High-resolution intravascular magnetic resonance quantification of atherosclerotic plaque at 3T

In vitro CMR studies were performed with a 3T loopless "receive-only" detector fabricated from a UT-85C (outer diameter, OD = 2.2 mm) semi-rigid copper coaxial cable (Micro-coax, Inc., Pottstown, PA) to fit within the approximately 8 - 15 mm lumens of human vessel specimens studied (Figure 1). The length of the detector's coaxial cable was tuned to a quarter-wavelength (λ/4 ≈ 40 cm). The central conductor was extended an additional 4 cm beyond the cable portion to form a whip that maximized signal at the whip-cable junction [33]. The outer surface of the cable was insulated with 0.02 mm polyester (dielectric constant, ε = 3) heat-shrink tubing (Advanced Polymer Inc, Salem, NH). To determine its sensitivity in the plane of the cable for angiography and/or tracking, the absolute SNR of the device in a 0.35% saline phantom with radio-frequency (RF) electrical properties comparable to those of tissue (conductivity, σ = 0.6 S/m, dielectric constant ε = 80) was determined by the numerical method-of-moments (FEKO EM analysis software, Stellenbosch, South Africa) from the ratio of the circularly polarized RF field excited by a 1A current, to the square root of the antenna resistance in the saline [33]. Since the SNR is cylindrically symmetric about the cable's long axis, the computation was performed on a single plane on one side of the detector, over the range 0.1 ≤ r ≤ 40 mm from the detector outer conductor, and 0 ≤ z ≤ 39 cm from the detector tip, as well as in axial planes about the detector junction.

In vivo experiments were performed with a biocompatible nitinol loopless imaging detector (Figure 2) with the same coaxial structure as the detector used for the in vitro experiments. It was formed by modifying an obsolete 0.76 mm OD Intercept 1.5T guidewire (MRI Interventions, formerly Surgi-Vision Inc, Memphis TN) for 3T use by retuning the whip to 40 mm, and extending the cable length to 3λ/4 (for a total cable length of 120 cm) to permit practical in vivo use in the magnet bore. Device heating during CMR was controlled at ≤1°C with decoupling circuitry described previously, and demonstrated in safety studies on a saline gel phantom [33]. The detector was tuned, matched, and interfaced to a 3T Philips Achieva MR scanner equipped with an X-ray C-arm (Philips Healthcare, Cleveland, OH) [33].

The 3T loopless CMR detector was first used to identify suspected plaques inside a fresh ~10 cm-long human iliac artery specimen, complete with bifurcation, harvested at autopsy within 3 hours of death. All studies of decedent specimens were approved by this institution's Office of Human Subject Research, Institutional Review Board. The vessel was secured to a plastic frame inside a 0.35% saline phantom (with σ = 0.6 S/m, ε = 80). The 3T detector was inserted into the lumen, and a spin-lattice relaxation time (T₁)-weighted angiographic scan performed in the coronal plane along the luminal long-axis (three-dimensional fast field echo, 3D FFE; FOV = 120 x73 mm; resolution, 0.5 x0.5 x1 mm; array size = 256 × 256 × 10; repetition time/echo time, TR/TE = 25/2.6 ms; flip-angle, FA = 20°; scan time = 143 s). After locating suspected plaques on the coronal angiographic scans, high-resolution trans-axial spin-spin relaxation time (T₂)-weighted scanning of the vessel wall was performed (3D turbo spin-echo, TSE factor = 9; FOV = 20 × 20 mm; TR/TE = 1500/32 ms; resolution = 80 μm × 80 μm × 2 mm; FA = 90°; 4 averages; scan time = 1514 s) to reveal the fibrous cap. To determine the effect of image resolution on measurements of FCT, the same section of a human iliac specimen was imaged with the same sequence but at 80, 100 and 140 μm in-plane resolution.

