- Email: [email protected]
Noninvasive detection of intimal xanthoma using combined ultrasound, strain rate and photoacoustic imaging
Ultrasonics 52 (2012) 435–441
Contents lists available at SciVerse ScienceDirect
Ultrasonics journal homepage: www.elsevier.com/locate/ultras
Noninvasive detection of intimal xanthoma using combined ultrasound, strain rate and photoacoustic imaging Iulia M. Graf a, Seungsoo Kim a, Bo Wang a, Richard Smalling b, Stanislav Emelianov a,⇑ a b
Department of Biomedical Engineering, University of Texas at Austin, Austin, TX 78712, USA Division of Cardiology, University of Texas Health Science Center, Houston, TX 77030, USA
a r t i c l e
i n f o
Article history: Received 30 March 2011 Received in revised form 29 September 2011 Accepted 10 October 2011 Available online 18 October 2011 Keywords: Ultrasound Strain Photoacoustics Intimal xanthoma Atherosclerosis
a b s t r a c t Background and motivation: The structure, composition and mechanics of carotid artery are good indicators of early progressive atherosclerotic lesions. The combination of three imaging modalities (ultrasound, strain rate and photoacoustic imaging) which could provide corroborative information about the named arterial properties could enhance the characterization of intimal xanthoma. Methods: The experiments were performed using a New Zealand white rabbit model of atherosclerosis. The aorta excised from an atherosclerotic rabbit was scanned ex vivo using the three imaging techniques: (1) ultrasound imaging of the longitudinal section: standard ultrasound B-mode (74 Hz frame rate); (2) strain rate imaging: the artery was flushed with blood and a 1.5 Hz physiologic pulsation was induced, while the ultrasound data were recorded at higher frame rate (296 Hz); (3) photoacoustic imaging: the artery was irradiated with nanosecond pulsed laser light of low fluence in the 1210–1230 nm wavelength range and the photoacoustic data was recorded at 10 Hz frame rate. Post processing algorithms based on cross-correlation and optical absorption variation were implemented to derive strain rate and spectroscopic photoacoustic images, respectively. Results: Based on the spatio-temporal variation in displacement of different regions within the arterial wall, strain rate imaging reveals differences in tissue mechanical properties. Additionally, spectroscopic photoacoustic imaging can spatially resolve the optical absorption properties of arterial tissue and identify the location of lipid pools. Conclusions: The study demonstrates that ultrasound, strain rate and photoacoustic imaging can be used to simultaneously evaluate the structure, the mechanics and the composition of atherosclerotic lesions to improve the assessment of plaque vulnerability. Ó 2011 Elsevier B.V. All rights reserved.
1. Objective The objective of the study was to explore the relationship between arterial morphology, mechanics and composition by using combined vascular ultrasound, strain rate and photoacoustic imaging and to identify the contribution of each arterial characteristic in the characterization of atherosclerosis. 2. Motivation The etiology of atherosclerosis and its clinical sequelae have been the object of many research studies in the past years. Various markers (i.e. intima media thickness [1], Framingham score [2], stenosis score [3], etc.) and noninvasive imaging techniques (i.e. ultrasound [4] and MRI [5]) are currently used in clinical practice to diagnose atherosclerosis. The difficulty of assessing early ⇑ Corresponding author. E-mail address: [email protected] (S. Emelianov). 0041-624X/$ - see front matter Ó 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.ultras.2011.10.005
atherosclerotic lesions with respect to plaque vulnerability is related to the complexity of the disease. Thus a concomitant analysis of the artery wall’s structure or anatomy, tissue mechanical properties and composition could lead to a superior evaluation of early atherosclerosis, such as intimal xanthoma in relation with vessel remodeling and inflammation in response to atherosclerosis [6–13]. Compared to other approaches (i.e. MRI or CT), ultrasonography is attractive because of its non-invasive character, its portability and associated low-cost. Gray-scale B-mode ultrasound imaging is commonly used in clinical practice to assess tissue morphology and cardiovascular risk estimators, such as the Intima-Media Thickness (IMT) [3,14–16]. The structure of the arterial walls and early atherosclerotic lesions can be partly characterized by the amplitude of ultrasound echo signals in combination with observed tissue specific attenuation [17,18]. Regions with high calcium content exhibit hyperechogenicity, while the lipid core exhibits a low echogenicity [19]. In the case of atherosclerotic lesions with high calcium content, especially the ones attached both to the anterior and posterior walls, the tissue beneath dense
436
I.M. Graf et al. / Ultrasonics 52 (2012) 435–441
calcifications lies within a shaded region, which makes the tissue differentiation very difficult. An additional imaging modality is needed to improve the assessment of plaque’s risk of rupture. A possible way to enhance the characterization of atherosclerotic lesions would be to supplement the information on morphology, as derived from ultrasound, with other arterial characteristic, such as mechanics and composition. Elastography is a medical imaging tool which complements ultrasonography by providing information on tissue mechanical properties [20]. Current clinically available tools of arterial mechanical properties, which are based on noninvasive ultrasound imaging, require the calculation of functional parameters, such as distension, distensibility, compliance, etc. [1,21]. These parameters are generally derived from the averaged displacement of a few centimeters along arterial segment. Thus they do not offer any information on the variation of mechanical characteristics throughout the arterial wall (neither in the axial nor in the lateral direction). The advantage of deriving the mechanical characteristics of the arterial segments two-dimensionally (2D), i.e. as a strain rate map, is the identification of regions exposed to a high stress gradient, where plaque rupture preferentially occurs, and the detection of various plaque components, such as macrophage infiltration [22–26]. The change in strain over time, i.e. strain rate, emphasizes the mechanical characteristics of the investigated tissue (i.e., the atherosclerotic lesions). To describe the distribution of strain rate within a few mm thick arterial wall, i.e. the change in tissue length during a fraction of the cardiac cycle, requires a spatial resolution in the order or below 1 mm [3]. This spatial resolution (<1 mm) in strain rate distribution also allows the atherosclerotic lesion from that of the adjacent wall to be discerned. The acquisition of ultrasound radio-frequency (RF) signals with high temporal (>200 Hz) and spatial resolution (<1 mm) and post-processing allows displacement detection, and hence, strain rate assessment [27,28]. The temporal resolution of the strain rate depends on the selected temporal window size, which should be a fraction of the systolic upstroke duration. Consequently, the use of a high spatial and temporal resolution ultrasound system in combination with a dedicated ultrasound signal processing algorithm would offer the advantage of revealing non-invasively the underlying mechanics of the early atherosclerotic lesions and adjacent arterial walls Table 1. Moreover, it is known that the plaque’s vulnerability to rupture is directly related to its composition. Clinical studies focused on the relationship between the degree of plaque calcification, fibrous cap and symptomatology, suggested that an advanced atherosclerotic lesion with a lipid core accounting for >40% of the plaque’s total volume covered by an inhomogeneous cap with a thickness of <100 lm is most vulnerable to rupture [29]. Recent investigations indicate that photoacoustic imaging can provide additional information about vascular composition, such as lipid accumulation [30–33]. Photoacoustic imaging combines modern ultrasound and pulsed laser technologies to deliver the high resolution and penetration depth (several cm) of ultrasound imaging with high contrast through optical processes.
