Influences of Aortic Motion and Curvature on Vessel Expansion in Murine Experimental Aneurysms
Objective—To quantitatively compare aortic curvature and motion with resulting aneurysm location, direction of expansion, and pathophysiological features in experimental abdominal aortic aneurysms (AAAs).
Methods and Results—MRI was performed at 4.7 T with the following parameters: (1) 3D acquisition for vessel geometry and (2) 2D cardiac-gated acquisition to quantify luminal motion. Male 24-week-old mice were imaged before and after AAA formation induced by angiotensin II (AngII)–filled osmotic pump implantation or infusion of elastase. AngII-induced AAAs formed near the location of maximum abdominal aortic curvature, and the leftward direction of expansion was correlated with the direction of suprarenal aortic motion. Elastase-induced AAAs formed in a region of low vessel curvature and had no repeatable direction of expansion. AngII significantly increased mean blood pressure (22.7 mm Hg, P<0.05), whereas both models showed a significant 2-fold decrease in aortic cyclic strain (P<0.05). Differences in patterns of elastin degradation and localization of fluorescent signal from protease-activated probes were also observed.
Conclusion—The direction of AngII aneurysm expansion correlated with the direction of motion, medial elastin dissection, and adventitial remodeling. Anterior infrarenal aortic motion correlated with medial elastin degradation in elastase-induced aneurysms. Results from both models suggest a relationship between aneurysm pathological features and aortic geometry and motion.
Abdominal aortic aneurysm (AAA) is a complex disease that leads to significant morbidity and mortality in the United States.1 AAAs are commonly defined as a 1.5-fold or larger increase in vessel diameter due to a pathological dilation.2 Diagnosis and monitoring are usually performed using noninvasive ultrasonography, but only surgical options exist to prevent continued vessel growth and reduce the risk of rupture. This “wait-and-see” approach is partially because of a lack of understanding of the mechanisms that lead to AAA development and expansion.
As a way to better study disease etiology and progression, murine AAA models have been created that mimic aspects of the human disease.3,4 In particular, 2 chemically induced murine models have become commonly used. The first model is initiated by subcutaneous systemic delivery of angiotensin II (AngII) into hyperlipidemic apolipoprotein E–deficient (apoE−/−) mice, leading to suprarenal AAAs.3 The predictable formation of these aneurysms above the renal arteries in this model is of particular interest because most human AAAs are infrarenal. The rationale for the development of the second model, induced by intraluminal infusion of elastase into the murine infrarenal aorta,4 was based on the disrupted nature of elastin in human AAAs.5–8 Although both of these models produce AAAs, there are significant differences. The AngII apoE−/− model is associated with atherosclerosis and often develops a hematoma,9 both of which are not commonly seen in elastase AAAs. Vessel remodeling is also different between the models because medial elastin ruptures at 1 circumferential location in AngII-infused apoE−/− mice (creating a dissection between the medial and adventitial layers)9 but is degraded in a more diffuse manner in the elastase model.4 This is likely because of differences in underlying mechanisms for each model.
Despite the extensive efforts made in developing murine AAA models, the methods for quantifying disease progression remain inadequate. Most published results, using in situ or ex vivo measurements of aortic diameter, have been limited to 1 point per animal. These studies6,10 have used either a calibrated ocular grid to measure external AAA diameter or histomorphometry with a “shrinkage index” to estimate luminal diameter. Others11 have used a categorical approach in which aneurysms were assigned 1 of 4 scores based on lumen dilation, presence of thrombus, and number of bulbous expansions. Although this approach helps to classify the degree of disease progression in each animal, it does not track growth of each aneurysm over time and is not a quantitative description of AAA shape. Those who have used in vivo ultrasonography12 or MR13 to characterize AAA progression did not quantify vessel motion, curvature, or 3D expansion.
This study characterizes the influence of vessel curvature and wall motion on the location and direction of aortic expansion in experimental aneurysms, with the hope of learning more about the relationship between in vivo biomechanics and AAA pathogenesis. Newly developed MR techniques14 were used to measure aortic curvature, motion, and expansion at multiple points in apoE−/− mice over 28 days. To create AAAs, animals were given either a systemic infusion of AngII through an osmotic pump or an intraluminal infusion of elastase into the infrarenal aorta. After euthanasia, protease activity was estimated by the distribution of activatable fluorescent probes. Finally, histological features revealed the circumferential locations of elastin degradation, hematoma formation, and general vessel remodeling. These data suggest a correlation between biomechanics and murine AAA formation and progression.