The locations of the high-resolution trans-axial images were registered to the vessels by first measuring their distance to the ends of the vessels as visualized on the coronal images. After CMR, vessels were decalcified for 48 hrs and cut into 5-10 mm sections, each facing the location of the corresponding trans-axial image to the nearest millimeter. The histological sections were visually inspected for antenna-related injury to the vessel wall, embedded in wax, microtomed, and stained with hematoxylin and eosin (H&E) at this institution's pathology service. The sections were then photographed with a microscope-mounted digital camera.

To test whether high-resolution 3T interventional CMR with a loopless detector could be used to provide accurate measures of FCT, thirty-one diseased carotid artery segments with atherosclerotic lesions identified by palpation, were harvested from sixteen adult human cadavers. These vessel segments were fixed in 10% formalin at 3°C for 24 hr. High resolution T₂-weighted CMR of the vessel wall was performed on vessels mounted in the saline phantom, followed by sectioning and histology as described above.

The FCT was measured in both the CMR and the histology sections between the two shoulder ends of the fibrous cap at an interval of 480 μm, or at about every 6 pixels in the 80 μm CMR. Distance was measured in histology sections using a movable scale overlaid on the digital micrograph, and in CMR by counting pixels and multiplying by the resolution. Measurements were restricted only to the regions with clear fibrous cap and lumen borders. The average of all of the FCT measurements from each image was recorded as the mean FCT for that image section. The mean FCT for the iliac specimen imaged at 80, 100 and 140 μm resolution was measured the same way.

High-resolution intravascular CMR was performed in vivo using the 3T biocompatible loopless detector in a healthy adult New Zealand white rabbit, and a 30-month old male Watanabe heritable hyper-lipidemic (WHHL) rabbit which is prone to spontaneously develop atherosclerotic lesions. The study was approved by this institution's Animal Care and Use Committee. The animals were sedated and positioned supine in the 3T scanner, and the detector inserted into the aorta via the femoral artery and a 5-French introducer under CMR guidance. The same T₁-weighted angiographic coronal scan as in the in vitro studies was used, except that the scans were cardiac-triggered. The location of the detector's junction in the animal relative to the renal bifurcation was determined by X-ray fluoroscopy and on the coronal T₁ angiographic images. The locations of the high-resolution trans-axial images relative to the detector's junction and to the center of the renal bifurcation were noted for subsequent co-registration.

High-resolution trans-axial CMR of the aortic wall was performed above the renal bifurcation. A T₂-weighted TSE sequence modified for black-blood with cardiac gating was used for the healthy rabbit (FOV = 80 × 40 mm; TR/TE = 3000/21 ms; resolution = 80 μm × 80 μm × 3 mm; array size = 1008 × 1008; FA: 90°; 4 averages; scan time = 597 s), and a 3D T₁ weighted TSE black-blood sequence with shortened scan-time was used for the hyperlipidemic rabbit (FOV = 20 x10 mm, TR/TE = 429/26; resolution = 80 μm × 80 μm × 2 mm; array size = 256 × 256 × 10; FA: 90°; 8 averages; scan time = 436 s). After CMR, the rabbits were humanely sacrificed. The aortas were harvested and cut into 5 mm sections facing the locations at which trans-axial CMR was performed, referenced relative to the renal bifurcation. The sections were visually inspected for injury, then H&E stained for histology as above.

The computed absolute SNR distribution of the loopless detector is shown in Figure 1. The highest SNR occurs at the detector's whip-cable junction at a radial distance r = 0.1 mm, falling to 17% of the peak at r = 5 mm. However, excluding the tip [35], the absolute SNR of the device for small r is preserved to at least half of the peak value over an extended region in the longitudinal plane for distances z ≤ 80 mm from the device junction.