Table 1 Qualitative comparison between vascular ultrasound, strain rate and photoacoustic imaging. Imaging technique
Ultrasound
Strain rate
Photoacoustics
Contrast Depth penetration Spatial resolution Frame rate
Anatomy High Excellent Real time
Tissue mechanic High Good Real time
Tissue components Medium Excellent Near real time
3. Novelty The feasibility of integrating conventional ultrasound with elasticity and photoacoustic imaging has been previously demonstrated as a concept for the diagnosis of cancer and its functionality was validated on phantoms [30,34]. These studies suggested that the three imaging modalities can be effectively integrated in a single imaging system by using the same ultrasound transducer and associated electronics. The difference between the previous and the current research, lies within the application area and the validation procedure. A bench-top imaging system combining the three modalities was constructed and optimized for noninvasive vascular imaging. The experiments were conducted ex-vivo using atherosclerotic rabbit arteries. Algorithms to process the RF data to derive the 2D strain rate map and the photoacoustic images were developed. The relationship between the strain rate and photoacoustic signal of two regions with similar echogenicity were directly compared. The obtained results suggest that the three imaging modalities are complementary and their combination may provide an advanced characterization of arterial segments. 4. Approach 4.1. Animal model The experiments were performed using an animal model of atherosclerosis [35]. Specifically, a New Zealand white rabbit was fed chow supplemented with 0.25% cholesterol, a diet promoting development of advanced atherosclerosis in aorta, for 20 months. The rabbit aorta was excised post-euthanasia and preserved in saline damped gauze at 4 °C. The excised artery was 6 cm long and had a lumen diameter of 4–6 mm. Imaging experiments were performed within 24 h after euthanizing the rabbit. The animal study protocol was approved by the Animal Welfare Committee (AWC) of the University of Texas Health Science Center at Houston. 4.2. Data acquisition The core of the imaging setup was a 12 MHz linear array ultrasound transducer (Vermon): 128 elements, 12.5 mm aperture, 0.2 mm pitch, and 60% bandwidth. This ultrasound transducer was selected for its compatibility with two ultrasound imaging systems: Sonix RP (Ultrasonix Medical Corp.) and Cortex (WinProbe Corp.), which allowed the integration of the three imaging modalities. The Sonix RP is a diagnostic ultrasound system with 128 transmit channels and 32 receive channels. The Research Interface of Sonic RP allows for the acquisition of ultrasound pulse-echo RF data at various frame rates; therefore it was selected for ultrasound and strain imaging. The Cortex ultrasound imaging system housed 64 independent transmit/receive channels (each with a pulser, a pre-amplifier, and a 12-bit 40 MHz A/D converter). The Cortex was programmed to allow synchronization with a pulsed laser and to acquire photoacoustic signals. Standard ultrasound B-mode imaging was used to identify a 25 mm region of interest (ROI), centered on the most prominent stenosis. The transducer was positioned 1 cm above the anterior wall, and the acquisition parameters were modified to optimize each imaging mode. For ultrasound imaging the data was captured at 74 Hz frame rate. Each frame was composed of 128 echo lines covering 12.5 mm laterally. For strain rate imaging, the ultrasound RF data were recorded at 296 Hz (32 echo lines covering 12.0 mm laterally) over 5 carotid
I.M. Graf et al. / Ultrasonics 52 (2012) 435–441
437
Fig. 1. (a) Schematic and (b) photo of the setup used for vascular ultrasound, strain rate and photoacoustic imaging, illustrating the position of the atherosclerotic artery and the orientation of the ultrasound transducer and the axis of the laser beam. The artery was mounted in a glass cuvette filled with saline, 1.5 cm away from the wall penetrated by the laser beam. Each end of the arterial sample was secured to a rigid connector attached to the wall of the cuvette. The artery was minimally stretched lengthwise to prevent sagging.
pressure pulse cycles produced by a pulsatile blood pump (Harvard Apparatus), which mimicked a true carotid pressure pulse, operated at a frequency of 1.5 Hz. For photoacoustic imaging, the Cortex ultrasound system was interfaced with the laser beam output from a tunable OPO laser system (Spectra-Physics). To irradiate the ROI of the excised artery, a free-space laser beam was directed perpendicularly to the ultrasonic imaging plane (Fig. 1a). The optical wavelength was varied over 1210–1230 nm, corresponding to the peak absorption of the lipids for the identification of intimal xanthoma [36]. The experiments were performed using a spot size of 1 cm and an average laser fluence of 8 mJ/cm2. Because the laser irradiation spot was smaller than the ultrasound transducer’s aperture the laser beam was laterally shifted 5 mm (50% of the size of irradiation spot) and a second photoacoustic data set was acquired (Fig. 1c). Each photoacoustic frame contained 128 echo lines. The acquisition frame rate was limited to 10 Hz due to the pulse repetition rate of the laser system. The data recorded with each imaging modality using the previously described acquisition procedures covered an arterial length of 12.5 mm. To image 25 mm of the arterial length the transducer was translated twice by 6.25 mm in the lateral direction using a motorized system consisting of a stepper motor and a motion controller (National Instruments); resulting in three overlapping data sets corresponding to three different locations along the artery (Fig. 2).
4.3. Data processing The acquired raw RF data was processed offline to obtain ultrasound, strain rate and photoacoustic images. For ultrasound images post-acquisition processing included noise removal (1–20 MHz bandpass filtering) and Hilbert transformation. The RF
Fig. 2. Diagram of the combination of the acquired data sets to form ultrasound, strain rate and photoacoustic images covering a lateral width of 25 mm, which required (a) for ultrasound images auto-correlation of RF lines to identify lateral overlap, (b) for strain rate images auto-correlation of RF lines at various moments in the cardiac cycle to identify the corresponding cardiac phase and (c) for photoacoustic images averaging of frames acquired in the same position followed by the integration of the three data sets into a single one.
