Three-dimensional dynamic contrast-enhanced MRI for the accurate, extensive quantification of microvascular permeability in atherosclerotic plaques

Abstract

Atherosclerotic plaques that cause stroke and myocardial infarction are characterized by increased microvascular permeability and inflammation. Dynamic contrast-enhanced MRI (DCE-MRI) has been proposed as a method to quantify vessel wall microvascular permeability in vivo. Until now, most DCE-MRI studies of atherosclerosis have been limited to two-dimensional (2D) multi-slice imaging. Although providing the high spatial resolution required to image the arterial vessel wall, these approaches do not allow the quantification of plaque permeability with extensive anatomical coverage, an essential feature when imaging heterogeneous diseases, such as atherosclerosis. To our knowledge, we present the first systematic evaluation of three-dimensional (3D), high-resolution, DCE-MRI for the extensive quantification of plaque permeability along an entire vascular bed, with validation in atherosclerotic rabbits. We compare two acquisitions: 3D turbo field echo (TFE) with motion-sensitized-driven equilibrium (MSDE) preparation and 3D turbo spin echo (TSE). We find 3D TFE DCE-MRI to be superior to 3D TSE DCE-MRI in terms of temporal stability metrics. Both sequences show good intra- and inter-observer reliability, and significant correlation with ex vivo permeability measurements by Evans Blue near-infrared fluorescence (NIRF). In addition, we explore the feasibility of using compressed sensing to accelerate 3D DCE-MRI of atherosclerosis, to improve its temporal resolution and therefore the accuracy of permeability quantification. Using retrospective under-sampling and reconstructions, we show that compressed sensing alone may allow the acceleration of 3D DCE-MRI by up to four-fold. We anticipate that the development of high-spatial-resolution 3D DCE-MRI with prospective compressed sensing acceleration may allow for the more accurate and extensive quantification of atherosclerotic plaque permeability along an entire vascular bed. We foresee that this approach may allow for the comprehensive and accurate evaluation of plaque permeability in patients, and may be a useful tool to assess the therapeutic response to approved and novel drugs for cardiovascular disease.

Keywords: atherosclerosis; compressed sensing; dynamic contrast enhanced (DCE) magnetic resonance imaging (MRI); permeability; rabbits.

Figure 1

A and B, curved maximum intensity projection (MPR) of the abdominal aorta of one rabbit imaged using 3D TFE and TSE DCE-MRI, respectively. Orange lines indicate the left renal artery and the iliac bifurcation. C and D, one representative axial slice of the abdominal aorta of the same rabbit shown in A and B, imaged using 3D TFE and TSE DCE-MRI, respectively. E, one representative axial slice of the same rabbit, imaged using 2D TSE DIR. Red arrow indicates the abdominal aorta.F, time-intensity curves for an aortic ROI in the same axial slice showed in C-E for all 3 sequences. Blue line, 3D TFE DCE-MRI. Red line, 3D TSE DCE-MRI. Green line, 2D DIR TSE DCE-MRI. Gray bars indicate the data points from which temporal SNR and CNR were calculated. X axis, time (min). Y axis, signal intensity (a.u.).

Figure 2

Average temporal signal-to-noise ratio (tSNR, panel A) and temporal contrast-to-noise ratio (tCNR, panel B) for 3D TFE (blue bar), 3D TSE (red bar) and 2D DIR (green bar) DCE-MRI. Black lines represent the standard deviation of the mean. Black stars indicate significant difference (p<0.05).

Figure 3

Difference between AUC (panel A) and K^trans (panel B) for atherosclerotic and control animals. For each sequence darker colors represent atherosclerotic animals, while lighter shades represent the control rabbits. Black lines, error bars (standard deviation of the mean). Asterisk indicates significant difference.

Figure 4

Difference between permeability in atherosclerotic (A) and control (B) animals. Hotter colors in AUC maps and EB images indicated higher permeability. Yellow lines indicate the left renal artery and the iliac bifurcation. The higher ex vivo permeability of the atherosclerotic rabbit (EB) is mirrored in the “hotter” AUC maps. On the contrary the control animal shows poor Evans Blue uptake and very low AUC maps.

Figure 5

Correlations between in vivo permeability by DCE-MRI and ex vivo permeability by Evans Blue near infra red fluorescence. Panel A, 3D TFE DCE-MRI (blue). Panel B, 3D TSE DCE-MRI (red). Panel C, 2D DIR TSE DCE-MRI (green). Black dashed line, regression line.

Figure 6

A, vessel wall/lumen contrast is shown to decrease with increasing acceleration rate, indicating blurring and poorer wall/lumen delineation with increasing under-sampling. Red dashed line, vessel wall/lumen contrast within 5% of the fully sampled acquisition. Gray dashed line, vessel wall/lumen contrast within 10% of the fully sampled acquisition. Vertical black lines, standard deviation. B, representative images of the dynamic series (averaged in the time dimension) reconstructed with compressed sensing after different degrees of under-sampling. Vessel wall is bright after contrast agent injection. Images show increased blurring at high accelerations. S, fully sampled scanner reconstruction, used as the reference image. Based on data shown in panel A, reconstruction from data under-sampled 8 fold show vessel wall/lumen delineation within 10% of the reference images (S).

Figure 7

A, normalized root mean square error between curves derived from under-sampled and fully sampled (scanner) reconstructions. N_RMSE is shown to increase with increasing acceleration. Gray dashed line, N_RMSE within 10% of scanner reconstruction. Vertical black lines, standard deviation. B, representative signal intensity curves from acceleration 2,4,6 and 15. S, scanner reconstruction.