Figure 3 exemplifies T₁-weighted angiographic scout CMR, and high-resolution trans-axial images acquired from human iliac artery specimens in vitro. In coronal angiographic images, calcified plaques appeared as regions of low intensity or signal voids, such as those between the vessel walls below the iliac bifurcation in Figure 3(A). The guide-wire CMR detector itself contains no mobile protons and also appears as a signal void: the absence of any image decoupling artifact around the probe attests to the efficacy of the tune/match/decoupling circuitry [33, 35]. The extended region of sensitivity permits high-resolution imaging virtually anywhere in these lesions, and the trans-axial 80 μm image of the iliac vessel wall (Figure 3B) reveals a calcified plaque as a void separated from the fluid-filled lumen by the fibrous cap. The vessel wall morphology is confirmed by microscopy of the corresponding histological section (Figure 3C).

CMR of a human iliac vessel acquired at 80, 100, and 140 μm resolution is presented in Figure 4 with corresponding histology. Mean FCT was 360, 243, and 206 μm, when measured with 140, 100, and 80 μm resolution intravascular CMR, respectively (Figure 4A-C). Mean FCT from histology was lower at 186 μm (Figure 4D), consistent with the view that reducing the CMR spatial resolution (increasing voxel size) decreases the accuracy of FCT measurements.

All 31 vessel segments exhibited lesions with fibrous caps based on histology. High-resolution CMR of carotid vessel walls at 80 μm also revealed atheroma in all vessel segments, permitting 3-13 individual FCT measurements from each CMR section, and 3-12 FCT measurements in each of the corresponding histological sections, from which the mean FCT was determined. The mean FCT measured from CMR is plotted against the mean FCT from histology in Figure 5. The mean FCT by histology ranged from 74 μm to 1.1 mm. The two sets of measurements are highly correlated (correlation coefficient, R = 0.96). Even limiting the data to those samples with mean FCT < 650 μm (24 out of 31 samples) as measured by histology, preserves the correlation (y = 0.921x + 29.8, R = 0.91, P = 0.98). Bland-Altman analysis (Figure 6) shows excellent agreement between the FCT measured by CMR and that measured by histology (-1.3 ± 68.9 μm).

The placement of a biocompatible 3T loopless detector in the New Zealand White rabbit aorta in vivo under CMR angiography is exemplified in Figure 7(A). This shows the detector's whip-cable junction where signal intensity is highest (Figure 1), 4 cm above the renal bifurcation. Unlike CMR endoscopic probes [33, 34], the receive-only intravascular detector used here reveals the entire aorta and some adjacent anatomical structures, including the renal bifurcation. X-ray fluoroscopy confirmed the location of the whip-cable junction 4 cm above the renal-bifurcation (Figure 7B). An 80 μm trans-axial high-resolution CMR of the aorta delineates the vessel wall in the rabbit (Figure 7C). The vessel wall thickness was measured from the image at 12 equi-angular locations around the vessel, commencing at a vessel feature identifiable on both CMR and histological images (solid arrows, Figure 7C, D). The average wall thickness from high-resolution intravascular 3T CMR was 0.54 mm ± 0.06 (SD) (Figure 7C), in agreement with an average thickness of 0.55 ± 0.07 mm measured the same way in the histological section (Figure 7D).

In the hyperlipidemic rabbit, in vivo 80 μm high-resolution T₁-weighted CMR revealed a region of elevated signal intensity on the inner wall of the vessel extending into the lumen (Figure 8A, B), consistent with a pattern of fibrous atheroma tissue [29]. A small atherosclerotic lesion was confirmed in the corresponding histological section from the same location (Figure 8C). This shows a fairly homogeneous layer of fibrous tissue, with a higher lipid content rendering it paler than the adventitia. The area of the plaque region measured from CMR (Figure 8B) is 0.36 mm², while the plaque area in the histological section is 0.33 mm².

Visual inspection of the 5-10 mm sections harvested from the human vessel specimens and the rabbits prior to fixation revealed no evidence of vessel perforation, heat injury or plaque rupture attributable to the intravascular CMR detector.