data sets collected at the three different locations of the transducer were combined. The lateral overlap between the RF data sets was identified by auto-correlating RF lines from 2 subsequent data sets. The maximum auto-correlation coefficient was used as a marker of the data set overlap. The RF data from the three measurements were merged into a single data set by averaging the signal of the overlapping regions. Finally, the envelope of the ultrasound signals was calculated and ultrasound gray-scale images were displayed. The location of the walls was manually identified in the first recorded frame. Two regions of similar echogenicity were identified in the ultrasound image to explore the correlation to strain rate and spectroscopic photoacoustic imaging. The raw RF data sets acquired for strain imaging were displayed in M-mode and all cycles with periodical displacement were included in the analysis. Two cycles with out of plane movement created by the blood pump were excluded from analysis. For strain imaging the artery was pulsating; therefore, the overlap of the recorded data was identified by applying auto-correlation in both lateral and temporal direction. The resulting RF data set was then divided into three-dimensional volumes with a size of 0.22 mm in depth by 0.75 mm in the lateral direction by 33 ms in time. Adjacent volumes had 50% overlap in all directions. The axial displacement of each volume (laterally averaged, thus converted to a 2D segment) was calculated using a cross-correlation function R(k,i) based on a model proposed by others [37]:
438
I.M. Graf et al. / Ultrasonics 52 (2012) 435–441
Rðk; iÞ ¼ ðS þ Ni Þ cosð2pfnc ðk ui ÞÞ:
ð1Þ
The cross-correlation function R(k,i) was calculated for each volume with the spatial shift (k) and the temporal shift (i) equal to 0 or 1 to solve the unknown parameters: S (signal power), Ni (noise power), fnc (normalized spatial frequency) and u (dimensionless velocity). Consequently, the equation defining the normalized spatial frequency was:
fnc ¼
1 a cosðRð1; 0Þ=Rð0; 0; ÞÞ; 2p
ð2Þ
which was further used to estimate the dimensionless velocity:
1 0:5ðRð1; 1Þ Rð1; 1ÞÞ : u¼ a tan 2 2pfnc Rð0; 1Þ sinða cosðRð1; 0Þ=Rð0; 0ÞÞÞ
ð3Þ
The dimensionless velocity of volumes axially spaced at 0.45 mm was subtracted and normalized to the spatial kernel dimension (0.22 mm) giving the strain rate expressed in %/s. The temporal variation of strain rate was analyzed and compared to the ultrasound M-mode for identification and removal of any possible motion artifacts. The obtained data set was spline interpolated in the lateral direction with a factor of 4, in order to obtain the same number of sample points in the lateral direction as the ultrasound and photoacoustic images. To reveal the variation in mechanical characteristics throughout the arterial wall the strain rate map and the ultrasound image were overlayed. For photoacoustic imaging, the average of the two data sets acquired at the same ultrasound transducer position was calculated. Following, the three data sets corresponding to the three ultrasound transducer positions were merged into a single set. For spectroscopic photoacoustic imaging the lipid content in the imaged tissue was identified from the multi-wavelength photoacoustic data using an algorithm based on the lipid’s optical absorption spectra [38,39]. Within the 1210–1230 nm wavelength range, the absorption coefficient of lipids decreases nearly linearly in contrast to the relatively constant optical absorption spectrum of waterbased tissue. The envelope-detected amplitude of the photoacoustic signal was spatially low-pass filtered using a moving average filter. Then, at each location, the slope of the photoacoustic signal amplitude from 1210 nm to 1230 nm was calculated. The negative slopes within 30° and 60° were extracted as potential markers of lipid regions [36]. For display purposes, the spectroscopic photoacoustic image was combined with the ultrasound image to reflect the estimated location of lipid pools with respect to the vessel wall anatomy. The selected dynamic range for all images was 40 dB. No estimates were discarded after the data processing.
Fig. 3. Conventional B-mode image of the excised artery centered on a plaque located at both anterior and posterior wall. This and all other images cover area measuring 10 mm in depth and 25 mm laterally.
5. Results The conventional ultrasonic image, presented in Fig. 3, shows the anatomy of the excised artery. The ultrasound B-mode image indicates severe lumen narrowing, a traditional marker of plaque formation (highlighted by the inner yellow and red markers). Additionally, the ultrasound image shows inhomogeneous lateral echogenicity; the posterior wall appears hyperechogenic (fibrotic tissue) on the left side (0–13 mm) and predominantly hypoechogenic on the right side. Regions 1 and 2 were chosen in sections of the posterior and the anterior wall, which presented similar echogenicity (Fig. 3). The 2D map of axial strain rate, presented in Fig. 4a reveals the existence of tissue mechanical inhomogeneities. Note that in the strain rate map the intense red is associated with the maximum strain rate (8%/s) recorded for the artery. The shades of red indicate softer regions of the arterial wall (such as Region 2 in the anterior wall) and the shades of blue–black denote harder tissue. The strain
4.4. Histology studies At the completion of the imaging session, an arterial segment (2 mm wide 3 mm thick 25 mm length) corresponding to the posterior wall of the imaged vessel was selected for histology studies. The arterial sample was fixed in paraformaldhehyde (4% in phosphate-buffered saline). Then the central part of the arterial segment was sectioned through intima, media and adventitia in 5 mm wide 3 mm thick 15 mm long slices. The arterial segments were stained with hemalum and eosin Y (H&E) to detect the anatomical features of the vessel with atherosclerotic lesion [40]. This staining method involves application of hemalum, which colors the nuclei of cells (and a few other objects, such as keratohyalin granules) blue. The nuclear staining is followed by counterstaining with an aqueous or alcoholic solution of eosin Y, which colors eosinophilic structures (generally composed of intracellular or extracellular proteins) in pink.
Fig. 4. (a) Combined ultrasound and strain rate map corresponding to the peak systolic pressure, indicating a stiffer tissue in the center of the posterior wall (Region 1). (b) Temporal variation of the strain rate in Region 2 in the anterior wall.
I.M. Graf et al. / Ultrasonics 52 (2012) 435–441
rate around Region 2 in the anterior wall reached maximum in the intima-media region and then slowly decreased throughout the adventitia (Fig. 4a). The temporal variation of the strain rate in Region 2 of the anterior wall illustrates the compression of the tissue throughout the induced pulsations (Fig. 4b) and it demonstrates the robustness of the technique in displacement tracking as well as the absence of significant motion artifacts. The average strain rate of the anterior wall (4.7%/s) was higher than in the posterior wall (2.3%/s), indicating that the posterior wall was affected by atherosclerosis more than the anterior wall. Additionally the posterior wall presented large inhomogeneities, suggesting a variation in arterial mechanical properties. The strain rate was minimum (0.2%/s) in the central region of the posterior wall (Region 1), encompassing the hypoechogenic area of the ultrasound image (Fig. 3). The low-deformation region extends from the intima to the adventitia layer (Fig. 4a). Globally, the 2D map of axial strain rate illustrates inhomogeneous mechanics along the walls; the standard deviation of the strain rate was 1.36%/s and 1.27%/s in the anterior and posterior wall, respectively, quantitatively indicating large inhomogeneities, associated to the presence of plaques. The photoacoustic responses of the imaged arterial section at 1210 nm (wavelength corresponding to high optical absorption of lipids) and 1230 nm (wavelength corresponding to reduced optical absorption of lipids) are presented in Fig. 5a and b. The amplitude of the photoacoustic signal recorded at 1210 nm was relatively high in the Region 1 of the posterior wall. The photoacoustic signal amplitude detected in Region 1 of the posterior wall quasi-linearly decreased between 1210 and 1230 nm, suggesting the formation of intimal xanthoma in that region (Fig. 5c). The photoacoustic signal amplitude in Region 2 of the anterior wall remained almost constant, which suggested reduced lipid content (Fig. 5c).
439
The spectroscopic photoacoustic image illustrates the most probable location of intimal xanthoma (Fig. 5d). Thus the area highlighted by spectroscopic photoacoustics was a fraction of the area, which produced photoacoustic signals at 1210–1230 nm. Globally, the spectroscopic photoacoustic image suggests that lipids distribution is dominantly larger in the posterior wall (including Region 1) than in the Region 2 within the anterior wall (Fig. 5d). The microscopy image of the H&E stained diseased arterial sample shows interrupted inner elastic lamina and inhomogeneous intimal thickening, confirming the existence of plaques (Fig. 6). 6. Discussion The aim of the present study was to analyze the advantages of integrating ultrasound, strain rate and spectroscopic photoacoustic imaging, targeting superficial large arteries, such as the common carotid artery. The results show that ultrasound, strain rate and spectroscopic photoacoustic imaging of the same arterial segment is feasible. Additionally, the current results demonstrate that the three imaging modalities – ultrasound, strain rate and spectroscopic photoacoustics, are complementary and their combination provides an advanced characterization of arterial segments. Ultrasound images can assist vascular diagnosis by providing information on arterial anatomy such as the presence and the dimension of intimal xanthoma. A noninvasive ultrasound-based 2D map of axial strain rate illustrates tissue mechanical properties, for example it can indicate regions exposed to the highest stress. Spectroscopic photoacoustic imaging provides information about tissue composition such as the location of lipid-rich regions. The first part of our study derives a 2D map of axial strain rate using noninvasive ultrasound imaging. Clinical studies have previously demonstrated that decreased arterial elasticity and
Fig. 5. Combined ultrasound and photoacoustic image obtained at (a) 1210 nm and (b) 1230 nm wavelength. Per optical absorption spectra of lipid [25,26], 1210 nm wavelength corresponds to higher optical absorption of lipids compared to 1230 nm wavelength. (c) Normalized photoacoustic signal amplitude in 1210–1230 nm spectral range from the two regions of the vessel wall. For lipid-rich region (1), the photoacoustic signal quasi-linearly decreased from 1210 nm to 1230 nm, while the water-based arterial region (2) had a relatively constant photoacoustic response. (d) Combined ultrasound and spectroscopic photoacoustic image illustrating the most probable location of lipids.
