Validating a bimodal intravascular ultrasound (IVUS) and near-infrared fluorescence (NIRF) catheter for atherosclerotic plaque detection in rabbits

: Coronary artery disease is characterized by atherosclerotic plaque formation. Despite impressive advances in intravascular imaging modalities, in vivo molecular plaque characterization remains challenging, and different multimodality imaging systems have been proposed. We validated an engineered bimodal intravascular ultrasound imaging (IVUS) / near-infrared fluorescence (NIRF) imaging catheter in vivo using a balloon injury atherosclerosis rabbit model. Rabbit aortas and right iliac arteries were scanned in vivo after indocyanine green (ICG) injection, and compared to corresponding ex vivo fluorescence and white light images. Areas of ICG accumulation were colocalized with macroscopic atherosclerotic plaque formation. In vivo imaging was performed with the bimodal catheter integrating ICG-induced fluorescence signals into cross-sectional IVUS imaging. In vivo ICG accumulation corresponded to ex vivo fluorescence signal intensity and IVUS identified plaques. system. Indocyanine green fluorescence signal intensity in vivo was corroborated by corresponding ex vivo fluorescence and white light images of macroscopic atherosclerotic plaques.


Introduction
Cardiovascular diseases are the leading cause of morbidity and mortality worldwide [1], and their burden is expected to rise in the near future given the aging of the population and the increasing rates of obesity and diabetes. Cardiovascular diseases are mainly due to atherosclerotic changes within the vessel wall [2], with lipoprotein accumulation and inflammatory cell proliferation considered as key factors in atherosclerotic plaque formation [3,4].
Despite impressive advances in non-invasive and invasive imaging approaches, the identification of high-risk coronary artery plaques prior to their rupture, and thereby identification of patients at risk for acute events, remains an unmet challenge. As vulnerable plaques are often not flow-limited, they remain undetected by conventional coronary angiography [3]. Different imaging modalities including intravascular ultrasound (IVUS) and optical coherence tomography (OCT) have been established to overcome these limitations [5][6][7]. Although such catheter-based technologies provide high-resolution cross-sectional images of the coronary artery wall, allowing for a detailed visualization of atherosclerotic plaques and corresponding vessel responses [8], information with regard to plaque composition and activity is limited. In recent years, emerging molecular imaging approaches such as nearinfrared fluorescence (NIRF) imaging have been proposed for an improved plaque characterization at a cellular or molecular level. Utilizing a variety of molecular probes, fluorescence imaging has been demonstrated to provide complementary information with regard to plaque activity and inflammation [9][10][11][12][13]. Among others, indocyanine green (ICG), given its ability to accumulate in lipid-loaded macrophages [14] and its established clinical applications [15,16], has been proposed as an attractive agent for molecular fluorescence imaging of cardiovascular diseases [14,17]. ICG is a near-infrared fluorochrome known to bind to albumin, which is ingested by macrophages [18]. However, for in vivo imaging, sensitive molecular technologies need to be combined with high-resolution microstructural imaging tools to allow for a correct interpretation of the fluorescence signal. Advantages of integrated multimodality catheter designs and simultaneous imaging have been recognized [17,19]. Fluorescence imaging catheters with integrated ultrasound guidance have been designed previously [20,21] and validated in vivo in swine arteries [22]. However, previous in vivo applications relied on balloon occlusion to remove blood from the field of view.
In this study, we validated our recently engineered bimodal IVUS/NIRF imaging catheter without balloon occlusion in vivo using a rabbit model of atherosclerosis [23]. Further, in vivo fluorescence imaging was compared to ex vivo ICG accumulation and macroscopic plaque formation, and a blood attenuation compensation algorithm for fluorescence imaging was proposed.

Animal model of atherosclerosis
Inflamed atheroma formation was induced in 9 male New Zealand White rabbits (2.8-3 kg, aged 12-13 weeks; Charles River Laboratories, Saint-Constant, QC, Canada) using a 14week-cholesterol-enriched diet (0.5% cholesterol, Harlan Techlad Diets, Madison, WI; n = 3) or an artery balloon injury (n = 3), or both (n = 3). Balloon injury was performed in the abdominal aorta and the right iliac artery at week 2. At week 14, in vivo and ex vivo imaging procedures were performed, and the animals were sacrificed thereafter by exsanguination under anesthesia.
The experimental protocol was approved by the animal ethics committee of the Montreal Heart Institute Research Center according to the guidelines of the Canadian Council on Animal Care.