A more detailed methods section is included in the supplemental material (available online at http://atvb.ahajournals.org). Experiments were performed with local institutional animal care and use committee approval. All animals used in this study were 24-week-old, male, apoE−/− mice bred on a C57BL/6 background to aid in comparisons between models.
AngII-induced AAAs were created using a previously described murine model.3 Osmotic pumps filled with AngII were implanted in 31 mice, of which 16 survived to day 28. Of these mice, 11 developed suprarenal AAAs, defined as an expansion of at least 50% over the normal lumen diameter. Pumps filled with saline were used as a control (n=6). A detailed description of the intra-aortic elastase infusion was previously published.4,5 Briefly, the infrarenal aorta is exposed, isolated, and infused with 4.5-U/mL type I porcine pancreatic elastase for 5 minutes (specific activity, 5 U/mg protein at 100 mm Hg; E1250; Sigma Chemical Co, St Louis, MO) (n=28 total). Of these, 12 mice survived to day 28, all of which developed infrarenal AAAs. A control group underwent aortic infusion with heat-inactivated elastase (n=6).4 Buprenorphine, 0.05 mg/kg, was given to each mouse in two 50-μL subcutaneous injections before and after surgery.
MR Vessel Imaging
All animals underwent imaging before surgery (day 0) and on days 3, 7, 14, 21, and 28. The following MRI procedures were previously described in more detail.14 Briefly, all MRI was performed at 4.7 T. A 3D time-of-flight sequence, which highlights flowing blood in the aorta, was used to acquire data above and below the renal arteries. These data were used to quantify centerline shift and curvature. Maximum intensity projections of these images were used to plan orthogonal imaging planes for a cardiac-gated 2D time-of-flight sequence with 12-cine frames. These data were used to quantify cyclic strain and the direction of centroid motion.
Image Data Analysis
The magnitude and direction of AAA expansion were determined at the location with the greatest shift between the aneurysm centerline and the centerline of a theoretically healthy vessel (which follows the natural curvature of the aorta). The magnitude of the shift is the actual displacement, and the angle is defined relative to anterior (0°), such that a positive angle corresponds to a leftward shift and a negative angle corresponds to a rightward shift. Methods for quantifying geometric curvature from vessel centerlines were previously described.14,15 Geometric curvature (κ) was defined by the inverse of the radius of a circle as follows:
Curvature was calculated in increments of 0.1 mm down the length of the abdominal aorta before aneurysm formation, and the location of maximum curvature was calculated in reference to the right renal artery. The lumen centroids were calculated from aortic segmentations created with a thresholding technique.16 The centroid of the initial timeframe was defined as the origin, and the magnitude and angle of centroid motion were calculated in reference to this location. The perimeter (P) of each segmentation was calculated by summing the distances between consecutive points around the circumference.16 Maximum Green-Lagrange circumferential cyclic strain was defined as follows: where Psys and Pdias are the perimeters of the vessel at systole and diastole, respectively.
Blood Pressure and Ultrasonography
Blood pressure was noninvasively measured with a volume pressure recording sensor and an occlusion tail cuff.17 Systolic, mean, and diastolic pressures were collected at each point. The same animals were imaged at day 28 with a small animal US system (Vevo770). Systolic (maximum) and diastolic (minimum) diameters at the maximum AAA location were measured from M-mode tracings for ≥6 cardiac cycles per mouse.
All aortas were striped with colored dyes for circumferential orientation and then processed and embedded in paraffin. Axial sections (thickness, 5 μm) were stained with hematoxylin-eosin and elastic–Van Gieson (EVG 87017) to visualize elastic lamina. Other sections were stained for smooth muscle cells using an anti–smooth muscle actin antibody (NB600–531) and macrophages using an F4/80 antibody (MCA497GA). With the AngII model, the circumferential location of adventitial hematoma and the octant corresponding to the middle of the medial elastin rupture were recorded. With the elastase model, the amount of medial elastin area was quantified by drawing regions of interest in both the anterior and posterior sections and then calculating an elastin area over the inner region of interest perimeter ratio (in micrometers).
