Measurement of enhanced vasa vasorum density in a porcine carotid model using photon counting detector CT

Abstract.

Purpose

The onset of atherosclerosis is preceded by changes in blood perfusion within the arterial wall due to localized proliferation of the vasa vasorum. The purpose of this study was to quantify these changes in spatial density of the vasa vasorum using a research whole-body photon-counting detector CT (PCD-CT) scanner and a porcine model.

Approach

Vasa vasorum angiogenesis was stimulated in the left carotid artery wall of anesthetized pigs ($n=5$) while the right carotid served as a control. After a 6-week recovery period, the animals were scanned on the PCD-CT prior to and after injection of iodinated contrast. Annular regions of interest were used to measure wall enhancement in the injured and control arteries. The exact Wilcoxon-signed rank test was used to determine whether a significant difference in contrast enhancement existed between the injured and control arterial walls.

Results

The greatest arterial wall enhancement was observed following contrast recirculation. The wall enhancement measurements made over these time points revealed that the enhancement was greater in the injured artery for 13/16 scanned arterial regions. Using an exact Wilcoxon-signed rank test, a significantly increased enhancement ratio was found in injured arteries compared with control arteries ($p=0.013$). Vasa vasorum angiogenesis was confirmed in micro-CT scans of excised arteries.

Conclusions

Whole-body PCD-CT scanners can be used to detect and quantify the increased perfusion occurring within the porcine carotid arterial wall resulting from an increased density of vasa vasorum.

1. Introduction

Vasa vasorum are microscopic vessels that perfuse the muscular wall of large vessels and are known to play an important role in the development of atherosclerotic changes within arterial walls.^1–6 It has been shown that atherosclerotic plaque formation is affected by local increases in vasa vasorum spatial density within the arterial wall. Further, such plaque formation can lead to luminal stenosis and reduced blood flow to target organs.^6–9 An increase in vasa vasorum density tends to precede the development of plaques, thus, we propose that the quantification of arterial wall vasa vasorum density could provide a metric for early detection of plaque development prior to luminal stenosis.^6,8

There are several technical challenges involving vasa vasorum quantification in human subjects. Visualization of vasa vasorum using intra-arterial ultrasound and optical coherence tomography can enable counting of the number/density of vasa vasorum.^10,11 Invasive methods of this variety often involve selective catheterization of arteries of interest. Use of computed tomography (CT) angiography to image the passage of an intravenously administered bolus of contrast agent through the vasa vasorum would be less invasive and could enable three-dimensional (3D) visualization of the arteries and plaques. However, CT angiography imaging faces challenges with respect to limited spatial resolution, partial volume averaging, luminal contrast blooming, poor signal-to-noise ratio in arterial walls, and radiation exposure.

While clinical CT scanners can generally resolve high-contrast objects of $\sim 300$ to $500\mu \mathrm{m}$, vasa vasorum diameters typically range between only 5 and $150\mu \mathrm{m}$, making it unlikely that these vessels can be directly resolved and quantified using conventional CT.^6,12,13 Finite CT detector pixel sizes also cause partial volume averaging to occur at the interface between the arterial lumen and wall in the reconstructed images.^14,15 Such partial volume averaging causes blooming artifacts to manifest within the arterial wall during contrast perfusion. Compared with the luminal contrast enhancement, signals originating from enhanced vasa vasorum are relatively weak, yielding only about 10% to 15% of the total luminal enhancement.^6,16 In addition, the regions of interest (ROIs) available for signal measurements are very small, as the carotid artery wall thickness is only about 30% of the carotid lumen radius.^6,8,17 Further, due to the patchy nature of plaque formation, encountering heterogeneous signals in the arterial wall is very likely.¹⁸

Reproducible assessment of vasa vasorum enhancement is essential for successful application in research and clinical settings. This requires a CT system and imaging scenario that yield high signal-to-noise properties. However, care must be taken to ensure that the radiation dose and injected volume of contrast required to achieve these criteria are within clinically acceptable levels. To address these challenges, we propose to use a photon-counting detector CT (PCD-CT) system, as PCD technology has demonstrated benefits in terms of decreased electronic noise influence and improvements in CT number accuracy, contrast-to-noise ratio, and dose efficiency.^19–31 Specifically, the research whole body PCD-CT system used in this work (SOMATOM CounT) has previously demonstrated benefits in terms of improved spatial resolution,^32,33 increased iodine signal-to-noise ratio,^34,35 and increased dose efficiency.^34–36 Our approach seeks to indirectly quantify the spatial density of the vasa vasorum in the carotid artery wall by measuring the cumulative contrast enhancement in the wall, rather than by resolving individual vasa vasorum.³⁷ The purpose of this study, therefore, was to quantify increased vasa vasorum density in the arterial walls of an in vivo swine model of vasa vasorum proliferation using a whole-body PCD-CT system.

2. Methods

2.1. Research Whole-Body, Photon-Counting-Detector CT Scanner

The investigational PCD-CT scanner used in our study was a Somatom CounT (Siemens Healthcare, Forchheim, Germany), which was built upon a modified 128-slice, second-generation dual-source CT scanner (Somatom Definition Flash, Siemens Healthcare).^35,38,39 The A-subsystem was equipped with a conventional energy-integrating detector (EID) while the B-subsystem was equipped with a CdTe PCD. Both x-ray sources were Straton tubes (Siemens Healthcare). In this study, we used the ultra-high resolution (UHR) acquisition mode, which featured a $0.7\mathrm{mm}\times 0.7\mathrm{mm}$ focal spot size, a $0.25\mathrm{mm}\times 0.25\mathrm{mm}$ effective pixel pitch (at isocenter), and two energy thresholds. The in-plane field-of-view (FOV) was 275 mm for the PCD and 500 mm for the EID. Data truncation occurred in the axial plane while scanning objects larger than the PCD FOV; this was corrected using a data completion scan performed using the EID subsystem.⁴⁰