440
I.M. Graf et al. / Ultrasonics 52 (2012) 435–441
Fig. 6. H&E staining of the arterial wall segment (20X magnification) underlines the inhomogeneous distribution of the plaque.
inhomogeneous arterial mechanical properties are indicative of generalized atherosclerosis [26,41]. The quality of the strain rate imaging could be further improved if the high frame rate (>200 Hz) required for accurate tracking of the arterial displacement could be achieved without decreasing the lateral sampling (i.e., the inter-distance between the echo lines), possible by using ultrafast or parallel beam imaging techniques [42]. Using such an acquisition protocol, the described strain rate method could be implemented to differentiate the variation of tissue mechanics within small regions in the order of hundreds of lm. The experimental conditions used here led to the measurement of significantly lower strain rate values (maximum 8%/s) than in vivo, which are due to the fact that the arterial tissue becomes stiffer postmortem [43–45]. Additionally, the absence of structures outside the arterial wall affected the propagation of the induced mechanical waves. Consequently in vivo studies are expected to provide a more precise assessment of lateral and axial mechanical inhomogeneities and will allow the use of cumulative strain, which is traditionally linked to the arterial elasticity. Previous in vivo studies of vascular diagnosis identified an association between mean arterial mechanics and the degree of atherosclerosis [41,46,47]. In these studies, the mechanical properties (i.e. compliance, distension, etc.) were derived from ultrasound data and averaged over the imaged arterial segments and statistically correlated to other biomarkers of atherosclerosis. Different from a global mechanical parameter, the strain rate analysis could assist in the delineation of a region affected by atherosclerosis and possibly in identifying the severity of the disease. Additionally, the ultrasound and strain rate imaging results were compared to and combined with spectroscopic photoacoustic images of the same arterial segment. For photoacoustic imaging, the artery was irradiated externally with a laser beam perpendicular to the imaging plane thus producing higher laser fluence at the periadventitia. Despite the exponential decline of laser fluence with imaging depth, the highest amplitude of the photoacoustic signals was detected in the thickened intima-media layer of the posterior wall. For in vivo studies the irradiation of the atherosclerotic artery should be parallel to the ultrasound beam, which requires a special imaging probe combining an ultrasound transducer and parallel aligned optical fibers. Previous intravascular ultrasound and multi-wavelength photoacoustic studies demonstrated that other ultrasound transducer and laser beam configurations can also be used to identify lipid-rich regions in atherosclerotic vessels [48]. Considering the absorption properties of lipids in 1210–1230 nm wavelength range, our current results suggest that spectroscopic vascular photoacoustic imaging based on slope analysis may be used to non-invasively identify the lipid-rich regions in atherosclerotic vessels. Quantitative comparisons between lipid-rich regions identified from spectroscopic photoacoustic images and Oil Red O histological sections of the tissue will be evaluated in future studies. Furthermore, the proposed spectroscopic photoacoustic imaging method should be tested on human atherosclerotic vessels
which have the potential to develop vulnerable atherosclerotic plaques as a difference to the New Zealand white rabbit models of atherosclerosis [49,50]. However, the validation of the system using samples excised from animal models of the studied disease is a standard scientific approach. Currently, the most frequently used animal models of atherosclerosis are New Zealand White rabbits, [35] Wantanabe Hereditary Hyperlipedimic (WHHL) rabbits [51] and ApoE deficient mice [52]. For this study we selected a rabbit atherosclerotic aorta, which is of the same order of magnitude as a human common carotid artery. The selection of the rabbit strain was dictated by the fact that we were more successful in developing large atherosclerotic plaques in New Zealand White rabbits, which coped better with long-term high cholesterol diet than the WHHL rabbits. In future studies, the performance of noninvasive spectroscopic photoacoustic imaging to identify lipid accumulation in an artery covered by a thick layer of fatty tissue will be investigated to prove the feasibility of the method for clinical patients [53,54]. In the current setting the laser beam crossed a 1.5 cm layer of water-based solution before reaching the artery. In the 1210–1230 nm range, the optical absorption of lipids is approximately 1.7 times the optical absorption of water-based tissue [36]. Consequently we expect that the proposed noninvasive spectroscopic photoacoustic imaging will be generally applicable to the majority of patients, but the precise limit of depth penetration must still be identified, especially in relation to tissue scattering. In conclusion, the information derived from the combination of the three complementary imaging methods could be used in the future to create a new parameter to assess the risk of plaque vulnerability. We will develop this new tool based on the clinical knowledge of the relationship between arterial characteristics (i.e. plaque’s geometry, mean and local mechanical properties, and volume of lipid accumulation) and risk of vascular events (i.e., transient ischemic attacks, stroke, myocardial infarction, etc.) [3,41,55] and [3,12,38,41,52,53]. Such an advanced risk score might prove to be superior to current estimates of cardiovascular events, such as IMT, which are derived from a more limited arterial characterization (i.e. only tissue morphology). Thus, the proposed system provides a variety of new noninvasive possibilities in diagnosis and monitoring of atherosclerosis. Acknowledgements We would like to thank James Amirian from the University of Texas Health Science Center at Houston for providing the excised rabbit aorta sample. This work was supported in part by the National Institutes of Health under Grant HL096981. References [1] A. Harloff, T. Zech, A. Frydrychowicz, M. Schumacher, J. Schollhorn, J. Hennig, C. Weiller, M. Markl, Carotid intima-media thickness and distensibility measured by MRI at 3 T versus high-resolution ultrasound, Eur. Radiol. 19 (2009) 1470– 1479. [2] T.Z. Naqvi, F. Mendoza, F. Rafii, H. Gransar, M. Guerra, N. Lepor, D.S. Berman, P.K. Shah, High prevalence of ultrasound detected carotid atherosclerosis in subjects with low Framingham risk score: potential implications for screening for subclinical atherosclerosis, J. Am. Soc. Echocardiogr. 23 (2010) 809–815. [3] I.M. Graf, F.H. Schreuder, J.M. Hameleers, W.H. Mess, R.S. Reneman, A.P. Hoeks, Wall irregularity rather than intima-media thickness is associated with nearby atherosclerosis, Ultrasound Med. Biol. 35 (2009) 955–961. [4] F.H. Schreuder, M. Graf, J.M. Hameleers, W.H. Mess, A.P. Hoeks, Measurement of common carotid artery intima-media thickness in clinical practice: comparison of B-mode and RF-based technique, Ultraschall. Med. 30 (2009) 459–465. [5] T.S. Hatsukami, R. Ross, N.L. Polissar, C. Yuan, Visualization of fibrous cap thickness and rupture in human atherosclerotic carotid plaque in vivo with high-resolution magnetic resonance imaging, Circulation 102 (2000) 959–964. [6] M. Hennerici, W. Steinke, C. Klotzsch, T. Els, Flow changes at the carotid bifurcation–an ultrasound study in the human, Vasa Suppl. 32 (1991) 66–71.