Two activatable fluorescent probes allowed for imaging of the spatial distribution of relative protease activity (MMPSense 680 and ProSense 750). Probes were administered through a tail or jugular vein injection 24 hours before euthanasia at day 28. Fluorescent images of the excised aortas from both ventral and dorsal surfaces were acquired (FX Pro). The mean signal intensity was measured in the AAA, the normal thoracic aorta, and the background noise. A ratio of signal intensity was used to estimate the relative fold change as follows:
Each AngII AAA was then divided into 4 approximately equal quadrants to determine probe accumulation patterns within each AAA.
Leftward AngII AAA Expansion and Varied Elastase Expansion
AngII-induced aneurysms developed directly above the right renal artery (Figure 1A). Vessels expanded asymmetrically, forming saccular aneurysms on the left side of the aorta between the celiac and superior mesenteric arteries (supplemental Figure I). From 11 mice that developed AAAs, 1 developed an aneurysm between days 0 and 3, 1 between days 3 and 7, 8 between days 7 and 14, and 1 between days 14 and 21. At day 28, the maximum aortic diameter was located 1.09±0.57 mm above the right renal artery. In contrast, all 12 mice infused with elastase developed aneurysms at the site of the intraluminal infusion midway between the left renal artery and the aortic trifurcation (supplemental Figure II). A small initial lumen expansion was immediately seen at day 3, likely because of the mechanical damage from the pressurized infusion (Figure 1B). None of the 6 heat-inactivated control mice developed AAAs (day 0, 0.85±0.03 mm; day 3, 0.92±0.05 mm; day 28, 1.00±0.06 mm). Although vessels in both models continued to expand over 28 days, the average volume/length ratio was stable (Table 1 and Table 2). Aneurysm centerlines were calculated for both AAA models from consecutive 2D segmentations of the 3D time-of-flight acquisition (Figure 1C). Starting at day 14, the AngII AAAs shifted leftward while elastase aneurysms had no common direction of growth (Figure 1D). The elastase centerline shift magnitude increased as the vessel expanded but varied in direction.
Maximum Curvature and AngII AAA Formation Occur at Similar Locations
The maximum geometric curvature of the abdominal aorta (supplemental Figure III) was 0.16±0.02 mm−1 and directed leftward (supplemental Table I). This occurred slightly above the right renal artery (2.43±0.54 mm) and is located at a position where AngII-induced aneurysms form. The geometric curvature in the infrarenal aorta was anterior in direction and significantly less than the suprarenal curvature (0.09± 0.02 mm−1, P<0.05). The direction of vessel curvature did not change significantly in either model before aneurysm formation (data not shown).
Changes in Lumen Centroid Motion Differ Between Models
Aortic centroid motion changed more substantially in the AngII model. Suprarenal aortic motion decreased in magnitude significantly between day 7 and 14 and stayed reduced through day 28 (Figure 2 A). These were significant reductions of 52%, 43%, and 44% for days 14, 21, and 28, respectively, compared with day 0 (P<0.05). The angle of centroid motion also changed at day 14, shifting roughly 30° more anterior after aneurysm formation. In the elastase model, the magnitude of centroid motion did not change dramatically (a 15% reduction at day 28 compared with day 0; P=0.27). Yet, the angle of motion shifted rightward after surgery, with a maximum shift of 36° at day 21. This change may be because of manipulation of the peritoneum or separation of the aorta from the vena cava during the infusion procedure.
Direction of Aortic Motion and AAA Expansion Are Similar in the AngII Model
Mice given AngII showed a similar direction of suprarenal aortic motion at day 0 (86.9°±8.5°) and aneurysm expansion at day 28 (77.0°±24.1°). When plotted against each other, all 11 AngII AAA mice fell within the upper right quadrant (Figure 2B), corresponding to both leftward suprarenal aortic motion and leftward vessel expansion. In contrast, elastase AAA expansion was much more variable, suggesting little correlation between day 0 infrarenal motion (−3.9°±30.2°, approximately anterior) and day 28 vessel growth (17.7°±107.8°). Supplemental Table II shows the correlation factors between AAA expansion, aortic motion, and aortic curvature for both models. AngII AAAs showed a much closer correlation between all 3 than elastase aneurysms.
