Ty O Easley1, Federico S Pineda1, Byol S Kim2, Rina S Foygel-Barber2, Chengyue Wu3, Thomas E Yankeelov4, Xiaobing S Fan1, Deepa S Sheth1, David S Schacht1, Hiro S Abe1, and Gregory S Karczmar1
1Radiology, University of Chicago, Chicago, IL, United States, 2Statistics, University of Chicago, Chicago, IL, United States, 3Biomedical Engineering, University of Texas at Austin, Austin, TX, United States, 4Institute for Computational and Engineering Sciences, Diagnostic Medicine, and Oncology, University of Texas at Austin, Austin, TX, United States
Synopsis
Ultrafast DCE-MRI detects sparse
enhancement during the early phase of contrast media uptake [8]. This
facilitates reconstruction of arteries and lesions using partial k-space data
to obtain even higher temporal resolution. In addition, new approaches to
tracking blood vessels in breast [10] identify arteries feeding
suspicious lesions and possibly also draining veins. As a result, tumor blood flow can be
accurately measured from propagation of the contrast media bolus along arteries
that supply tumors. This approach avoids
assumptions and artifacts that are inherent in pharmacokinetic analysis, and
facilitates measurement of important biomarkers, e.g. capillary permeability
and interstitial pressure.
INTRODUCTION
Measurements of tumor blood flow
based on pharmacokinetic analysis of contrast media kinetics depends on compartmental
models for tissue [1, 2]. These models do
not match the true breast tissue structure [3, 4], and this can cause significant errors in data
analysis. Here we identify arteries
feeding the lesion, track the speed of propagation of the bolus down each
artery, and measure arterial diameter to measure blood flow to the tumor. We use ultrafast data acquisition [3, 5-8] and partial k-space
reconstruction to obtain bilateral images of enhancing arteries and lesions
with high temporal resolution. This
method, combined with pharmacokinetic measurements [9] of Ktrans (volume
transfer coefficient) may allow
measurement of tumor permeability. In
addition, tumor interstitial pressure can be evaluated from the speed of bolus
propagation [10, 11]. These are two
important biomarkers that cannot be measured by conventional pharmacokinetic
measurements alone. This method was
tested using simulations based on breast models constructed from actual ultrafast
DCE-MRI (dynamic contrast enhanced MRI) data.THEORY AND METHODS
Bilateral breast images ultrafast (3.5
seconds/image) DCE images (contrast agent dose = 0.1 mM/kg) were acquired on a
Philips 3T scanner. These images were used to produce a breast model that
simulated the Parker AIF [12] propagating
along internal mammary arteries and the single artery feeding the cancer with 50
msec resolution. A full k-space dataset was acquired every 3.5
seconds. Each ¼ second section of k-space,
calculated from the 50 msec simulation, was used to estimate a bilateral image by
optimizing over a regularized temporal smoothing constraint via gradient descent (analogous to [13]).RESULTS
Image reconstructions using 1/14th
of a complete k-space dataset closely matched the ‘gold standard’ fully sampled
images simulated with 50 msec temporal resolution. 98% of enhancing voxels had
error (difference between gradient descent reconstructions and ‘gold standard’
images) of 2% or less (see Figure 1). Gradient subtraction
images (each image subtracted from the subsequent image) reconstructed from 1/14th of k-space (Figure 2), clearly show the single artery feeding the breast cancer,
as well as other arteries and the suspicious lesion.
Figure 3 shows the time of arrival of the
contrast media bolus at each point along each artery (blue = early arrival; red
= later arrival). Based on these
measurements the speed of propagation of the AIF, i.e. the speed of blood flow,
is ~2.2 cm/sec near the lesion. The
diameter of the artery feeding the lesion is 1.5 mm and plasma flow (0.55 *
blood flow) to the lesion is approximately 1.3 mls per minute. The volume of the lesion, based on
enhancement in later post contrast images is ~1.57 grams, so lesion perfusion
is ~0.83 mls of plasma per gram of tissue per minute. This is equivalent to an average lesion Ktrans of 0.83 min-1, assuming
high permeability, i.e. extraction fraction of 1.0. CONCLUSION
Ultrafast
DCE-MRI facilitates reconstruction of images from partial k-space data is effective because the difference images during
initial enhancement are very sparse [7,
8, 14]. The gradient descent [13] method provided accurate
reconstructions at ¼ sec temporal resolution (Figures 2 and 3). This was critical for accurate measurement of
the speed of blood flow in arteries. Although a full contrast agent dose
was used here, a much lower dose could be used for more accurate detection of
the bolus [15].