I.M. Graf et al. / Ultrasonics 52 (2012) 435–441 [7] J.J. Wentzel, R. Corti, Z.A. Fayad, P. Wisdom, F. Macaluso, M.O. Winkelman, V. Fuster, J.J. Badimon, Does shear stress modulate both plaque progression and regression in the thoracic aorta? Human study using serial magnetic resonance imaging, J. Am. Coll. Cardiol. 45 (2005) 846–854. [8] H.C. Groen, F.J. Gijsen, A. van der Lugt, M.S. Ferguson, T.S. Hatsukami, A.F. van der Steen, C. Yuan, J.J. Wentzel, Plaque rupture in the carotid artery is localized at the high shear stress region: a case report, Stroke 38 (2007) 2379–2381. [9] M. Bardelli, Ultrasonographic investigation of the mechanics of vulnerable atherosclerotic plaques: significance of the volume strain, J. Hypertens. 27 (2009) 219–220. [10] J. Andersson, J. Sundstrom, L. Kurland, T. Gustavsson, J. Hulthe, A. Elmgren, K. Zilmer, M. Zilmer, L. Lind, The carotid artery plaque size and echogenicity are related to different cardiovascular risk factors in the elderly: the Prospective Investigation of the Vasculature in Uppsala Seniors (PIVUS) study, Lipids 44 (2009) 397–403. [11] J. Heliopoulos, K. Vadikolias, P. Mitsias, D. Artemis, G. Tripsianis, C. Piperidou, N. Artemis, A three-dimensional ultrasonographic quantitative analysis of non-ulcerated carotid plaque morphology in symptomatic and asymptomatic carotid stenosis, Atherosclerosis 198 (2008) 129–135. [12] E. Falk, Pathogenesis of atherosclerosis, J. Am. Coll. Cardiol. 47 (2006) C7–12. [13] D. Vela, L.M. Buja, M. Madjid, A. Burke, M. Naghavi, J.T. Willerson, S.W. Casscells, S. Litovsky, The role of periadventitial fat in atherosclerosis, Arch. Pathol. Lab. Med. 131 (2007) 481–487. [14] S.E. Elias-Smale, M. Kavousi, G.C. Verwoert, M.T. Koller, E.W. Steyerberg, F.U. Mattace-Raso, A. Hofman, A.P. Hoeks, R.S. Reneman, J.C. Witteman, Common carotid intima-media thickness in cardiovascular risk stratification of older people: the Rotterdam Study, Eur. J. Cardiovasc. Prev. Rehabil. (2011). [15] W. Xie, L. Liang, L. Zhao, P. Shi, Y. Yang, G. Xie, Y. Huo, Y. Wu, Combination of carotid intima-media thickness and plaque for better predicting risk of ischaemic cardiovascular events, Heart 97 (2011) 1326–1331. [16] J.H. Stein, Carotid intima-media thickness progression and cardiovascular disease risk, J. Am. Coll. Cardiol. 57 (2011) 2291 [author reply 2292]. [17] M. Sareen, K. Waters, A. Nair, D.G. Vince, Normalization and backscatter spectral analysis of human carotid arterial data acquired using a clinical linear array ultrasound imaging system, in: Conf Proc IEEE Eng Med Biol Soc, 2008, pp. 2968–2971. [18] S.K. Kakkos, J.M. Stevens, A.N. Nicolaides, E. Kyriacou, C.S. Pattichis, G. Geroulakos, D. Thomas, Texture analysis of ultrasonic images of symptomatic carotid plaques can identify those plaques associated with ipsilateral embolic brain infarction, Eur. J. Vasc. Endovasc. Surg. 33 (2007) 422–429. [19] S. Yoshida, M.S. Nieminen, T. Paananen, A. Kahri, Can high frequency ultrasound and MRI diagnose malignant atheromatous plaque in vitro?, Jpn Heart J. 36 (1995) 235–245. [20] B.S. Garra, E.I. Cespedes, J. Ophir, S.R. Spratt, R.A. Zuurbier, C.M. Magnant, M.F. Pennanen, Elastography of breast lesions: initial clinical results, Radiology 202 (1997) 79–86. [21] E.C. Godia, R. Madhok, J. Pittman, S. Trocio, R. Ramas, D. Cabral, R.L. Sacco, T. Rundek, Carotid artery distensibility: a reliability study, J. Ultrasound Med. 26 (2007) 1157–1165. [22] L. Zhang, Y. Liu, P.F. Zhang, Y.X. Zhao, X.P. Ji, X.T. Lu, W.Q. Chen, C.X. Liu, C. Zhang, Y. Zhang, Peak radial and circumferential strain measured by velocity vector imaging is a novel index for detecting vulnerable plaques in a rabbit model of atherosclerosis, Atherosclerosis 211 (2011) 146–152. [23] P.F. Zhang, H.J. Su, M. Zhang, J.F. Li, C.X. Liu, S.F. Ding, Y. Miao, L. Chen, X.N. Li, X. Yi, Y. Zhang, Atherosclerotic plaque components characterization and macrophage infiltration identification by intravascular ultrasound elastography based on b-mode analysis: validation in vivo, Int. J. Cardiovasc. Imaging 27 (2011) 39–49. [24] A.P. Hoeks, K.D. Reesink, E. Hermeling, R.S. Reneman, Local blood pressure rather than shear stress should be blamed for plaque rupture, J. Am. Coll. Cardiol. 52 (2008) 1107–1108 [author reply 1108–1109]. [25] J. Folts, Arterial blood pressure gradient across vulnerable plaque might increase rupture, J. Am. Coll. Cardiol. 50 (2007) 2440. [26] H. Ribbers, R.G. Lopata, S. Holewijn, G. Pasterkamp, J.D. Blankensteijn, C.L. de Korte, Noninvasive two-dimensional strain imaging of arteries: validation in phantoms and preliminary experience in carotid arteries in vivo, Ultrasound Med. Biol. 33 (2007) 530–540. [27] A.P. Hoeks, J.M. Meinders, R. Dammers, Applicability and benefit of arterial transfer functions, J. Hypertens. 21 (2003) 1241–1243. [28] T.A. Whittingham, Medical diagnostic applications and sources, Prog. Biophys. Mol. Biol. 93 (2007) 84–110. [29] C.M. Wahlgren, W. Zheng, W. Shaalan, J. Tang, H.S. Bassiouny, Human carotid plaque calcification and vulnerability. Relationship between degree of plaque calcification, fibrous cap inflammatory gene expression and symptomatology, Cerebrovasc. Dis. 27 (2009) 193–200. [30] S.Y. Emelianov, S.R. Aglyamov, A.B. Karpiouk, S. Mallidi, S. Park, S. Sethuraman, J. Shah, R.W. Smalling, J.M. Rubin, W.G. Scott, Synergy and applications of
[31]
[32]
[33] [34]
[35]
[36] [37]
[38] [39] [40]
[41]
[42]
[43]
[44] [45] [46]
[47]
[48]
[49]
[50]
[51]
[52]
[53]
[54] [55]
441
ultrasound, elasticity, and photoacoustic imaging, in: Proceedings of IEEE Ultrasonics Symposium, 2006, pp. 405–415. P.C. Beard, T.N. Mills, Characterization of post mortem arterial tissue using time-resolved photoacoustic spectroscopy at 436, 461 and 532 nm, Phys. Med. Biol. 42 (1997) 177–198. A.A. Oraevsky, S.L. Jacques, G.H. Pettit, R.A. Sauerbrey, F.K. Tittel, J.H. Nguy, P.D. Henry, XeCl laser-induced fluorescence of atherosclerotic arteries. Spectral similarities between lipid-rich lesions and peroxidized lipoproteins, Circ. Res. 72 (1993) 84–90. T.J. Allen, P.C. Beard, A. Hall, A. Dhillon, J.S. Owen, Photoacoustic imaging of lipid rich plaques in human aorta, in: SPIE, 2010, pp. 75640C–75641C. S. Park, J. Shah, S.R. Aglyamov, A.B. Karpiouk, S. Mallidi, A. Gopal, H. Moon, X.J. Zhang, W.G. Scott, S.Y. Emelianov, Integrated system for ultrasonic, photoacoustic and elasticity imaging, in: S. Emelianov, W.F. Walker (Eds.) Proceedings of SPIE, 2006, pp. 61470H. F.D. Kolodgie, A.S. Katocs Jr., E.E. Largis, S.M. Wrenn, J.F. Cornhill, E.E. Herderick, S.J. Lee, R. Virmani, Hypercholesterolemia in the rabbit induced by feeding graded amounts of low-level cholesterol. Methodological considerations regarding individual variability in response to dietary cholesterol and development of lesion type, Arterioscler. Thromb. Vasc. Biol. 16 (1996) 1454–1464. O. Header, Opt. Absorp. Spectra (2001). P.J. Brands, A.P. Hoeks, L.A. Ledoux, R.S. Reneman, A radio frequency domain complex cross-correlation model to estimate blood flow velocity and tissue motion by means of ultrasound, Ultrasound Med. Biol. 23 (1997) 911–920. S.A. Prahl, Opt. Prop. Spectra (2001). C.L. Tsai, J.C. Chen, W.J. Wang, Near-infrared absorption property of biological soft tissue constituents, J. Med. Biol. Eng. 21 (2001) 7–14. E.M. Rubin, A.A. Martin, S.N. Thung, M.A. Gerber, Morphometric and immunohistochemical characterization of human liver regeneration, Am. J. Pathol. 