Cyclic Strain Decreases in Both AAA Models
Green-Lagrange cyclic strain waveforms for both models calculated at 12 points through the cardiac cycle are shown in Figure 3. The shape is similar for all waveforms, and a significant decrease in maximum cyclic strain is seen in both models by day 28. Mice given AngII had their maximum suprarenal strain decrease from 20.8%±4.2% at day 0 to 10.0%±2.4% at day 28 (a 52% decrease). Likewise, elastase-infused mice had their maximum infrarenal strain decrease from 19.4%±2.5% at day 0 to 10.7%±1.6% at day 28 (a 45% decrease). Despite these similar reductions, the rate of change in maximum strain differed between the 2 models. AngII aneurysms had a large decrease in strain between day 7 (19.4%±7.7%) and day 14 (10.8%±1.5%), coinciding with the period of fastest vessel expansion. Elastase-induced AAAs had a more gradual decrease in strain over 28 days, with the largest decrease occurring between day 0 (19.4%±2.5%) and day 3 (15.2%±3.5%). Ultrasonographic measurements at day 28 produced similar data when compared with the MR maximum strain measurements (AngII, 8.9%±1.2%; elastase, 8.4%±2.6%; solid and dashed horizontal lines in post-AAA formation plots).
AngII Induces Hypertension and Moderately Increased Pulse Pressure
Systemic infusion of AngII over 28 days increased blood pressure (supplemental Figure IV). In mice infused with AngII, systolic blood pressure increased significantly by day 7 compared with day 0 (P<0.05), reaching a constant increase of approximately 30% (or 25 mm Hg) by day 14. Diastolic pressure showed a similar increase in days 7 through 28. Pulse pressure (calculated as systolic minus diastolic pressure) significantly increased by approximately 9 mm Hg at days 21 and 28 (P<0.05).
AngII AAA Elastin Breakdown Is Focal and Leftward
Histological analysis confirmed vessel expansion and dissection of the aorta, with significant formation of adventitial hematomas in many AngII AAAs. Our in vivo MR images (supplemental Figure VA) were qualitatively confirmed by histological features (supplemental Figure VB) because flowing blood in the lumen and organizing hematoma both produced a signal.18 Often, a large hematoma formed between the medial and adventitial layers, leading to a collection of stationary blood and layered fibrin (sections 6 and 7). Other axial locations show both true and false lumens (asterisks), where blood circulated (sections 1, 2, and 3). Colored striping around the circumference (Figure 4A and 4B) showed that most AAAs experienced focal medial elastin breakdown in the left anterior region (supplemental Table III). Although the adventitial hematomas most often formed on the left side of each vessel, the right side of these AAAs appeared normal. Many of the AngII aneurysms also showed intimal thickening, smooth muscle cell migration out of the media, and prominent macrophage accumulation in the media and adventitia (supplemental Figure VI).
Elastase AAA Elastin Breakdown Is Diffuse and Greatest in the Anterior Wall
Elastin degradation, smooth muscle apoptosis, intimal thickening, and inflammatory infiltrate were present in the histological analysis of the elastase aneurysms. Expanded lumens are seen in both the histological sections and MRIs (supplemental Figure VII), and some AAAs had noticeable intimal hyperplasia, creating thickened vessel walls (sections 6, 7, and 8). Vessels showed significant breakdown of medial lamellar elastin, with more degradation in the anterior wall than the posterior wall (Figure 4C and 4D). Finally, immunohistochemistry revealed smooth muscle cell apoptosis in the medial layer and a mixed inflammatory infiltrate of macrophages and other mononuclear cells in the adventitia, media, and neointima (supplemental Figure VIII).
Activatable Fluorescent Probes Colocalize to Regions of Vascular Injury
Near-infrared fluorescent signals from matrix metalloproteinases and cathepsin-activatable fluorescent probes colocalize to regions of abdominal aortic expansion in both ventral and dorsal images (Figure 5 A). Aneurysms from both AngII and elastase showed a significant 2-fold or greater increase in signal ratio compared with saline pump and heat-inactivated elastase controls (P<0.05, Figure 5B). A fluorescent signal may be related to vascular injury because animals infused with AngII who did not develop AAAs did not show a similar increase in the suprarenal aorta signal ratio. Furthermore, a similar pattern of activated probe accumulation was seen in most AngII AAAs (supplemental Figure IX). The left inferior quadrant showed a significant increase in normalized signal ratio compared with all other quadrants (P<0.01). The fluorescent signal from elastase AAAs was homogeneous and formed no repeatable pattern.