The speed of blood flow and arterial
lumen diameter measured from high temporal resolution images were used to
determine the rate of blood flow. With
real data, capillary permeability would be determined by comparing Ktrans
determined from arterial blood flow to the average Ktrans from
pharmacokinetic modeling. In the data
shown here the single artery feeding the tumor is easily identified. In more complicated cases, the vessel
tracking method developed by Wu et al
[10] can identify arteries
feeding breast cancers, and possibly draining veins. This approach to measurement of tumor blood
flow avoids errors due to use of compartmental models [2,
9] that are
not a good match for tissue microstructure [2,
16], and errors
due to the assumption that all compartments are well mixed [4]. In addition, tumor interstitial pressure can
be measured based on speed of blood flow in arteries feeding tumors versus
arteries feeding normal tissue [11,
17]. These biomarkers are important indicators of
cancer malignancy and response to therapy.
Finally, this approach may allow measurement of tumor blood flow with
low doses of contrast agent, since bolus propagation can be measured more
accurately and with high SNR at very low contrast media concentrations [15].Acknowledgements
This research is
supported by National Institutes of Health (R01 CA172801-01, R01 CA218700-01,
and 5U01 CA142565-09, and U01CA174706), and the Segal Foundation.References
1. Yankeelov, T.E. and J.C. Gore, Dynamic Contrast Enhanced Magnetic Resonance
Imaging in Oncology: Theory, Data Acquisition, Analysis, and Examples. Curr
Med Imaging Rev, 2009. 3(2): p.
91-107.
2. Sourbron, S.P. and D.L. Buckley, Classic models for dynamic contrast-enhanced
MRI. NMR Biomed, 2013. 26(8): p.
1004-27.
3. Jansen, S.A., et al., Ductal carcinoma in situ: X-ray fluorescence
microscopy and dynamic contrast-enhanced MR imaging reveals gadolinium uptake
within neoplastic mammary ducts in a murine model. Radiology, 2009. 253(2): p. 399-406.
4. Woodall, R.T., et al., The effects of intravoxel contrast agent
diffusion on the analysis of DCE-MRI data in realistic tissue domains. Magn
Reson Med, 2018. 80(1): p. 330-340.
5. Mann, R.M., et al., A novel approach to contrast-enhanced breast
magnetic resonance imaging for screening: high-resolution ultrafast dynamic
imaging. Invest Radiol, 2014. 49(9):
p. 579-85.
6. Pinker, K., et al., A combined high temporal and high spatial
resolution 3 Tesla MR imaging protocol for the assessment of breast lesions:
initial results. Invest Radiol, 2009. 44(9):
p. 553-8.
7. Abe, H., et al., Kinetic Analysis of Benign and Malignant
Breast Lesions With Ultrafast Dynamic Contrast-Enhanced MRI: Comparison With
Standard Kinetic Assessment. AJR Am J Roentgenol, 2016. 207(5): p. 1159-1166.
8. Pineda, F.D., et al., Ultrafast Bilateral DCE-MRI of the Breast
with Conventional Fourier Sampling: Preliminary Evaluation of Semi-quantitative
Analysis. Acad Radiol, 2016. 23(9):
p. 1137-44.
9. Tofts, P.S. and A.G. Kermode, Measurement of the blood-brain barrier
permeability and leakage space using dynamic MR imaging. 1. Fundamental
concepts. Magn Reson Med, 1991. 17(2):
p. 357-67.
10. Wu, C., et
al., Quantitative analysis of vascular
properties derived from ultrafast DCE-MRI to discriminate malignant and benign
breast tumors. Magn Reson Med, 2018.
11. Wu C, P.F.,
Hormuth DA, Karczmar G, Yankeelov TE. . Computational
fluid dynamics characterizing flow through breast tumors using patient-specific
geometries. . in Annual meeting of BMES. 2018.
12. Parker, G.J.,
et al., Experimentally-derived functional
form for a population-averaged high-temporal-resolution arterial input function
for dynamic contrast-enhanced MRI. Magn Reson Med, 2006. 56(5): p. 993-1000.
13. Ryan Wen Liu,
L.S., Simon Chun Ho Yu, Naixue Xiong,Defeng Wang, Reconstruction of Undersampled Big Dynamic MRI Data Using Non-Convex
Low-Rank and Sparsity Constraints
Sensors, 2017. 17(3): p. 509.
14. Jansen, S.A.,
et al., Characterizing early contrast
uptake of ductal carcinoma in situ with high temporal resolution dynamic
contrast-enhanced MRI of the breast: a pilot study. Phys Med Biol, 2010. 55(19): p. N473-85.
15. He, D., et
al., Feasibility of Dynamic
Contrast-Enhanced Magnetic Resonance Imaging Using Low-Dose Gadolinium:
Comparative Performance With Standard Dose in Prostate Cancer Diagnosis.
Invest Radiol, 2018. 53(10): p.
609-615.
16. Sourbron,
S.P. and D.L. Buckley, On the scope and
interpretation of the Tofts models for DCE-MRI. Magn Reson Med, 2011. 66(3): p. 735-45.
17. Liu,
L.J., et al., Estimation of Tumor
Interstitial Fluid Pressure (TIFP) Noninvasively. PLoS One, 2016. 11(7): p. e0140892.