2125

Accurate detection of calcified carotid plaques on quantitative susceptibility mapping – A CT validation study
Thanh D Nguyen1, Jingwen Du2, Yan Wen1, Ajay Gupta1, Pascal Spincemaille1, Qi Yang2, and Yi Wang1
1Weill Cornell Medicine, New York, NY, United States, 2Xuanwu Hospital of Capital Medical University, Beijing, China

Synopsis

In this study, we used CT angiography (CTA) as a reference standard to demonstrate that carotid quantitative susceptibility mapping (QSM) provides accurate detection of calcified atherosclerotic plaques in the carotid arteries. Calcification is strongly diamagnetic and appears uniquely hypointense on carotid QSM. Inclusion of QSM in carotid MRI would aid characterization of carotid plaque.

INTRODUCTION

Calcification is a common feature of carotid atherosclerotic plaques, which indicates plaque stability and may reduce stroke risk in half (1). Heavily calcified plaques are associated with less favorable outcomes of carotid revascularization. Therefore, accurate assessment of plaque calcification is critically important to triaging patients for medication vs. revascularization intervention. CT angiography (CTA), currently the gold standard for in vivo calcification detection, requires harmful radiation and a large dose of iodine contrast agent but does not provide characterization of other important intraplaque components including hemorrhage of high risk. Multi-contrast MRI (mcMRI), useful for characterizing intraplaque components without CTA limitations (2), allows detection of calcification based on hypointensity on T1w/T2w images. However, accurate interpretation of calcification on mcMRI is difficult, as hemosiderin-rich intraplaque hemorrhage (IPH), perivascular fat, and blood vessels also appear hypointense on the black blood fat-suppressed mcMRI images. This mcMRI ambiguity can be resolved on quantitative susceptibility mapping (QSM), because calcification is a uniquely strong diamagnetic source in the body (3-5) and is easily distinguishable from paramagnetic sources such as hemorrhage and fat. The objective of this study was to evaluate the improvement in diagnostic accuracy of calcification detection in carotid plaques by adding QSM to mcMRI, using CTA as the reference standard.

METHODS

Carotid plaque mcMRI. The clinical imaging protocol was based on ASNR Vessel Wall Imaging Study Group recommendations (6) and consisted of 3D TOF, 2D black blood fat-suppressed T1w/T2w TSE, and 3D magnetization-prepared rapid gradient echo (MPRAGE) sequences with spatial resolution of 0.6x0.6x2 mm3 and 6 cm longitudinal coverage of the carotid bifurcation.
Carotid QSM. A multi-echo 3D GRE sequence was optimized for carotid QSM to match the spatial resolution and coverage of mcMRI with 5 min scan time. Four echo times were acquired per TR with 4.7 ms in-phase echo spacing to allow optimal field estimation in the presence of fat at 3T (7). A nonlinear preconditioned total field inversion algorithm was developed to calculate the susceptibility map by minimizing the following cost function:
$$x^*=arg min_x{\parallel}w(e^{-i\alpha f}-e^{-id*Px}){\parallel}^2_2+{\lambda\parallel}M_G{\triangledown}Px{\parallel}_1+{\lambda_{A}\parallel}M_{A}P(x-\overline{x}^{A}){\parallel}^2_2$$
Here the first two terms are the data fidelity term and structure consistency regularization term, where $$$w$$$ is the SNR weighting, $$$d$$$ is the dipole kernel, $$$M_G$$$ is the edge mask, and $$$\triangledown$$$ is the gradient operator. The third term enforces susceptibility homogeneity of the well-mixed arterial blood (8), $$$M_{A}$$$ is the arterial mask (obtained using a semi-automated region-growing algorithm), and $$$\overline{x}^{A}$$$ is the mean susceptibility within the mask. $$$\alpha$$$ is a scalar that scales down $$$f$$$ to avoid phase wraps. The preconditioner $$$P$$$ is used to accelerate convergence (9). The final susceptibility map is $$$Px^*/\alpha$$$.
Imaging study. Ten patients (mean age 66 years ± 5) had both MRI and CTA scans with median follow-up interval of 3 days. A neuroradiologist with 13 years of carotid imaging experience independently reviewed CTA and mcMRI (without and with QSM image) to identify plaque calcification on a per vessel basis. Calcification was detected as region with high attenuation on CTA (Hounsfield unit>100), hypointense signal (compared to the adjacent muscle) on mcMRI, and strongly negative susceptibility (<-0.5 ppm) on QSM.

RESULTS

Figure 1 shows an example of concordant depiction of a heavily calcified plaque at the carotid bifurcation on mcMRI, QSM and CTA images. Compared to mcMRI, the unambiguous visual contrast between calcification and the surrounding tissues on QSM allows easy identification and improves diagnostic confidence. This benefit is highlighted in Figs. 2 and 3, which show examples of small calcification nodules with dark signal that could not be prospectively identified by the experienced reader on mcMRI but well captured by QSM (approximately -2 ppm susceptibility) in excellent agreement with CTA. Out of 17 vessels with calcified plaques detected by CTA, mcMRI only captured 11 calcifications (64.7% sensitivity). However, with the addition of QSM, all calcifications were identified (100% sensitivity). Both mcMRI and QSM correctly identified the 3 non-calcified vessels (100% specificity). Additionally, Figure 4 demonstrates the versatility of QSM for detecting paramagnetic intraplaque hemorrhage consistent with low attenuation (Hounsfield unit <60) on CTA image.