2.2. Animal Model

This animal study was approved by our Institutional Animal Use and Care Committee. We used a porcine model of increased vasa vasorum density within the arterial wall originally developed for use in rabbits.¹⁰ The animals ($n=5$) were anesthetized (telazol/ketamine/xylazine) and the carotid arteries were exposed via a mid-line incision of the neck. The left carotid’s arterial wall was injected with 0.25 ml of autologous blood medially and laterally at four sequential locations (total of 2 ml for eight injections) between the carotid artery bifurcation and a caudal location about 6 to 8 cm proximal (Fig. 1). The right carotid artery was kept intact and served as the study’s control. The neck incision was closed, and the animal recovered from anesthesia. Hereafter, the left carotid is referred to as the “injured” artery and the right carotid is referred to as the “control” artery.

Fig. 1.

(a) Midline incision along the pig neck that was used to access the common carotid arteries and administer the injections of autologous blood. (b) Injection method used in this study: eight injections were performed down the length of the left carotid artery in four pairs of medial/lateral injections. (c) Scar on ventral surface of neck occurring along the midline incision 6 weeks after injury (red arrows mark suture endings). (d) Anterior topogram acquired with the CT scanner on scan day. Four distinct $\sim 8\text{-}\mathrm{mm}$ scan regions (I to IV) were selected along the $\sim 8\text{-}\mathrm{cm}$ carotid region associated with the autologous blood injections.

The animals were given a 6-week post-surgery recovery period to allow for vasa vasorum proliferation at injection sites. Animals were socially housed and were provided with various enrichment material. Their diet included twice daily feeding of a grain-based dry chow mix. After 6 weeks, the animals were re-anesthetized, and their necks were scanned at four independent locations on the research PCD-CT scanner described in Sec. 2.1. A $1\text{-}\mathrm{ml}/\mathrm{kg}$ body weight bolus of iodine-based contrast agent [Omnipaque^® 350 (iohexol), GE Healthcare, Inc., Chicago, Illinois)] followed by a 30-ml saline chaser, was injected into the femoral vein at $10\mathrm{ml}/\mathrm{s}$. Immediately following contrast injection, repeated axial scans (no table translation) were acquired every 3 s over a period of 60 s, resulting in 20 “time point” scans per location I to IV [Fig. 1(d)]. Scan parameters included 140-kV tube potential, 341-mA tube current, 1-s gantry rotation time, $32\times 0.25\mathrm{mm}$ detector collimation, and energy threshold settings of 30 and 70 keV. The scan and contrast injection procedure was repeated at each carotid region so as to sequentially cover 6 to 8 cm of the artery (i.e., the range over which the carotid was previously injected with autologous blood). Fifteen minutes were allowed between successive scans to permit contrast washout from the arterial walls.⁴¹

After scan completion, the animals were euthanized. The carotid arteries were exposed and flushed with heparinized saline. The arteries were then injected at 100 mmHg pressure with Microfil^® (Flow Tech, Inc., Carver, Massachusetts), a lead-doped silicon-based polymer that hardens after several hours.

2.3. Ex Vivo Micro-CT Scans

Micro-CT scans of the carotid arteries were acquired to visually confirm increased vasa vasorum spatial density in the injured arteries relative to the control arteries. First, the Microfil^® containing arteries were excised from the animal and submerged in formalin for 24 h. The arteries were then placed into a thin-walled plastic cylinder and were embedded within paraffin wax. Ex vivo scans of the specimens were acquired on a custom-built micro-CT system.⁴² The detector was a Pixis-XB detector (Pixis-XB: 1300, Princeton Instruments), which had a pixel pitch $0.02\times 0.02\mathrm{mm}$, a CsI scintillator, and Tb fiber optic plate. The traditional x-ray source (XRD C-Tech Tube, PANalytical, Eindhoven, The Netherlands) featured a molybdenum anode ($K_{\alpha }$ emission at 17.5 keV) and a zirconium foil filter, which restricted the spectrum to $17.5\pm 1\mathrm{keV}$. The corresponding images were reconstructed using a Feldkemp filtered backprojection algorithm and a voxel size of $20\times 20\times 20{\mu \mathrm{m}}^3$. Image data were loaded into Analyze 12.0 (AnalyzeDirect, Overland Park, Kansas), a 3D viewing software, which was used to generate volume renderings of each carotid artery pair.

2.4. CT Image Reconstruction and Postprocessing

The PCD-CT images were reconstructed using an iterative reconstruction algorithm (SAFIRE, Siemens Healthcare) and a sharp quantitative kernel (Q65). Neck images displaying both the left (injured) and right (control) carotid arteries were reconstructed with a 1 mm thickness and 0.5 mm increment. The reconstruction FOV was 80 mm and the image matrix size was $1024\times 1024$, resulting in voxel sizes of $0.0781\times 0.0781\times 1{\mathrm{mm}}^3$. The reconstruction range along the $z$ axis included the central 7 mm for each scan region I to IV, which yielded a total of 13 CT images per carotid scan region per time point.

To reduce image noise, an in-house developed multienergy nonlocal means (MENLM) algorithm was applied. The denoising performance of the MENLM algorithm was investigated in previous work by our group.⁴³ Briefly, this algorithm exploits spatiospectral data redundancy of the multienergy data sets to reduce image noise.