Balloon dilation procedures
For balloon injury procedures, anesthesia was induced with intramuscular injections of ketamine (35 mg/kg) and buprenorphine (10 µg/kg), and maintained with inhaled isoflurane (3 to 5% v/v, Baxter, Deerfield, IL) and supplemental oxygen [10]. Then, a 5F transradial introducer (Cordis Corporation, Fremont, CA) was installed into the right carotid artery, and a 0.014" guidewire (Abbott Vascular, Santa Clara, CA) advanced to the aorta. Balloon injury was performed under fluoroscopic guidance utilizing a percutaneous transluminal coronary angiography (PTCA) balloon dilatation catheter (Boston Scientific, Marlborough, MA) inflated to tension. Balloon injury was performed in the proximal 20 mm of the right iliac artery (2.5-mm balloon), and in the distal 40 mm of the abdominal aorta above the iliac bifurcation (4-mm balloon). Three pullbacks were performed in each arterial segment. The left iliac artery served as non-injured controls. After balloon injury and removal of the introducer, the right carotid artery was ligated.

Bimodal intravascular ultrasound (IVUS) / near-infrared fluorescence (NIRF) catheter system
The imaging system consisting of an imaging catheter linked to an optical assembly and custom-made electronics, has previously been described and is shown in Fig. 1 [23]. It combined an 400 µm optical fiber having a numerical aperture of 0.22 for 2-dimensional, 360-degree fluorescence imaging and photoacoustic excitation, and an ultrasound transducer (extracted from a commercially available IVUS 45 MHz catheter; Revolution, Volcano Therapeutics, Rancho Cordova, CA), for acoustic imaging and photoacoustic signal detection at frame rates of up to 30 images/second. A 500 µm right angle prism with an aluminized hypotenuse was glued to the optical fiber for side-fire illumination. With a diameter of 1.4 mm, the system was utilized with 5F introducers. In this work, we focused exclusively on IVUS and NIRF.
A dedicated electronic circuit synchronized acquisition of the different imaging modalities with 2 motors driving the rotating/translating catheter assembly. The pullback speed was set to 0.5 mm/second, with a total pullback length of 100 mm and a frame rate of 10 images/second. A custom Matlab (The MathWorks, Inc, Natick, MA) user interface was used for USB data acquisition, real-time signal filtering, reconstruction, and image display. Optical and electrical connections between the fixed and rotating assemblies used an optical rotary joint for fluorescence imaging (Doric Lenses, Quebec, QC, Canada) and a slip ring to transmit and receive ultrasound signals. A 20 dB preamplifier was designed to rotate with the ultrasound transducer, in order to amplify the signal close to the source. The main ultrasound acquisition circuit included a 5 to 55 dB variable gain amplifier with customizable time-gain compensation and a 14-bit, 200 MegaSamples/second analog-to-digital converter.
Fluorescence excitation was performed with a 780 nm laser diode combined with a narrow-band optical filter (769 ± 20 nm) and emission was detected by a photomultiplier tube (H5783-20, Hamamatsu Photonics, Hamamatsu City, Japan) combined with a narrow-band optical filter (832 ± 19 nm). A dichroic filter (cut-off wavelength: 801 nm) was used to separate the excitation and emission light paths. The optical power at the tip of the catheter was 25 mW. In our phantom validations, we detected 100 nM of ICG through 1.4 mm of blood-mimicking solution. The lateral resolution through blood was 500 µm for an object at 1.3 mm of the outer surface of the catheter.

In vivo intravascular imaging procedures
Imaging procedures were performed as previously described. Animals were anesthetised with intramuscular injections of ketamine (35 mg/kg) and buprenorphine (10 µg/kg), followed by isoflurane (3 to 5% v/v) inhalation. For intravascular imaging procedures, the fluorochrome ICG (10 mg/kg) was utilized, and injected via the right ear vein at the beginning of the procedures. While the fluorochrome was circulating, a 5F introducer was inserted into the left carotid artery, and the bimodal imaging catheter was advanced over the 0.014" guide wire under fluoroscopic guidance. The guide wire was removed prior to image acquisition. Forty minutes after ICG injection, automated imaging pullbacks were performed using the bimodal IVUS/NIRF imaging catheter both in the right iliac artery and the abdominal aorta.