The results of this study suggest that aortic curvature and pulsatile lumen expansion influence the location and direction of vessel expansion in experimental aneurysms. AngII AAAs form at the location of maximum abdominal curvature; and the direction of lumen expansion, curvature of the vessel, and shape of these saccular AAAs were all leftward in the suprarenal region. The elastase-induced aneurysms contribute as a significant negative control to the hypothesis that highly curved aortic locations with large asymmetrical centroid motion may strongly influence the direction of AAA expansion because curvature and centroid motion were much less in the infrarenal region. Thus, the finding that elastase-induced AAAs had variable directions of expansion was not surprising. The periods of largest decrease in cyclic strain coincided with the points of greatest vessel expansion in both models. The histological analysis in both models suggests a correlation between vessel breakdown and aortic motion because regions of increased elastin degradation and the direction of lumen expansion were similar in both models. Finally, the near-infrared fluorescence images suggest that protease-activated probes colocalize to regions of vascular injury for both models.
The formation of AngII AAAs at the location of maximum abdominal curvature suggests a correlation between the amount of vessel curvature and the location of aortic dissection. Although differences in the elastin/collagen ratio or the origins of the medial cells may also play a role in AngII AAA localization,19,20 simple vessel geometry may influence the location of aneurysms as well. A correlation between the location of murine AAA formation and maximum curvature may be because of increased local strain within the left side of the suprarenal aorta or the effects of asymmetrical blood flow. Indeed, previous research21 in patients who experience above-knee amputation showed a propensity of aneurysms to form in the same direction as the lost limb, possibly because of flow asymmetry in the infrarenal region from the unilateral reduction of iliac artery caliber. Our results agree with these findings because flow into the superior murine right renal artery may affect the left side of the suprarenal aorta. Although these results are intriguing, caution should be taken when associating vessel curvature and asymmetrical flow with aneurysm formation because other portions of the aorta (eg, aortic arch) are more curved than the abdominal region. Yet, the effects of curvature are intriguing because others have recently observed significant lumen dilatations of the ascending aorta of AngII-infused apoE−/− mice,22 suggesting that vessel curvature may also play a role in thoracic aortic expansion.
The difference in AAA shape between models may be because of the underlying mechanisms that lead to vessel expansion. Our histological analysis showed aortic dissection and medial elastin breakage on the left side of the vessel in AngII AAAs, often creating true and false lumens that are associated with the formation of adventitial hematomas (supplemental Figure V). This saccular formation was seen in none of the elastase AAAs, where diffuse elastin degradation was observed around the circumference. Thus, although chemotaxis of inflammatory cells into the aortic wall occurs in both models, elastin degradation (and the mechanisms that lead to vessel expansion) is markedly different between AngII- and elastase-induced AAAs.
The pulsatile aortic motion analysis showed that strain was inversely proportional to aortic expansion. However, the rate of this decrease differed between AngII and elastase AAAs. AngII mice showed an abrupt and significant decrease temporally coinciding with AAA formation (day 14). Conversely, the elastase-infused mice had a more gradual decrease in maximum strain values, with the largest decrease between days 0 and 3. This is likely because of mechanical damage from the elastase infusion itself and not vascular remodeling because the aortic diameter had not increased significantly by day 3. The decrease in strain from days 3 to 28 may also be related to the significant increase in transmural pressure seen in the AngII model (leading to increased wall stress) or the smooth muscle cell migration and proliferation in the intimal layer we observed in the elastase AAAs (supplemental Figure VIII).
The enlargement we observed in elastase-induced AAAs (60% and 81% increases at days 21 and 28, respectively) did not reach the 100% threshold proposed by others.4 Although unexpected, the MR techniques used in this study quantified a different AAA metric. Previous research typically measured the external aortic diameter at 1 location by exposing the vessel via laparotomy. The MR technique described herein measures the in vivo internal lumen diameter of the aorta when surrounding tissue can provide restraint and does not easily measure vessel thickening.