2.5. Enhancement Measurement Procedure

The arterial wall enhancement, corresponding to the perfusion of iodinated contrast agent into the vasa vasorum, was measured using annular ROIs. Wall enhancement comparisons between the injured and control vessels were made using an identical segmentation and analysis procedure, which included: (1) carotid lumen segmentation and annular ROI generation, (2) blooming correction, (3) vessel masking, and (4) lipid tissue masking. This processing pipeline is outlined in Fig. 2 and the steps are further described in Secs. 2.5.1–2.5.4.

Fig. 2.

Image processing and segmentation workflow. The four main steps occurring between CT image denoising and carotid wall enhancement analysis include: (1) lumen segmentation and annular ROI generation, (2) blooming correction, (3) vessel masking, and (4) lipid masking.

2.5.1. Lumen segmentation and annular ROI generation

Iodine perfusion characteristics were assessed throughout the 20 time point scans in all carotid regions. For each region, the time point scan capturing the peak iodine enhancement in the arterial lumen was used to identify the wall–lumen boundary (Fig. 3). Identification of the wall–lumen boundary was performed using a half-maximum threshold (HMT) segmentation technique.^37,44 To calculate the HMT, the user first manually positioned a small circular ROI within both the enhanced carotid lumen and the soft tissue surroundings. Next, the HMT was calculated as the halfway point between the respective mean attenuation values measured in lumen and the soft tissue surroundings [Fig. 3(c)]. An annular ROI was then generated just beyond the wall–lumen boundary to encompass the artery wall [Fig. 3(d)]. Concentric single voxel thick annular rings were generated within the wall, expanding outward from the wall–lumen boundary, and were used to evaluate the wall attenuation in their respective areas. The concentric annuli receiving the largest contrast signal would be selected for the ensuing wall enhancement measurements in this study.

Fig. 3.

HMT segmentation technique. (a) The time point scan capturing the peak contrast enhancement was selected (window level/width: 230/750 HU). (b) The exact boundary of the enhanced carotid lumen cannot be easily delineated. (c) The HMT segmentation procedure used to define the approximate lumen boundary. (d) An annular wall ROI generated beyond the approximated wall-lumen boundary to encompass the carotid artery wall.

2.5.2. Blooming correction forward model

Annular wall ROIs were processed with a custom blooming correction code³⁷ implemented in MATLAB (vers. 2018b, The MathWorks Inc., Natick, Massachusetts). Briefly, a 30-cm anthropomorphic chest phantom was used to measure the iodine blooming magnitude occurring around vessels of three different diameters (2, 4, and 8 mm) containing two different concentrations (21 and $35\mathrm{mg}/\mathrm{ml}$) of the same iodinated contrast described in Sec. 2.2. The CT acquisition and reconstruction parameters were the same as noted in Secs. 2.2 and 2.4. The vessel lumen boundaries were segmented as described in Sec. 2.5.1. Annular ROIs were generated radially outward from the lumen boundary to produce seven 1-voxel-thick concentric annuli. The mean enhancement occurring within each annular ROI was measured for each vessel diameter and for both iodine concentrations. A look-up-table (LUT) was created from the measured blooming data and was then used to predict the magnitude of blooming contamination expected to occur within voxels neighboring the iodine-enhanced vessel lumen. The model’s accuracy was verified using a vessel phantom (Appendix, Sec. 6.1).

In this study, the carotid artery lumen was segmented, the attenuation due to iodine enhancement was recorded, and the segmented diameter was calculated. Next, the blooming expected within each wall voxel was predicted based on the target voxel’s radial distance from the lumen, and the mean iodine enhancement within the arterial lumen. Interpolation was used between data points in the LUT. The predicted portion of attenuation due to iodine blooming was then subtracted from the voxel.

2.5.3. Vessel mask

Noncarotid vessels could interfere with attenuation measurements due to anatomical variations in their proximity to the carotid arteries. During iodine enhancement, such nearby vessels could falsely contribute to the attenuation measurements within the carotid wall ROI. This situation occasionally arose with the jugular vein [Fig. 4(a)] and small branch vessels of the carotid artery [Fig. 4(b)]. Masks were generated for these vessels and were used to suppress attenuation contributions to the carotid wall ROI. Any voxels within the mask’s segmented area were omitted from downstream enhancement measurements.

Fig. 4.

Anatomical variations in vessel and tissue locations occasionally required additional preprocessing to enable consistent measurements. Such anatomical variations were observed from animal to animal, carotid region to carotid region, and even between the right and left carotids of the same animal and region. (a) Enhanced jugular veins (black arrows) immediately adjacent to the unenhanced carotid arteries (white arrows) for right and left carotid artery (right and left panes, respectively) of the same CT image slice. The jugular veins become enhanced at later time points than the arteries. (b) Small branch vessels of the carotid artery (black arrows) immediately adjacent to the carotid artery lumen. The proximity of the branch vessels to the carotid lumen varied. (c) Unenhanced carotid arteries (black arrows) showing different amount of neighboring lipid tissue between the right and left carotids. (d) Carotid bifurcation leading to a noncircular lumen shape (black arrows).

2.5.4. Lipid mask

Lipid tissue residing around the carotid artery could interfere with attenuation measurements due to anatomical variations in proximity to the carotid artery wall [Fig. 4(c)]. In addition, the fraction of lipid tissue contained within the wall ROI could be different for the right and the left arteries. Further, minor pulsatile motion between adjacent time point scans could lead to additional variability. Simple masks were generated for such lipid tissue via thresholding and were used to suppress contributions to the downstream carotid wall ROI measurements. The attenuation of the unenhanced soft tissue composing the carotid arterial wall was expected to range from 20 to 40 HU,⁴⁵ while by comparison, unenhanced lipid tissue was expected to range from $-30$ to $-190\mathrm{HU}$.^46,47 In this study, lipid masks were created by thresholding voxels with values of $\le -30\mathrm{HU}$.