Intravascular ultrasound (IVUS)/ near-infrared fluorescence (NIRF) image analysis
For the 100 mm pullback, manual segmentation of the vessel wall was performed on recorded IVUS images using 1 slice/mm, and interpolation was used to obtain the distance for the remaining slices in between (1900 slices). A dedicated fluorescence blood attenuation compensation algorithm based on the Beer-Lambert law [24] was developed to compensate the fluorescence signal amplitude for the calculated distance between the imaging catheter and the arterial wall. We observed compensation to be dependent on vessel size. Utilizing an effective extinction coefficient of 2 mm −1 was found to be optimal for the vessel sizes that were imaged. For fluorescence detection, a minimum catheter-vessel wall distance of 0.1 mm based on the limitation of IVUS imaging, and a maximum distance of 0.75 mm to avoid background noise were utilized. The blood attenuation compensation algorithm was applied to all the results presented below, unless otherwise specified. Assuming fluorescence to be preferentially accumulated at the vessel lumen surface, an algorithm was applied on the NIRF data for each longitudinal position to account for light attenuation by blood in the lumen. Due to the expected homogeneity of the media, a simple Beer-Lambert law was assumed and is given by, where F c is the corrected fluorescence signal, F m is the measured fluorescence signal at each angular position θ, µ is the extinction coefficient and d is the distance between the catheter and the vessel wall.

Ex vivo imaging
After the sacrifice, animals were perfused with Lactated Ringers solution. Aortas and iliac arteries were excised, conserved in 4% formaldehyde and imaged within 30 minutes in a fluorescence imaging scanner (IVIS Lumina II, Caliper Life Sciences, Hopkinton, MA). The abdominal aorta and iliac arteries of one randomly chosen rabbit per group were opened longitudinally, pinned on a black tray and imaged en face. White light images captured with a camera (D3100, Nikon Corporation, Tokyo, Japan) at an exposure time of 1/15 second for macroscopic atherosclerotic plaque visualization and fluorescence imaging utilizing the fluorescence scanner for ex vivo ICG accumulation detection were performed. For fluorescence, excitation and emission wavelengths were set to 745 nm and 800 nm, respectively. The exposure time was 1 second and no binning was applied.

Ex vivo near-infrared fluorescence (NIRF) imaging: validation of the model
Ex vivo fluorescence imaging was performed for the entire aorta, as well as for the right balloon-injured and the left non-injured iliac arteries (Fig. 2). An increased NIRF signal in balloon-injured arterial segments was observed both for rabbits fed a regular diet and rabbits fed a cholesterol-enriched diet (3 per group; Fig. 2(a)-2(c) and 2(g)-2(i), respectively) as compared to non-injured arterial segments of rabbits fed a cholesterol-enriched diet ( Fig.  2(d)-2(e). The strongest NIRF signal was observed in balloon-injured arterial segments of rabbits fed a cholesterol-enriched diet ( Fig. 2(g)-2(i)). The increased NIRF signal was quantified and is illustrated in Fig. 3, where the signal ratio between right and left iliac arteries and between distal 0-40 mm and 40-80 mm abdominal aorta is compared for each group. Substantial ICG accumulation was observed in the aortic arch and the descending thoracic aorta (Fig. 2(a)-2(i)), as both regions are prone to atherosclerotic plaque formation given the increased shear stress oscillation [25]. The ICG accumulation in these arterial segments was further enhanced in the cholesterol-enriched diet group. En face fluorescence and white light imaging are shown in Fig. 4. In rabbits fed a cholesterol-enriched diet, ICG-induced signal enhancement on fluorescence imaging was colocalized with severely diseased atherosclerotic vessel regions as identified by their opaque and white macroscopic appearance on white light (Fig. 4(c)-4(f)). Similarly, at sites of balloon injury, increased ICG accumulation was observed at sites of scar tissue formation ( Fig. 4(a)-4(b), and 4(e)-4(f)).

