₂), an additional measurement of blood flow velocity is required. T₂-relaxation-under-spin-taggging (TRUST) is a well-established T2O method for quantifying whole-brain SvO₂ by measuring T₂ in the superior sagittal sinus of the brain ^[2]. However, TRUST is sensitive to physiologic motion because two scans, a “control” and a “label,” are required to remove stationary background tissue and isolate intravascular blood signal. Moreover, it is important to minimize the TE in TRUST because venous blood and tissue accrue phase at different rates. This can contribute to the persistence of background tissue despite subtraction between “control” and “label” images. TRUST also requires a separate measurement of blood flow velocity to quantify MRO₂ ^[3]. Here, a T2O method is reported that interleaves phase-contrast before background-suppressed T₂-prepared echoplanar imaging (EPI). Background-suppression reduces signal from stationary tissue while inflowing venous blood will provide signal from which T₂ (and SvO₂) can be determined. Importantly, the proposed method interleaves phase-contrast MRI to simultaneously estimate SvO₂ and blood flow velocity, which are both required to quantify MRO₂.

METHODS

The pulse sequence begins by resetting the magnetization with a global saturation, followed by radial phase-contrast MRI and background-suppressed T₂-prepared EPI modules. Background-suppression significantly reduces the partial volume effect by suppressing static tissue signal and takes advantage of signal enhancement from inflowing blood with a combination of slice-selective and non-selective adiabatic inversion pulses ^{[4, 5]}. T₂-preparation consists of an MLEV-4 scheme to provide T₂-weighting with five TE's (0/48/96/144/192ms) ^[6]. In each of the five TE's, 55 interleaved velocity-encoded radial views are acquired. A golden-angle radial acquisition was used because each view passes through the center of k-space, which increases robustness to motion. The total acquisition time is approximately 18 seconds. The pulse sequence timeline is shown in Figure 1.

The velocity-encoding module comprises the following parameters: VENC=60cm/s, golden-angle increment=111.3°, TR/TE=9/5.8ms, 55 views per interleave, and voxel size=0.67x0.67x5 mm³. Background-suppression is achieved with the following RF pulses: a slice-selective 90° saturation pulse, two slice-selective adiabatic inversion pulses, and two non-selective adiabatic inversion pulses. The EPI module comprises the following parameters: 5/8^th partial Fourier, flip angle of 90°, voxel size=3.2x3.2x5 mm³, and includes a calibration scan for phase correction. Radial reconstruction is performed by combining the sets of 55 radial views from the five interleaves, for a total of 275 radial views. The pulse sequence results in five images of varying T₂-weighting and one velocity map. Pulse sequence parameters are summarized in Table 1.

The pulse sequence was tested in two healthy subjects, ages 24.7 and 23.9 years. After informed consent was obtained, the subjects were imaged at 3 T (Siemens Prisma) using a 20-channel head and neck coil, at a slice in the neck transecting the internal carotid and vertebral arteries and internal jugular vein. The pulse sequence was repeated six times. Data are reported as an average of the six repetitions.

Cerebral MRO₂ (CMRO₂) was quantified on the basis of the conservation of mass equation (Fick’s Principle): $$$CMRO_2=C_{RBC}\cdot Hct\cdot tCBF\cdot (SaO_2-SvO_2)$$$ ^[7]. C_RBC is 22.02 μmol O₂/mL RBC ^[8], the O₂ carrying capacity of a red blood cell (RBC). Hct represents hematocrit and is determined by finger prick (Hb 201+, Hemocue, Sweden). tCBF is total blood flow rate normalized by brain mass. Velocity-encoded images were generated by phase difference, from which tCBF was determined by integrating blood flow velocity over vessel area (internal carotid and vertebral arteries). Brain mass was computed by estimating brain volume with a T₁-weighted MPRAGE sequence ^[9] and applying a density of 1.05g/mL. SaO₂ and SvO₂ are arterial and venous oxygen saturation, respectively. SaO₂ was assumed to be 98%. T₂ of the blood water protons in the internal jugular vein was estimated and converted to SvO₂ via a calibration curve ^[10]. Figure 2 illustrates the workflow.

RESULTS

Table 2 lists the physiologic parameters obtained in the two test subjects, yielding CMRO₂ of 127±14 and 132±12 μmol/min/100g obtained via Fick’s Principle. Figure 3 shows a representative background-suppressed magnitude image of the internal jugular vein at TE=0ms, a panel of the internal jugular vein at increasing T₂-weighting, and an inset of a phase-contrast velocity image of the feeding arteries, and a graph of the T₂-relaxation curve.

DISCUSSION & CONCLUSION

The pulse sequence quantified CMRO₂ values that are comparable to those reported in literature (132±20^[3], 127±7^[11] μmol/min/100g). T₂-based oximetry at the neck allows for measuring total blood flow velocity from the internal carotid and vertebral arteries and SvO₂ from the internal jugular vein at a single location. Acquisition at the neck also enables direct measurement of total cerebral blood flow, whereas methods at the superior sagittal sinus often use flow-upscaling to estimate total cerebral blood flow ^[12]. In conclusion, this rapid pulse sequence simultaneously acquires the key physiological parameters (SvO­₂ and blood flow velocity) to quantify whole-organ MRO₂ in 18 seconds scan time. Further work will systematically validate and compare this pulse sequence to other oximetry methods, including TRUST and susceptometry-based oximetry. This method should lend itself for deriving MRO₂ in abdominal organs, including liver and kidney.

Acknowledgements

This work was supported by NIH Grant T32 EB020087.