2.6. Wall Enhancement Analysis

A baseline correction procedure, based-on the methodology developed previously by Sheikhzadeh et al.¹⁵ for quantifying enhanced microvascular density in myocardium specimens, was applied to quantify the vasa vasorum density within the carotid artery wall (Appendix, Sec. 6.2). The volume of vasa vasorum in the arterial wall, after normalization by the volume of the wall, would indicate the spatial density of vasa vasorum. This quantity could be practically calculated using CT image data by taking the ratio of iodine enhancement measured in the carotid wall compared to that measured in the lumen, as follows:

$$F=\frac{{\mathrm{CT}}_{\text{wall}}-{\mathrm{CT}}_{\text{wall}}(t=0)}{{\mathrm{CT}}_{\text{lumen}}-{\mathrm{CT}}_{\text{lumen}}(t=0)},$$ (1)

where $F$ is the fractional enhancement ratio, ${\mathrm{CT}}_{\text{wall}}$ is the attenuation during iodine enhancement as measured in the carotid wall ROI, ${\mathrm{CT}}_{\text{lumen}}$ is the attenuation during iodine enhancement as measured in the carotid lumen ROI, ${\mathrm{CT}}_{\text{wall}}(t=0)$ is the unenhanced attenuation of the carotid wall as measured in the carotid wall ROI, and ${\mathrm{CT}}_{\text{lumen}}(t=0)$ is the unenhanced attenuation as measured in the carotid lumen ROI. For a given scan region, this enhancement ratio was calculated image-by-image for both the injured and control artery.

Due to the patchy nature of vasa vasorum proliferation within the arterial wall, our study also sought to compare the cumulative enhancement spanning the total length of each sampled carotid region. Thus, the mean of the image-specific enhancement ratios was also calculated and was used to approximate the wall enhancement experienced along the length of the scanned region. Comparisons between the mean enhancement of the injured and the control artery are reported.

2.7. Statistical Analysis

For each of the evaluable carotid artery regions, the injured and control wall enhancement ratio of each individual CT slice were displayed graphically using box plots and identify the median, 25th and 75th percentiles. For each region, the mean and standard deviation of the slice-specific wall enhancement ratios were also calculated for the injured and control arteries. The mean wall enhancement values were tested for increased enhancement ratio in the injured arteries compared with the control arteries using an exact Wilcoxon-signed rank test. A $p\text{-}\text{value}<0.05$ was considered statistically significant.

3. Results

3.1. Animal CT Scans

Physiological characteristics of the animals used in this study are summarized in Table 1. For simplicity, the animals are numbered (pig #1 through pig #5) and the four scan regions per animal are also numbered (R1 through R4). A portion of two scan regions (pig #2 R1 and pig #3 R1) were acquired too distally and resulted in a nonideal partial scan of the carotid bifurcation [Fig. 4(d)]. For both regions, the most distal 9/13 CT slices, which captured the bifurcation, were omitted from downstream enhancement measurements. Two of the animals had a negative physiological reaction to the injected iodinated contrast agent. These animals were administered vasopressin to promote recovery from shock. The first animal (pig #1) was unable to be resuscitated and thus data were unable to be collected from the three unscanned regions (italic text in Table 1). The later animal (pig #3) was stabilized after onset of reaction, and the three remaining regions were scanned (bold text in Table 1).

Table 1.

Summary of animal physiological characteristics. R1 to R4 correspond to identified carotid artery scan regions 1 to 4. Italic used to identify regions left unscanned due to animal death. Bold used to identify scanned regions occurring after vasopressin administration.

Pig	Sex	Weight (kg)	Iodine reaction	Heart rate (BPM)				Blood pressure (mmHg)
				R1	R2	R3	R4	R1	R2	R3	R4
1	M	30	Y	80	N/A	N/A	N/A	143/101	N/A	N/A	N/A
2	F	26.5	N	96	94	87	87	80/40	79/37	81/40	98/54
3	M	31	Y	122	98	98	98	128/85	89/52	86/50	83/48
4	M	29	N	123	121	98	87	132/72	110/58	99/50	96/52
5	M	27	N	120	110	112	109	116/66	121/71	118/68	116/65

3.2. Iodine Perfusion Characteristics

Iodine perfusion characteristics were evaluated throughout the 20 discrete time point scans acquired at each carotid region. All profiles presented a large initial contrast enhancement curve in the carotids, followed by a smaller secondary enhancement curve due to contrast recirculation. Sixteen of the 17 scanned regions were discovered to have similar perfusion curves. The contrast perfusion profile corresponding to these 16 regions is shown in Fig. 5, where the solid line denotes the mean enhancement profile, and the gray shading denotes the standard deviation. An abnormal perfusion curve was observed in a single scan region, belonging to pig #3 R2, which was acquired immediately following the iodine reaction and successful resuscitation with vasopressin. The secondary enhancement curve due to contrast recirculation deviated in timing and shape from other regions (dashed line in Fig. 5). This scan region was excluded from the data analysis to avoid introducing uncertainty due to this significant change in physiology.

Fig. 5.

Time enhancement profile observed during contrast delivery as measured within the injured carotid artery lumen. The successive sequential time point scans were acquired in 3 s intervals. While the magnitude of attenuation due to contrast enhancement varied among the scanned regions, the primary (peak) and secondary (recirculation) contrast enhancement curves were found to have similar timing and shape among 16/17 scanned regions. The black solid line denotes the average of these profiles, and the gray shading represents the standard deviation. The time enhancement profile of the scan affected by an iodine reaction and vasopressin administration is denoted by a dotted line. The primary (peak) and secondary (recirculation) enhancement curves differed in timing and shape compared to the other regions, and the data from this region were excluded from subsequent analysis.