In vivo intravascular ultrasound (IVUS) / near-infrared fluorescence (NIRF) imaging
The integration of fluorescence and IVUS imaging in a single catheter allows for simultaneous in vivo molecular and anatomical imaging of the vessel wall. The previously engineered dual-modality IVUS/NIRF imaging catheter system was utilized to assess whether ICG-induced fluorescence can be integrated into IVUS cross-sectional arterial images. Indocyanine green-induced 360-degree fluorescence signals, known to target macrophagerich inflamed plaque areas [14], were integrated into cross-sectional IVUS images of the vessel wall (Fig. 5). IVUS images were reviewed by a trained technician, blind to the NIRF images. NIRF signal enhancement correlated with sites of increased echogenicity (Fig. 5(a), 5(e)), and intimal thickening (Fig. 5(b), 5(c)) identified by the technician on IVUS imaging. Indocyanine green-induced signal enhancement in vivo was colocalized with fluorescence ex vivo as illustrated in Fig. 6. The cholesterol-enriched diet group that did not undergo balloon injury was excluded from this analysis as in this group, neither macroscopic plaque formation nor significant ICG-accumulation was observed in the abdominal aorta.
We observed that in some arterial segments, in vivo ICG-accumulation was stronger as compared to ex vivo fluorescence signal intensity while in others it was weaker ( Fig. 7(a), 7(c), 7(d), 7(f)). These false positive signals may be due, at least in part, to blood attenuation given an increased distance between the catheter and the vessel wall due to increased vessel dimensions or an eccentric position of the catheter within the artery. Therefore, a compensation algorithm to adjust the fluorescence signal amplitude for blood attenuation was developed, and IVUS images were used to measure the distance between the catheter and the vessel wall at any angle and pullback position. Compensated ICG-induced fluorescence signals are shown in Fig. 7(b), and 7(e), 7(h). False positive fluorescence detection was reduced after applying the blood attenuation compensation algorithm for in vivo fluorescence imaging. For the artery segments presented in Fig. 7, a correlation of the fluorescence signal intensity was performed between in vivo and ex vivo measurements, before and after the blood attenuation compensation, and shows a slightly increased R 2 value (from 0.1 to 0.12) with residuals oscillating around zero following compensation, but not prior. Despite compensation, we observed a remaining variability, potentially explained by catheter eccentricity and depth distribution of the fluorophore in the vessel wall.

Discussion
This study validates our recently developed bimodal IVUS/NIRF catheter imaging system for in vivo plaque imaging in a rabbit model of atherosclerosis. The bimodal intravascular imaging system integrating both ICG-induced fluorescence and ultrasound imaging provided the capability to simultaneously acquire molecular and structural plaque characteristics. Further, a blood attenuation compensation algorithm to normalize the fluorescence signal intensity to the distance between the catheter and the arterial wall as measured by IVUS was introduced to enhance the accuracy of the imaging system. Indocyanine green fluorescence signal intensity in vivo was corroborated by corresponding ex vivo fluorescence and white light images of macroscopic atherosclerotic plaques.

Indocyanine green (ICG) and molecular imaging of plaque
Acute coronary syndromes are mainly due to atherosclerotic plaque rupture, thrombus formation, and subsequent vessel occlusion [26]. Emerging molecular imaging strategies aim at identifying inflammatory plaque features, and thereby detecting high-risk plaques prone to rupture, and different molecular agents have been proposed. For over three decades, the fluorochrome ICG has been widely used for different clinical applications including retinal and choroidal vessel imaging or intraoperative evaluation of coronary artery bypass graft patency with proven safety and efficacy [18,27]. Moreover, ICG has previously been demonstrated to target lipid-loaded macrophages in vitro, and when utilized as molecular imaging agent to detect lipid-rich, inflamed coronary-sized plaques containing abundant macrophages in atherosclerotic rabbits in vivo [14,17,28,29]. Different animal models of atherosclerosis have been established, and are mainly based on pro-atherogenic diets and different types of vessel injury [10,30]. Since rabbit abdominal aorta and iliac arteries suit the typical dimensions of human coronary arteries, rabbit atherosclerosis models are widely used for intravascular imaging studies [31]. Therefore, in this study, different rabbit atherosclerosis models, consisting of cholesterol-enriched diet and/or balloon injury, were utilized to investigate ICG imaging with our bimodal catheter. Balloon injury substantially enhanced ICG accumulation within the vessel wall, and, as expected, the combination of both atherosclerosis inducers was the most effective with this regard. However, ICG accumulation was also correlated with plaques in animals that did not undergo the balloon procedure confirming its potential to image macrophages in this model despite the absence of scar tissue. This is in distinction to previous studies using ICG with balloon-injured aortas [14], as we demonstrated ICG accumulation in a more natural plaque development environment. Moreover, ex vivo ICG accumulation, corresponding with in vivo fluorescence signals, was colocalized with visible macroscopic inflamed plaque formation on aorta specimen. When integrating in vivo fluorescence into cross-sectional IVUS images of the arterial wall, enhanced fluorescence was mainly observed at sites of increased echogenicity, but some regions with lower echogenicity also displayed increased fluorescence. As plaque characterization solely by IVUS imaging is limited, fluorescence imaging adds important information about inflammation features on top of structural plaque characterization.