147 (1995) 397–404. A. Paini, P. Boutouyrie, D. Calvet, M. Zidi, E. Agabiti-Rosei, S. Laurent, Multiaxial mechanical characteristics of carotid plaque: analysis by multiarray echotracking system, Stroke 38 (2007) 117–123. S. Park, S.R. Aglyamov, W.G. Scott, S.Y. Emelianov, Strain imaging using conventional and ultrafast ultrasound imaging: numerical analysis, IEEE Trans. Ultrason. Ferroelectr. Freq. Control 54 (2007) 987–995. S.L. Wilens, The postmortem elasticity of the adult human aorta. its relation to age and to the distribution of intimal atheromas, Am. J. Pathol. 13 (1937) 811– 834. B.S. Gow, C.D. Hadfield, The elasticity of canine and human coronary arteries with reference to postmortem changes, Circ. Res. 45 (1979) 588–594. R.T. Yen, S.S. Sobin, Elasticity of arterioles and venules in postmortem human lungs, J. Appl. Physiol. 64 (1988) 611–619. I.M. Graf, F.H. Schreuder, W.H. Mess, R.S. Reneman, A.P. Hoeks, Spatial distension variations are associated with focal atherosclerotic plaques, Cerebrovasc. Dis. 29 (2009) 199–205. T. Montalcini, G. Gorgone, C. Gazzaruso, A. Pujia, Carotid distension and distensibility impairment in individuals affected by familial combined hyperlipidemia, Atherosclerosis 212 (2010) 177–180. B. Wang, J.L. Su, J. Amirian, S.H. Litovsky, R. Smalling, S. Emelianov, Detection of lipid in atherosclerotic vessels using ultrasound-guided spectroscopic intravascular photoacoustic imaging, Opt. Express 18 (2010) 4889–4897. X. Niu, V. Zammit, J.M. Upston, R.T. Dean, R. Stocker, Coexistence of oxidized lipids and alpha-tocopherol in all lipoprotein density fractions isolated from advanced human atherosclerotic plaques, Arterioscler. Thromb. Vasc. Biol. 19 (1999) 1708–1718. J.H. Rapp, W.E. Connor, D.S. Lin, T. Inahara, J.M. Porter, Lipids of human atherosclerotic plaques and xanthomas: clues to the mechanism of plaque progression, J. Lipid Res. 24 (1983) 1329–1335. S.P. Roux, H. Kuhn, H. Lengsfeld, O.H. Morand, Effects of chronic aortic coarctation on atherosclerosis and arterial lipid accumulation in the Watanabe hereditary hyperlipidemic (WHHL) rabbit, Atherosclerosis 93 (1992) 123–132. Z. Ma, A. Choudhury, S.A. Kang, M. Monestier, P.L. Cohen, R.A. Eisenberg, Accelerated atherosclerosis in ApoE deficient lupus mouse models, Clin. Immunol. 127 (2008) 168–175. P. Babusik, I. Duris, Comparison of obesity and its relationship to some metabolic risk factors of atherosclerosis in Arabs and South Asians in Kuwait, Med. Princ. Pract. 19 (2010) 275–280. P. Sucharda, Obesity and atherosclerosis – what’s the link? Vnitr. Lek. vol. 56, 2010, pp. 289–291. J. Cai, T.S. Hatsukami, M.S. Ferguson, W.S. Kerwin, T. Saam, B. Chu, N. Takaya, N.L. Polissar, C. Yuan, In vivo quantitative measurement of intact fibrous cap and lipid-rich necrotic core size in atherosclerotic carotid plaque: comparison of high-resolution, contrast-enhanced magnetic resonance imaging and histology, Circulation 112 (2005) 3437–3444.
Contents lists available at SciVerse ScienceDirect
Ultrasonics journal homepage: www.elsevier.com/locate/ultras
Noninvasive detection of intimal xanthoma using combined ultrasound, strain rate and photoacoustic imaging Iulia M. Graf a, Seungsoo Kim a, Bo Wang a, Richard Smalling b, Stanislav Emelianov a,⇑ a b
Department of Biomedical Engineering, University of Texas at Austin, Austin, TX 78712, USA Division of Cardiology, University of Texas Health Science Center, Houston, TX 77030, USA
a r t i c l e
i n f o
Article history: Received 30 March 2011 Received in revised form 29 September 2011 Accepted 10 October 2011 Available online 18 October 2011 Keywords: Ultrasound Strain Photoacoustics Intimal xanthoma Atherosclerosis
a b s t r a c t Background and motivation: The structure, composition and mechanics of carotid artery are good indicators of early progressive atherosclerotic lesions. The combination of three imaging modalities (ultrasound, strain rate and photoacoustic imaging) which could provide corroborative information about the named arterial properties could enhance the characterization of intimal xanthoma. Methods: The experiments were performed using a New Zealand white rabbit model of atherosclerosis. The aorta excised from an atherosclerotic rabbit was scanned ex vivo using the three imaging techniques: (1) ultrasound imaging of the longitudinal section: standard ultrasound B-mode (74 Hz frame rate); (2) strain rate imaging: the artery was flushed with blood and a 1.5 Hz physiologic pulsation was induced, while the ultrasound data were recorded at higher frame rate (296 Hz); (3) photoacoustic imaging: the artery was irradiated with nanosecond pulsed laser light of low fluence in the 1210–1230 nm wavelength range and the photoacoustic data was recorded at 10 Hz frame rate. Post processing algorithms based on cross-correlation and optical absorption variation were implemented to derive strain rate and spectroscopic photoacoustic images, respectively. Results: Based on the spatio-temporal variation in displacement of different regions within the arterial wall, strain rate imaging reveals differences in tissue mechanical properties. Additionally, spectroscopic photoacoustic imaging can spatially resolve the optical absorption properties of arterial tissue and identify the location of lipid pools. Conclusions: The study demonstrates that ultrasound, strain rate and photoacoustic imaging can be used to simultaneously evaluate the structure, the mechanics and the composition of atherosclerotic lesions to improve the assessment of plaque vulnerability. Ó 2011 Elsevier B.V. All rights reserved.