3.3. Enhancement Measurements

The time point scan found to capture the peak luminal iodine enhancement was identified from the perfusion curve of each carotid region and was used for lumen and wall ROI segmentation as outlined in Sec. 2.5.1. After segmentation, the average carotid lumen radius was measured as $2.18\pm 0.27\mathrm{mm}$ for the right artery and $2.15\pm 0.18\mathrm{mm}$ for the left artery. The carotid wall was therefore estimated to be about 0.35 mm thick (i.e., about 30% of the lumen radius). Concentric single voxel thick annular rings were generated throughout the expanse of the wall and indicated that the majority of contrast enhancement occurred within the inner half of the wall thickness (Fig. 6). Therefore, an annular wall ROI with a radial thickness of three voxels ($\sim 0.234\mathrm{mm}$) was used to measure the arterial wall enhancement in this study.

Fig. 6.

(a) CT image of animal neck region during peak enhancement. (b) Cropped region encompassing enhanced carotid artery. (c) Same image with seven superimposed single voxel thick annular ring ROIs. (d) A plot of the measured attenuation compared among the annuli for the unenhanced time point scan (dashed line), original enhanced scan (solid line with triangle markers), and blooming corrected data (solid line with square markers). The area between the unenhanced and blooming corrected data are taken to be the enhancement due to iodine perfusing the wall. The majority of wall enhancement is observed to occur within the inner half of the wall thickness.

Attenuation measurements were made within the carotid wall using the three-voxel thick annular wall ROI during iodine enhancement. In addition, the attenuation due to unenhanced anatomical tissue was measured using the same wall ROI within the initial (unenhanced) time point scan. These resulting attenuation measurements were used to calculate the fractional enhancement ratio, $F$, detailed in Sec. 2.6. The time point scans observed to capture the greatest enhancement ratio occurred after the arterial contrast recirculation curve (Fig. 7). In general, this period corresponded with the following range: $T+6$ to $T+16$, where $T$ is the time point capturing the peak luminal enhancement. The enhancement ratio was calculated image-by-image for all 16 scan regions over that time period and are presented in Fig. 8.

Fig. 7.

Average enhancement ratio calculated among all 16 scanned regions at each time point scan occurring after peak contrast enhancement (peak time point = $T$). The successive sequential time point scans were acquired in 3 s intervals. The largest difference between injured and control artery enhancement ratio followed the contrast recirculation curve. In general, this period corresponded to $T+6$ through $T+16$.

Fig. 8.

Enhancement ratio distribution for the injured and control artery among all images of the 16 evaluated regions. Each circle represents the enhancement ratio calculated for a single CT slice. (a), (c)–(e), (g)–(p) Most carotid regions had 13 consecutive CT image slices, while (b), (f) two regions had only 4. The $x$ axis includes both (left) the injured and (right) control arteries and the $y$ axis shows the dimensionless enhancement ratio magnitude. Box plots were generated to indicate additional information: central horizontal line indicates the median of the CT slices, upper and lower box edges indicate the 75th and 25th percentiles, respectively, upper and lower whiskers extend to: $q_3\pm 1.5*(q_3-q_1)$, where $q_1$ and $q_3$ are 25th and 75th percentiles, respectively. Any points beyond the whiskers were considered outliers and are marked as “+” along the box axis. All subplots were generated using Matlab.

3.4. Statistical Analysis

The injured and control wall enhancement ratios for each individual CT slice in each of the 16 evaluable carotid artery regions are shown in Fig. 8. The mean and standard deviation of the individual CT slice enhancement ratios for the injured and control arteries in all 16 evaluable regions are reported in Table 2. Using an exact Wilcoxon-signed rank test, a significantly increased enhancement ratio was found in the injured arteries compared to control arteries ($p=0.013$).

Table 2.

Enhancement ratio mean and standard deviation for all 16 scan regions. A larger magnitude of this ratio indicates a larger attenuation due to contrast enhancement was observed in the carotid arterial wall.

Scan region	Injured	Control	Scan region	Injured	Control
1	0.41 ± 0.02	0.40 ± 0.02	9	0.39 ± 0.04	0.46 ± 0.02
2	0.40 ± 0.01	0.37 ± 0.01	10	0.47 ± 0.01	0.43 ± 0.01
3	0.40 ± 0.01	0.38± 0.01	11	0.53 ± 0.02	0.50 ± 0.03
4	0.48 ± 0.00	0.42 ± 0.02	12	0.48 ± 0.01	0.47 ± 0.02
5	0.49 ± 0.01	0.44 ± 0.01	13	0.49 ± 0.01	0.45 ± 0.01
6	0.43 ± 0.01	0.43 ± 0.02	14	0.41 ± 0.02	0.39 ± 0.01
7	0.40 ± 0.01	0.40 ± 0.01	15	0.38 ± 0.01	0.36 ± 0.01
8	0.40 ± 0.01	0.39 ± 0.01	16	0.34 ± 0.00	0.35 ± 0.02

3.5. Micro-CT Evaluations

Volume-rendered micro-CT images of the excised carotid specimens are shown in Fig. 9. These images provide a visual realization of the vasa vasorum proliferation in the injured vessel relative to control. In all animals, there are regions along the length of the carotid arteries where the injured artery appears to have a greater spatial density of vasa vasorum than control; however, there are also spans where no notable difference was observed, which is consistent with the discrete localized injections of autologous blood.

Fig. 9.

(a)–(e) Volume renderings of all five contrast-infused carotid artery pairs investigated in this study. Micro-CT scan data of the individual excised and Microfil infused vessel specimens were used to generate these volume renderings. The display windows of the 3D volumes were set to the Microfil contrast, thus only the Microfil infused vascular lumens are visible, while the soft arterial tissue and surrounding paraffin wax are absent. Upon visual inspection, there are regions along the length of each animal’s injured vessel that appear to possess a larger proliferation of vasa vasorum than control.