Bimodal imaging systems
With traditional fluorescence imaging systems, morphological information about arterial dimensions is lacking. The combination of both fluorescence and IVUS imaging allows for plaque visualization both at the molecular and morphological level, and thereby provides anatomical guidance for fluorescence signal interpretation. With the engineered catheter imaging system, simultaneous image acquisition with high temporo-spatial resolution at frame rates of 10 images/second were achieved [23], while previously reported multimodality imaging systems mainly utilized lower frame rates for fluorescence imaging with larger diameter catheters.
During in vivo fluorescence imaging, blood absorption of the fluorescence signal occurs depending on the distance between the catheter and the vessel wall. In the iliac arteries, intense ICG-induced fluorescence signals were observed in vivo which were not confirmed ex vivo. Separately, in larger arteries and when the imaging catheter was not centered in the vessel, despite substantial ICG accumulation ex vivo, only low signal intensity was observed in vivo given the increased distance between the catheter and the vessel wall. Hence, molecular and morphological imaging data need to be integrated to allow for an accurate interpretation of in vivo measured fluorescence signal intensity. In particular, as the intensity of the fluorescence signal attenuates with increasing distance from the catheter, fluorescence signals need to be adjusted for vessel dimensions. In this study, a blood attenuation compensation algorithm was developed based on measurements of the distance between the imaging catheter and the vessel wall to adjust fluorescence signals not only for vessel dimensions, but also for eccentric catheter positions. Application of the blood attenuation compensation algorithm reduced false positive, and enhanced true positive fluorescence signals. Similarly, a distance compensation algorithm has previously been proposed for a bimodal OCT/NIRF imaging system [17].

Limitations
The proposed blood attenuation compensation algorithm based on IVUS co-registration aimed at reducing the above-described inherent technical limitations, but was not able to overcome them completely. With the engineered catheter system and ICG as fluorochrome, fluorescence could not be detected at catheter-vessel wall distance exceeding 1 mm, because the signal was too low. Therefore, plaque analysis in both modalities together could be assessed only where the distance between the catheter and the arterial wall was below 1 mm, and vessels with a diameter between 2 mm and 3 mm were ideal for this experiment. SNR improvements combined to a more robust compensation algorithm could resolve this issue. Saline flushing, potentially combined with partial or full vessel occlusion, might also be considered to increase the sensitivity of the catheter imaging system. In that case, faster pullback speeds should be investigated to limit the injected volume. Also, an over-the-wire approach may favor a centered catheter position. As only modest abdominal aorta and iliac plaque formation was observed in the cholesterol-enriched diet group, a more aggressive diet could have been used to further enhance plaque formation. It would also enable quantifying plaque volumes in IVUS and correlating it with NIRF signal enhancement.

Conclusion
The bimodal IVUS/NIRF imaging catheter system, previously engineered by our group, was validated for in vivo plaque imaging in a rabbit atherosclerosis model. Simultaneously acquired IVUS/NIRF images of the vessel wall were acquired and allowed for an integrated molecular and structural analysis. Integrating molecular and IVUS imaging in a single catheter, ICG-based fluorescence for in vivo molecular imaging was further advanced, and in near future, the identification of lipid-rich, inflamed plaques prone to rupture might become possible. Thereby, high-resolution multimodality imaging technologies may provide new avenues for coronary plaque characterization utilized both in clinical practice and research applications.