1. Objective The objective of the study was to explore the relationship between arterial morphology, mechanics and composition by using combined vascular ultrasound, strain rate and photoacoustic imaging and to identify the contribution of each arterial characteristic in the characterization of atherosclerosis. 2. Motivation The etiology of atherosclerosis and its clinical sequelae have been the object of many research studies in the past years. Various markers (i.e. intima media thickness [1], Framingham score [2], stenosis score [3], etc.) and noninvasive imaging techniques (i.e. ultrasound [4] and MRI [5]) are currently used in clinical practice to diagnose atherosclerosis. The difficulty of assessing early ⇑ Corresponding author. E-mail address: [email protected] (S. Emelianov). 0041-624X/$ - see front matter Ó 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.ultras.2011.10.005
atherosclerotic lesions with respect to plaque vulnerability is related to the complexity of the disease. Thus a concomitant analysis of the artery wall’s structure or anatomy, tissue mechanical properties and composition could lead to a superior evaluation of early atherosclerosis, such as intimal xanthoma in relation with vessel remodeling and inflammation in response to atherosclerosis [6–13]. Compared to other approaches (i.e. MRI or CT), ultrasonography is attractive because of its non-invasive character, its portability and associated low-cost. Gray-scale B-mode ultrasound imaging is commonly used in clinical practice to assess tissue morphology and cardiovascular risk estimators, such as the Intima-Media Thickness (IMT) [3,14–16]. The structure of the arterial walls and early atherosclerotic lesions can be partly characterized by the amplitude of ultrasound echo signals in combination with observed tissue specific attenuation [17,18]. Regions with high calcium content exhibit hyperechogenicity, while the lipid core exhibits a low echogenicity [19]. In the case of atherosclerotic lesions with high calcium content, especially the ones attached both to the anterior and posterior walls, the tissue beneath dense
436
I.M. Graf et al. / Ultrasonics 52 (2012) 435–441
calcifications lies within a shaded region, which makes the tissue differentiation very difficult. An additional imaging modality is needed to improve the assessment of plaque’s risk of rupture. A possible way to enhance the characterization of atherosclerotic lesions would be to supplement the information on morphology, as derived from ultrasound, with other arterial characteristic, such as mechanics and composition. Elastography is a medical imaging tool which complements ultrasonography by providing information on tissue mechanical properties [20]. Current clinically available tools of arterial mechanical properties, which are based on noninvasive ultrasound imaging, require the calculation of functional parameters, such as distension, distensibility, compliance, etc. [1,21]. These parameters are generally derived from the averaged displacement of a few centimeters along arterial segment. Thus they do not offer any information on the variation of mechanical characteristics throughout the arterial wall (neither in the axial nor in the lateral direction). The advantage of deriving the mechanical characteristics of the arterial segments two-dimensionally (2D), i.e. as a strain rate map, is the identification of regions exposed to a high stress gradient, where plaque rupture preferentially occurs, and the detection of various plaque components, such as macrophage infiltration [22–26]. The change in strain over time, i.e. strain rate, emphasizes the mechanical characteristics of the investigated tissue (i.e., the atherosclerotic lesions). To describe the distribution of strain rate within a few mm thick arterial wall, i.e. the change in tissue length during a fraction of the cardiac cycle, requires a spatial resolution in the order or below 1 mm [3]. This spatial resolution (<1 mm) in strain rate distribution also allows the atherosclerotic lesion from that of the adjacent wall to be discerned. The acquisition of ultrasound radio-frequency (RF) signals with high temporal (>200 Hz) and spatial resolution (<1 mm) and post-processing allows displacement detection, and hence, strain rate assessment [27,28]. The temporal resolution of the strain rate depends on the selected temporal window size, which should be a fraction of the systolic upstroke duration. Consequently, the use of a high spatial and temporal resolution ultrasound system in combination with a dedicated ultrasound signal processing algorithm would offer the advantage of revealing non-invasively the underlying mechanics of the early atherosclerotic lesions and adjacent arterial walls Table 1. Moreover, it is known that the plaque’s vulnerability to rupture is directly related to its composition. Clinical studies focused on the relationship between the degree of plaque calcification, fibrous cap and symptomatology, suggested that an advanced atherosclerotic lesion with a lipid core accounting for >40% of the plaque’s total volume covered by an inhomogeneous cap with a thickness of <100 lm is most vulnerable to rupture [29]. Recent investigations indicate that photoacoustic imaging can provide additional information about vascular composition, such as lipid accumulation [30–33]. Photoacoustic imaging combines modern ultrasound and pulsed laser technologies to deliver the high resolution and penetration depth (several cm) of ultrasound imaging with high contrast through optical processes.