DISCUSSION

Our preliminary results suggest that QSM substantially improves the diagnostic accuracy for detecting calcified carotid plaques. Our QSM validation study with CTA is ongoing, and the patient cohort size will be significantly increased in the future. QSM has the potential to become an instrumental part of noninvasive and quantitative carotid MRI for plaque characterization. QSM is known to provide sensitive detection of strong magnetic materials in tissue, including highly paramagnetic iron in hemorrhage and highly diamagnetic calcification deposits. Accordingly, QSM can reliably resolve the ambiguity of T1w hypointensity on traditional mcMRI, which can be the result of IPH and/or calcification. Future clinical studies with large patient sample sizes are warranted to establish the value of including QSM in carotid MRI protocol, which would enable confident triaging carotid patients who are at high risk and would benefit from surgical interventions to prevent stroke, from those with lower risk who would derive a similar risk reduction benefit from noninvasive medication treatment.

Acknowledgements

No acknowledgement found.

References

1. Baradaran H, Al-Dasuqi K, Knight-Greenfield A, Giambrone A, Delgado D, Ebani EJ, Kamel H, Gupta A. Association between carotid plaque features on CTA and cerebrovascular ischemia: A systematic review and meta-analysis. AJNR Am J Neuroradiol. 2017;38(12):2321-6.
2. Cai JM, Hatsukami TS, Ferguson MS, Small R, Polissar NL, Yuan C. Classification of human carotid atherosclerotic lesions with in vivo multicontrast magnetic resonance imaging. Circulation 2002;106(11):1368-73.
3. Ikebe Y, Ishimaru H, Imai H, Abe K, Izumo T, Morofuji Y, Ideguchi R, Morikawa M, Uetani M. Quantitative susceptibility mapping for carotid atherosclerotic plaques: A pilot study. Magn Reson Med Sci 2019 May 31.
4. Wang C, Liu S, Chen Y, Buch S, Fan Z, Haacke EM, Yang Q. Intraplaque hemorrhage and calcification detection with quantitative susceptibility mapping. ISMRM 2018;491.
5. Chen W, Zhu W, Kovanlikaya I, Kovanlikaya A, Liu T, Wang S, Salustri C, Wang Y. Intracranial calcifications and hemorrhages: characterization with quantitative susceptibility mapping. Radiology 2014;270(2):496-505.
6. Saba L, Yuan C, Hatsukami TS, Balu N, Qiao Y, DeMarco JK, Saam T, Moody AR, Li D, Matouk CC, Johnson MH, Jager HR, Mossa-Basha M, Kooi ME, Fan Z, Saloner D, Wintermark M, Mikulis DJ, Wasserman BA, Vessel Wall Imaging Study Group of the American Society of N. Carotid Artery Wall Imaging: Perspective and Guidelines from the ASNR Vessel Wall Imaging Study Group and Expert Consensus Recommendations of the American Society of Neuroradiology. AJNR Am J Neuroradiol 2018;39(2):E9-E31.
7. Pineda AR, Reeder SB, Wen Z, Pelc Cramér-Rao bounds for three-point decomposition of water and fat. Magn Reson Med 2005;54(3):625-35.
8. Liu Z, Spincemaille P, Yao Y, Zhang Y, Wang Y. MEDI+0: Morphology enabled dipole inversion with automatic uniform cerebrospinal fluid zero reference for quantitative susceptibility mapping. Magn Reson Med 2018;79(5):2795-803.
9. Liu Z, Kee Y, Zhou D, Wang Y, Spincemaille P. Preconditioned total field inversion (TFI) method for quantitative susceptibility mapping. Magn Reson Med 2017;78(1):303-15.

Figures

Figure 1. Example of concordant detection of a heavily calcified plaque at the carotid bifurcation with hypointense appearance on mcMRI and strongly negative susceptibility (-1 to -2 ppm) on QSM in good agreement with high attenuation on post-contrast CTA image. The unambiguous tissue contrast between vascular calcification and blood as well as surrounding tissues on QSM leads to higher diagnostic confidence compared to mcMRI.

Figure 2. In this example, a calcified nodule missed on mcMRI due to poor lesion contrast was detected with high confidence by the reader on QSM (-2 ppm susceptibility), which shows excellent agreement with CTA.

Figure 3. An example of a small calcified plaque in the internal carotid artery that was not detected by mcMRI but well visualized on QSM (-1.8 ppm susceptibility) and confirmed by CTA.

Figure 4. Example of a highly paramagnetic lesion (1.6 ppm) on QSM consistent with a hemosiderin-rich IPH. CTA image shows low attenuation suggesting a soft plaque with hemorrhage (Hounsfield unit <60). This lesion has ambiguous contrast and could not be prospectively identified on mcMRI images.

Proc. Intl. Soc. Mag. Reson. Med. 28 (2020)
2125