4. Discussion

In this work, we used a PCD-CT system to compare the attenuation due to contrast enhancement within the wall of injured (induced vasa vasorum proliferation) and control carotid arteries. Of the 16 carotid regions, 13 ($\sim 80\%$) were found to have a greater wall enhancement ratio for the injured artery compared to control [Figs. 8(a)–8(e), 8(g), 8(h), 8(j)–8(o)]. Conversely, 3/16 regions had a greater wall enhancement ratio for the control artery [Figs. 8(f), 8(i), and 8(p)]. Scan region #9 [Fig. 8(i)] had the largest control artery wall enhancement signal relative to the injured artery, this data were obtained from the most distal region (R1) of pig #4 and had carotid branch vessels in both the injured and control wall ROIs [Fig. 4(b) displays a slice obtained from this region]. While such branch vessel lumens were masked, it is possible that some residual contribution may have remained to influence the wall enhancement measurements. Such extreme branching may necessitate a more robust exclusion method. There were also 4/16 regions [Figs. 8(a), 8(f), 8(g), and 8(p)] where little apparent difference in wall enhancement ratio was observed between the arteries. One possible explanation is that the axial scans of these 4/16 carotid artery regions resided in-between the discrete injection sites intended to induce vasa vasorum proliferation [i.e., falling in between the blue arrows in Fig. 1 (b)].

The micro-CT scans of the excised arteries were performed at a much higher resolution ($20\mu \mathrm{m}$ effective pixel size) than the CT scans ($250\mu \mathrm{m}$ effective pixel size), which permitted visualization of individual vasa vasorum. Visual inspection of the volume renderings for each carotid artery pair (Fig. 9) reveals that a larger vasa vasorum density existed along the injured arteries compared with control. This finding is consistent with the results obtained from the CT data, which revealed that a statistically significant difference existed between the wall enhancement measurements of the injured and control vessels, and where the injured artery was observed to have a greater wall enhancement in the majority of paired carotid regions evaluated.

We chose the vasa vasorum as an index of early atherosclerosis for two reasons. First is the close relationship between the increase in vasa vasorum density in proportion to plaque size,⁷ which is important as this might allow detection of early plaque formation prior to their encroachment on the arterial lumen.¹⁷ As atherosclerosis is a systemic disease of the arterial wall, demonstration of vasa vasorum density increase in the carotid artery may be applicable to other arteries, in terms of incidence and severity of atherosclerosis.⁴⁸ Second, the newly formed (and therefore probably more fragile) vasa vasorum could be a major contributor to hemorrhage into the plaque and arterial lumen, i.e., the cause of the ruptured “vulnerable” plaque.^10,49,50 Although not present in the animal model of this study, calcification within the arterial wall is a common phenomenon in the presence of atherosclerosis.^51,52 Hence, in a clinical situation, the calcium must be distinguished from the opacification due to the contrast agent within the wall. Material decomposition can achieve this provided suitable spectral x-ray CT is available.^53,54

While there are several imaging methods used to detect and quantify atherosclerotic plaques,^54,55 CT as a method for quantifying vasa vasorum density is minimally invasive. The goal of our work was to develop a technique that detects presymptomatic atherosclerosis, i.e., before there is an acute arterial lumen occlusion or narrowing of the lumen by a large plaque. The use of photon counting detector multienergy CT imaging is associated with reduced electronic noise, higher iodine signal-to-noise-ratio, and improved spatial resolution, which all contribute to minimizing the radiation exposure needed. PCD-CT technology continues to be a topic of great research interest, and recently an FDA-approved system has become available and offers larger field of view and higher resolution than its predecessor.^56,57 In addition, the possible role of artificial intelligence and other algorithmic methods to reduce radiation exposure is currently a topic of widespread interest.^58–61

This study had several limitations. First, without a more precise method to synchronize the scanning locations with the injection sites, there were several instances where nonideal locations encompassing the carotid bifurcation were scanned. Second, two animals had a negative reaction to the injected iodine bolus following the first CT scan. This resulted in three regions not being scanned (due to animal death), one region having abnormal perfusion characteristics (due to the reaction and subsequent administration of vasopressin), and two normal regions. In this study, the region with the abnormal perfusion profile was omitted from the enhancement calculations. Third, nonidealities introduced by the excision and specimen preparation process prevented a one-to-one comparison of the micro-CT scan locations with the scanned CT regions. These included: vessel length shrinkage after removal, damage during excision at the cut artery ends, contrast agent leakage out of vessel lumen in damaged areas, and contortion of the arteries after exposure to hot wax. Finally, minor pulsatile motion was observed between time point scans of animals with high heart rates.

5. Conclusions

We demonstrated that the wall enhancement measured within the injured carotid arteries was greater than the control in the majority of the evaluated regions. The associated vasa vasorum densities within the arterial walls were able to be estimated within images obtained from whole-body UHR PCD-CT scans. The CT vasa vasorum density estimates were found to be comparable to the data resulting from the micro-CT evaluations of the excised specimens. These results suggest that UHR PCD-CT angiography can be used to detect vasa vasorum proliferation in porcine carotid arterial walls. Considering vasa vasorum proliferation is an early biomarker for atherosclerosis, these results warrant further investigation in a clinical evaluation to determine the viability for potential integration into diagnostic or screening procedures.