Many proteases, including cathepsins K, L, and S and matrix metalloproteinases 2 and 9, have all been associated with human AAA disease, making them interesting targets for potential therapeutic agents.23,24 Previous work has also shown that AngII-induced murine AAAs are linked to the presence of matrix metalloproteinases.25 Our results suggest that proteases are not evenly distributed throughout AngII AAAs (supplemental Figure IX). In general, the inferior half of these aneurysms had increased accumulation of activated probe compared with the superior half. Furthermore, increased signal in the left inferior region of suprarenal aneurysms agrees with our histological analysis, which also showed elastin degradation and adventitial remodeling on the left side. Previous atherosclerosis research26 has shown that the gelatinase activity associated with extracellular matrix remodeling activates the near-infrared fluorescence probes, which then colocalize with macrophage accumulation. Others27 have shown that regions of thin intraluminal thrombus are associated with enhanced proteolytic activity within human AAAs. Thus, our results could indicate increased vascular remodeling or accumulation of inflammatory cells within specific regions of each aneurysm. Future work should be focused on developing in vivo near-infrared fluorescence techniques to provide insight into how active protease levels change during AAA progression.
In summary, this study shows the ability of MR angiography to evaluate aneurysm development at multiple points, highlighting differences between the 2 experimental models. The location of AngII-induced AAAs is similar to the location of maximum abdominal aortic curvature, and the leftward direction of aneurysm expansion in this model appears to be correlated with the direction of pre-AAA motion. Conversely, elastase-induced AAAs form in a region of low vessel curvature and have no repeatable direction of vessel expansion. None of these geometric or dynamic end points would be possible without the recently developed noninvasive imaging methods described herein. Future work incorporating similar in vivo imaging techniques would be useful to quantify murine AAA progression and provide the potential for improving translation because analogous imaging techniques are found in the clinic. Indeed, further research will be needed to determine if vessel curvature and pulsatile motion have an effect on human AAA expansion. Thus, this quantitative comparison between 2 commonly used murine AAA models will hopefully lead to a greater understanding of how biomechanics affect aneurysm formation and pathogenesis and help to improve future clinical treatment.
Sources of Funding
This study was supported by American Heart Association Predoctoral Fellowship 0815293F (Dr Goergen); and grant 1 P50 HL 83800-04 from the National Institutes of Health (Drs Connolly, Dalman, Taylor, and Tsao).
D.Y. Kallop, A. Gogineni, Dr Weimer, and Dr Greve are employees of Genentech Inc.
We thank Maj Hedehus, PhD, for MRI acquisition assistance; and Diem Huynh, BS, Pauline Chu, MS, Kathleen Sanders, BS, and Jeffrey Eastham-Anderson, MS, for histology assistance.
- Received July 12, 2010.
- Accepted October 26, 2010.
- © 2011 American Heart Association, Inc.
- Brady AR,
- Thompson SG,
- Fowkes FG,
- Greenhalgh RM,
- Powell JT
- Johnston KW,
- Rutherford RB,
- Tilson MD,
- Shah DM,
- Hollier L,
- Stanley JC
- Azuma J,
- Asagami T,
- Dalman R,
- Tsao PS
- Sho E,
- Sho M,
- Nanjo H,
- Kawamura K,
- Masuda H,
- Dalman RL
- Saraff K,
- Babamusta F,
- Cassis LA,
- Daugherty A
- Turner GH,
- Olzinski AR,
- Bernard RE,
- Aravindhan K,
- Karr HW,
- Mirabile RC,
- Willette RN,
- Gough PJ,
- Jucker BM
- Goergen CJ,
- Barr KN,
- Huynh DT,
- Eastham-Anderson JR,
- Choi G,
- Hedehus M,
- Dalman RL,
- Connolly AJ,
- Taylor CA,
- Tsao PS,
- Greve JM
- Choi G,
- Shin LK,
- Taylor CA,
- Cheng CP
- Feng M,
- Whitesall S,
- Zhang Y,
- Beibel M,
- D'Alecy L,
- DiPetrillo K
- Wolinsky H,
- Glagov S
- Daugherty A,
- Rateri DL,
- Charo IF,
- Owens AP,
- Howatt DA,
- Cassis LA
- Wang YX,
- Martin-McNulty B,
- Freay AD,
- Sukovich DA,
- Halks-Miller M,
- Li WW,
- Vergona R,
- Sullivan ME,
- Morser J,
- Dole WP,
- Deng GG
- Deguchi JO,
- Aikawa M,
- Tung CH,
- Aikawa E,
- Kim DE,
- Ntziachristos V,
- Weissleder R,
- Libby P