Wen Cao1, Yulin Chang1, Suliman Barhoum1, Zachary B Rodgers1, Michael C Langham1, Erin K Englund1, and Felix W Wehrli1
1Department of Radiology, University of Pennsylvania, Philadelphia, PA, United States
Synopsis
The
OxFlow imaging approach allows simultaneous quantification of whole-brain
venous oxygen saturation and total cerebral blood flow for the cerebral
metabolic rate of oxygen. However, the current Cartesian rendition of the sequence is not ideal as the achievable temporal resolution is limited and
needs to be chosen upfront. Here, we designed a golden-angle radial (GAR)
encoding sequence that yields an effective temporal resolution of 1.29s and evaluated
it in six subjects who underwent a paradigm of repeated breath-holds. Good agreement exists between the two methods but GAR provided superior SNR and better
delineation of the temporal dynamics during the stimulus.Introduction
The recently introduced OxFlow sequence for rapid
measurement of the cerebral metabolic rate of oxygen (CMRO2) is
based on Cartesian k-space coverage1. OxFlow simultaneously yields superior sagittal sinus (SSS) flow via
phase-contrast and intravascular venous oxygen saturation (SvO2) via magnetic susceptometry2. Temporal
resolution is enhanced by employing a BRISK view-sharing scheme, yielding a
maximum temporal resolution of 3 seconds1,3. However, this approach
is not ideal as it provides limited flexibility in that the temporal resolution
has to be chosen a priori. In contrast, a golden-angle radial sampling strategy
allows for a variety of reconstruction options tailored to the specific
application’s needs.
Purpose:
To design and
validate a method for rapid quantification of whole-brain CMRO2 via simultaneous measurement of SvO2 and SSS blood flow using a velocity-encoded, multi-echo, golden-angle radial imaging sequence.
Methods
A radial OxFlow (rOxFlow)
sequence (Figure 1) was designed in
SequenceTree with sequence parameters closely matching the standard Cartesian
version (cOxFlow)1. Sequence parameters: TR/TE=19.2 ms/6.752 ms , flip
angle=15°, slice thickness=4 mm. Acquisition matrix size was 208×208 for cOxFlow
(80% phase-encoding coverage, 176×176 FOV) and 240×240 for radial (15790 radial
lines). Temporal resolution was 3s for cOxFlow
and 1.29s for rOxFlow. Radial images were
reconstructed every 34 views. The sequences were optimized for a protocol
involving continuous scanning for 10 minutes.
All experiments were conducted at 1.5T (Siemens AVANTO).
Results from rOxFlow were compared to cOxFlow in six volunteers during a
volitional apnea paradigm. All subjects were instructed to perform five
successive breath-holds, each 30s in duration, separated by 90s. For both
rOxFlow and cOxFlow, SvO2 was calculated from 1st and 3rd
echoes and phase-contrast velocity was computed from the first echo collected
with different first gradient moments (Figure
1). Total CBF (tCBF) was estimated by upscaling SSS velocity by a factor
acquired from a preceding calibration sequence1. Bland-Altman analysis4
was conducted to investigate inter-sequence bias. Means of baseline and peak SvO2
and velocity were averaged over 5 breath-hold and rest periods, respectively,
from both Cartesian and radial data, and compared via paired t-tests.
Results
Figure 2 shows a sagittal scout angiogram
and magnitude axial image indicating measurement locations. Sample velocity and
phase cOxflow and rOxFlow images and response time courses are displayed in
Figure 3 highlighting the similarity of the two sequences in both the parametric images
and over the entire time course of the breath-hold
response. A Bland-Altman plot comparing cOxFlow to rOxFlow in one
subject during the time-course of the experiment is given in
Figure 4. The data suggests minimal
bias given that >95% of the differences were within the ±1.96 SD limit.
Table 1 lists SvO
2 and tCBF
data in all six subjects as well as means and standard deviations. There was no significant difference
between the two methods from the paired
t-tests involving all six subjects for
any of the parameters measured (p>0.20 in all cases).
Discussion & Conclusion
In
general, the results from the two sequences were found to be in good agreement
with each other. While the data from the six subjects did not suggest a bias
between sequences, comparisons involving a larger number of subjects would be
required to corroborate this result. In
addition to providing superior effective temporal resolution, rOxFlow was found
to have better SNR than its Cartesian counterpart.
Acknowledgements
This work was supported by NIH R01-HL122754References
[1] Rodgers, et al. JCBFM 2013; [2] Fernández-Seara, et al. MRM 2006; [3] Rodgers, et al. JCBFM 2015; [4] Bland & Altman, Lancet 1986.