6. Appendix

6.1. Blooming Correction Forward Model Evaluation on Vessel Phantom

An in-house vessel phantom with known physical dimensions was used to provide a preliminary evaluation of the blooming correction technique’s accuracy. The phantom was composed of a 6-mm-diameter plastic straw (the vessel’s lumen) and a 0.5-mm-thick layer of attenuating tape wrapped around the outside of the straw (emulating enhanced “vasa vasorum”). This vessel phantom was scanned twice, the first time it was filled with deionized water (offering only the pure vasa vasorum signal), the second time filled with $22\mathrm{mg}/\mathrm{ml}$ of iodine contrast (providing a blooming corrupted signal in the “vasa vasorum”).

A total of seven concentric annular ROIs were generated around the contrast infused lumen in images from the iodine scan [Fig. 10(a)]. These same seven ROI were superimposed onto the images of the water scan [Fig. 10(b)] such that they could be used for determination of the vessel’s “ground truth” wall enhancement (free of luminal contrast blooming). A mean enhancement value was calculated from the voxels contained within each annular ROI. The resulting CT numbers were used to compare the vessel wall enhancement observed in the following images: (i) the iodine infused lumen scan, (ii) the water infused lumen scan, and (iii) the iodine infused lumen scan after blooming correction. The unique blooming contamination occurring within each ROI was subtracted from its mean CT number as described in Sec. 2.5.2. Wall enhancement measurements were graphically displayed to facilitate a visual comparison of this data [Fig. 10(c)]. It can be noted that the wall enhancement curve representing the iodine infused lumen scan becomes very similar to the ground truth curve after blooming correction was applied.

Fig. 10.

(a) CT image of the vessel phantom containing $22\mathrm{mg}/\mathrm{ml}$ iodine solution. The seven concentric annular ROIs are superimposed onto the image. (b) CT image of the vessel phantom containing deionized water. The same seven ROI are superimposed onto the image. In this phantom, the enhanced layer of tape appears to reside between the fourth and seventh ROI. (c) Effect of iodine blooming correction on the CT numbers measured in the ROI of the vessel phantom images. The corrected scan (circle markers) is nearly an exact match with the water scan (dashed line).

6.2. Baseline Correction for Iodine Signal from Vasa Vasorum in Porcine Artery Walls

Abbreviations

• •Iodine in wall: {I,wall}
• •Tissue in wall: {T,wall}
• •Artery wall: wall
• •Artery lumen: lumen
• •Water: water
• •Blood: B
• •Time point: $t=n$, $(n=0,1,\dots ,N)$
• •Vasa vasorum: vv
• •Computed tomography: CT
• •Micro-CT: $\mu \mathrm{CT}$
• •Injured carotid: inj
• •Control carotid: ctrl

Notations for physical units

• •Linear attenuation coefficient of a tissue/material $x$ is ${\mu }_x$
• •Mass attenuation coefficient of a tissue/material $x$ is $(\frac{\mu }{\rho }{)}_x$
• •Density is $\rho$, mean density is $\overline{\rho }$
• •Volume is $V$
• •Mass is $M$
• •
  CT number of a specific material (say iodine in wall) is given as

  $${\mathrm{CT}}_{\{I,\text{wall}\}}=\left(\frac{{\mu }_{\{I,\text{wall}\}}-{\mu }_{\text{water}}}{{\mu }_{\text{water}}}\right)\times 1000.$$

Baseline correction for iodine signal in artery wall

The mass of iodine in the wall is given as

$$M_{\{I,\text{wall}\}}={\overline{\rho }}_{\{I,\text{wall}\}}*V_{\text{wall}}.$$ (2)

Since the wall is comprised of tissue and vasa vasorum (vv), we can rewrite the above equation with respect to the vv component alone as

$$M_{\{I,\text{wall}\}}={\rho }_{\{I,\mathrm{vv}\}}*V_{\mathrm{vv}}.$$ (3)

Note in Eq. (3), the value of $\rho$ is the density of iodine in individual vasa vasorum, while the density measured in the wall as shown in Eq. (2) is a mean density.

The mean density of iodine in the wall can be related to the density of iodine in the lumen by realizing that the density of the iodine in the vv is equivalent to that in the lumen. By substitution:

$${\overline{\rho }}_{\{\mathrm{I},\text{wall}\}}*V_{\text{wall}}={\rho }_{\{\mathrm{I},\text{lumen}\}}*V_{\mathrm{vv}}.$$ (4)

The diffusion of iodine into the arterial wall tissue is not considered here because the CT measurements are taken as the bolus of iodine moves through the artery, which occurs before diffusion into the wall tissue.

The proportion of the wall containing vv (the fractional enhancement ratio, $F$) can then be obtained from Eq. (4):

$$F=\frac{V_{\mathrm{vv}}}{V_{\text{wall}}}=\frac{{\overline{\rho }}_{\{\mathrm{I},\text{wall}\}}}{{\rho }_{\{\mathrm{I},\text{lumen}\}}}.$$ (5)

The CT number measured in the wall using an annular ROI can be written as

$${\mathrm{CT}}_{\text{wall}}=\left(\frac{(\mu /\rho {)}_I*{\overline{\rho }}_{\{\mathrm{I},\text{wall}\}}+(\mu /\rho {)}_T*{\rho }_{\{\mathrm{T},\text{wall}\}}-{\mu }_{\text{water}}}{{\mu }_{\text{water}}}\right)\times 1000.$$ (6)

Equation (6) can be rewritten as

$${\mathrm{CT}}_{\text{wall}}=\left(\frac{(\mu /\rho {)}_I*{\overline{\rho }}_{\{\mathrm{I},\text{wall}\}}}{{\mu }_{\text{water}}}\right)\times 1000+\left(\frac{(\mu /\rho {)}_T*{\rho }_{\{\mathrm{T},\text{wall}\}}-{\mu }_{\text{water}}}{{\mu }_{\text{water}}}\right)\times 1000.$$ (7)

$${\mathrm{CT}}_{\text{wall}}=\left(\frac{(\mu /\rho {)}_I*{\overline{\rho }}_{\{\mathrm{I},\text{wall}\}}}{{\mu }_{\text{water}}}\right)\times 1000+{\mathrm{CT}}_{\text{wall}}(t=0).$$ (8)

Equation (8) can be further rewritten as

$$(\frac{(\mu /\rho {)}_I*{\overline{\rho }}_{\{\mathrm{I},\text{wall}\}}}{{\mu }_{\text{water}}}\times 1000)={\mathrm{CT}}_{\text{wall}}-{\mathrm{CT}}_{\text{wall}}(t=0).$$ (9)

Note that ${\mathrm{CT}}_{\text{wall}}$ corresponds to time points ($t>0$) unless specified explicitly as ${\mathrm{CT}}_{\text{wall}}(t=0)$, which corresponds to the baseline acquisition.

Further rearranging Eq. (9) with respect to mean iodine density in the wall:

$${\overline{\rho }}_{\{\mathrm{I},\text{wall}\}}=\frac{{\mathrm{CT}}_{\text{wall}}-{\mathrm{CT}}_{\{\mathrm{T},\text{wall}\}}(t=0)}{1000*{(\mu /\rho )}_I}\times {\mu }_{\text{water}}.$$ (10)

Similarly, the CT number of lumen can be written in terms of its attenuation and iodine density components as

$${\mathrm{CT}}_{\text{lumen}}=\frac{(\mu /\rho {)}_I*{\overline{\rho }}_{\{\mathrm{I},\text{lumen}\}}+(\mu /\rho {)}_B*{\rho }_B-{\mu }_{\text{water}}}{{\mu }_{\text{water}}}\times 1000,$$ (11)

$${\rho }_{\{\mathrm{I},\text{lumen}\}}=\frac{{\mathrm{CT}}_{\text{lumen}}-{\mathrm{CT}}_{\text{lumen}}(t=0)}{1000*(\mu /\rho {)}_I}\times {\mu }_{\text{water}}.$$ (12)

Combining Eqs. (5), (10), and (12), we get

$$F=\frac{V_{\mathrm{vv}}}{V_{\mathrm{wall}}}=\frac{{\overline{\rho }}_{\{\mathrm{I},\text{wall}\}}}{{\rho }_{\{\mathrm{I},\text{lumen}\}}}=\frac{{\mathrm{CT}}_{\text{wall}}-{\mathrm{CT}}_{\text{wall}}(t=0)}{{\mathrm{CT}}_{\text{lumen}}-{\mathrm{CT}}_{\text{lumen}}(t=0)}.$$ (13)

Note at ($t=0$), ${\mathrm{CT}}_{\text{wall}}$ and ${\mathrm{CT}}_{\text{lumen}}$ correspond to attenuation from wall tissue and luminal blood, respectively, without any iodine signal.

Biographies

Jeffrey F. Marsh received his BS degree in bioengineering from the University of Illinois at Chicago in 2016. He is now a biomedical engineer working in the Department of Radiology at Mayo Clinic in Rochester, Minnesota, USA. His current research projects involve photon counting detector computed tomography and the application of artificial intelligence in medical imaging.

Andrew J. Vercnocke received his BS degree in computer science from Winona State University in 2004. He is currently a medical imaging analyst at the Mayo Clinic in Rochester, Minnesota, USA. His research interests lie in the development of micro-CT imaging for clinical research applications, and he has helped researchers gather data by way of this modality for numerous studies.

Kishore Rajendran received his BE degree in biomedical engineering from Anna University in 2009, his MS degree in medical electronics from Coventry University in 2011, and his PhD in radiology from the University of Otago in 2016. Currently, he is an assistant professor in the Department of Radiology at Mayo Clinic, Rochester, Minnesota, USA. His current research focuses on photon-counting CT for early detection of vascular diseases, CT image reconstruction, and CT noise reduction.

Shengzhen Tao received his PhD in biomedical engineering and physiology from Mayo Clinic Graduate School of Biomedical Sciences in 2017. He is currently a medical physicist specialized in diagnostic imaging. His current research interests include MRI pulse sequence design, image reconstruction, MRI system, and clinical applications.

Jill L. Anderson recently retired from her role as a research supervisor and program manager at the Mayo Clinic, Rochester, Minnesota, USA. She received her animal health technology degree in 1980 at the University of MN-Technical College in Waseca, Minnesota, USA, and her MN Certified Vet Tech license in 1984. She received her BA degree in management and communication from Concordia College in 2000.

Erik L. Ritman, MD, PhD, is a recently retired physiologist who used CT to explore pathophysiological processes in vivo. In the 1970s and 1980s, he developed and used high-speed volume CT scanning, and in the 1990s and 2000s, he developed and used micro-CT for biomedical research applications.

Shuai Leng received his BS degree in engineering physics in 2001, his MS degree in engineering physics from Tsinghua University in 2003, and his PhD in medical physics from the University of Wisconsin Madison in 2008. He is an associate professor of medical physics at the Mayo Clinic in Rochester, Minnesota, USA. He has authored over 160 peer-reviewed articles. His research interest is in technical development and clinical application of x-ray and CT imaging.

Cynthia H. McCollough, PhD, is a professor of radiological physics and biomedical engineering at Mayo Clinic, where she directs the CT Clinical Innovation Center. Her research interests include CT dosimetry, advanced CT technology, and new clinical applications, such as dual-energy and multi-spectral CT. She is an NIH-funded investigator and is active in numerous professional organizations. She is a fellow of the AAPM, ACR, and AIMBE. She received her doctorate from the University of Wisconsin in 1991.

Disclosures

CH McCollough is the recipient of a research grant from Siemens Healthcare. The equipment and software discussed in this paper are not commercially available.