Table 1 Qualitative comparison between vascular ultrasound, strain rate and photoacoustic imaging. Imaging technique
Ultrasound
Strain rate
Photoacoustics
Contrast Depth penetration Spatial resolution Frame rate
Anatomy High Excellent Real time
Tissue mechanic High Good Real time
Tissue components Medium Excellent Near real time
3. Novelty The feasibility of integrating conventional ultrasound with elasticity and photoacoustic imaging has been previously demonstrated as a concept for the diagnosis of cancer and its functionality was validated on phantoms [30,34]. These studies suggested that the three imaging modalities can be effectively integrated in a single imaging system by using the same ultrasound transducer and associated electronics. The difference between the previous and the current research, lies within the application area and the validation procedure. A bench-top imaging system combining the three modalities was constructed and optimized for noninvasive vascular imaging. The experiments were conducted ex-vivo using atherosclerotic rabbit arteries. Algorithms to process the RF data to derive the 2D strain rate map and the photoacoustic images were developed. The relationship between the strain rate and photoacoustic signal of two regions with similar echogenicity were directly compared. The obtained results suggest that the three imaging modalities are complementary and their combination may provide an advanced characterization of arterial segments. 4. Approach 4.1. Animal model The experiments were performed using an animal model of atherosclerosis [35]. Specifically, a New Zealand white rabbit was fed chow supplemented with 0.25% cholesterol, a diet promoting development of advanced atherosclerosis in aorta, for 20 months. The rabbit aorta was excised post-euthanasia and preserved in saline damped gauze at 4 °C. The excised artery was 6 cm long and had a lumen diameter of 4–6 mm. Imaging experiments were performed within 24 h after euthanizing the rabbit. The animal study protocol was approved by the Animal Welfare Committee (AWC) of the University of Texas Health Science Center at Houston. 4.2. Data acquisition The core of the imaging setup was a 12 MHz linear array ultrasound transducer (Vermon): 128 elements, 12.5 mm aperture, 0.2 mm pitch, and 60% bandwidth. This ultrasound transducer was selected for its compatibility with two ultrasound imaging systems: Sonix RP (Ultrasonix Medical Corp.) and Cortex (WinProbe Corp.), which allowed the integration of the three imaging modalities. The Sonix RP is a diagnostic ultrasound system with 128 transmit channels and 32 receive channels. The Research Interface of Sonic RP allows for the acquisition of ultrasound pulse-echo RF data at various frame rates; therefore it was selected for ultrasound and strain imaging. The Cortex ultrasound imaging system housed 64 independent transmit/receive channels (each with a pulser, a pre-amplifier, and a 12-bit 40 MHz A/D converter). The Cortex was programmed to allow synchronization with a pulsed laser and to acquire photoacoustic signals. Standard ultrasound B-mode imaging was used to identify a 25 mm region of interest (ROI), centered on the most prominent stenosis. The transducer was positioned 1 cm above the anterior wall, and the acquisition parameters were modified to optimize each imaging mode. For ultrasound imaging the data was captured at 74 Hz frame rate. Each frame was composed of 128 echo lines covering 12.5 mm laterally. For strain rate imaging, the ultrasound RF data were recorded at 296 Hz (32 echo lines covering 12.0 mm laterally) over 5 carotid
I.M. Graf et al. / Ultrasonics 52 (2012) 435–441
437
Fig. 1. (a) Schematic and (b) photo of the setup used for vascular ultrasound, strain rate and photoacoustic imaging, illustrating the position of the atherosclerotic artery and the orientation of the ultrasound transducer and the axis of the laser beam. The artery was mounted in a glass cuvette filled with saline, 1.5 cm away from the wall penetrated by the laser beam. Each end of the arterial sample was secured to a rigid connector attached to the wall of the cuvette. The artery was minimally stretched lengthwise to prevent sagging.
pressure pulse cycles produced by a pulsatile blood pump (Harvard Apparatus), which mimicked a true carotid pressure pulse, operated at a frequency of 1.5 Hz. For photoacoustic imaging, the Cortex ultrasound system was interfaced with the laser beam output from a tunable OPO laser system (Spectra-Physics). To irradiate the ROI of the excised artery, a free-space laser beam was directed perpendicularly to the ultrasonic imaging plane (Fig. 1a). The optical wavelength was varied over 1210–1230 nm, corresponding to the peak absorption of the lipids for the identification of intimal xanthoma [36]. The experiments were performed using a spot size of 1 cm and an average laser fluence of 8 mJ/cm2. Because the laser irradiation spot was smaller than the ultrasound transducer’s aperture the laser beam was laterally shifted 5 mm (50% of the size of irradiation spot) and a second photoacoustic data set was acquired (Fig. 1c). Each photoacoustic frame contained 128 echo lines. The acquisition frame rate was limited to 10 Hz due to the pulse repetition rate of the laser system. The data recorded with each imaging modality using the previously described acquisition procedures covered an arterial length of 12.5 mm. To image 25 mm of the arterial length the transducer was translated twice by 6.25 mm in the lateral direction using a motorized system consisting of a stepper motor and a motion controller (National Instruments); resulting in three overlapping data sets corresponding to three different locations along the artery (Fig. 2).
4.3. Data processing The acquired raw RF data was processed offline to obtain ultrasound, strain rate and photoacoustic images. For ultrasound images post-acquisition processing included noise removal (1–20 MHz bandpass filtering) and Hilbert transformation. The RF
Fig. 2. Diagram of the combination of the acquired data sets to form ultrasound, strain rate and photoacoustic images covering a lateral width of 25 mm, which required (a) for ultrasound images auto-correlation of RF lines to identify lateral overlap, (b) for strain rate images auto-correlation of RF lines at various moments in the cardiac cycle to identify the corresponding cardiac phase and (c) for photoacoustic images averaging of frames acquired in the same position followed by the integration of the three data sets into a single one.
data sets collected at the three different locations of the transducer were combined. The lateral overlap between the RF data sets was identified by auto-correlating RF lines from 2 subsequent data sets. The maximum auto-correlation coefficient was used as a marker of the data set overlap. The RF data from the three measurements were merged into a single data set by averaging the signal of the overlapping regions. Finally, the envelope of the ultrasound signals was calculated and ultrasound gray-scale images were displayed. The location of the walls was manually identified in the first recorded frame. Two regions of similar echogenicity were identified in the ultrasound image to explore the correlation to strain rate and spectroscopic photoacoustic imaging. The raw RF data sets acquired for strain imaging were displayed in M-mode and all cycles with periodical displacement were included in the analysis. Two cycles with out of plane movement created by the blood pump were excluded from analysis. For strain imaging the artery was pulsating; therefore, the overlap of the recorded data was identified by applying auto-correlation in both lateral and temporal direction. The resulting RF data set was then divided into three-dimensional volumes with a size of 0.22 mm in depth by 0.75 mm in the lateral direction by 33 ms in time. Adjacent volumes had 50% overlap in all directions. The axial displacement of each volume (laterally averaged, thus converted to a 2D segment) was calculated using a cross-correlation function R(k,i) based on a model proposed by others [37]:
438
I.M. Graf et al. / Ultrasonics 52 (2012) 435–441
Rðk; iÞ ¼ ðS þ Ni Þ cosð2pfnc ðk ui ÞÞ:
ð1Þ
The cross-correlation function R(k,i) was calculated for each volume with the spatial shift (k) and the temporal shift (i) equal to 0 or 1 to solve the unknown parameters: S (signal power), Ni (noise power), fnc (normalized spatial frequency) and u (dimensionless velocity). Consequently, the equation defining the normalized spatial frequency was:
fnc ¼
1 a cosðRð1; 0Þ=Rð0; 0; ÞÞ; 2p
ð2Þ
which was further used to estimate the dimensionless velocity:
1 0:5ðRð1; 1Þ Rð1; 1ÞÞ : u¼ a tan 2 2pfnc Rð0; 1Þ sinða cosðRð1; 0Þ=Rð0; 0ÞÞÞ
ð3Þ
The dimensionless velocity of volumes axially spaced at 0.45 mm was subtracted and normalized to the spatial kernel dimension (0.22 mm) giving the strain rate expressed in %/s. The temporal variation of strain rate was analyzed and compared to the ultrasound M-mode for identification and removal of any possible motion artifacts. The obtained data set was spline interpolated in the lateral direction with a factor of 4, in order to obtain the same number of sample points in the lateral direction as the ultrasound and photoacoustic images. To reveal the variation in mechanical characteristics throughout the arterial wall the strain rate map and the ultrasound image were overlayed. For photoacoustic imaging, the average of the two data sets acquired at the same ultrasound transducer position was calculated. Following, the three data sets corresponding to the three ultrasound transducer positions were merged into a single set. For spectroscopic photoacoustic imaging the lipid content in the imaged tissue was identified from the multi-wavelength photoacoustic data using an algorithm based on the lipid’s optical absorption spectra [38,39]. Within the 1210–1230 nm wavelength range, the absorption coefficient of lipids decreases nearly linearly in contrast to the relatively constant optical absorption spectrum of waterbased tissue. The envelope-detected amplitude of the photoacoustic signal was spatially low-pass filtered using a moving average filter. Then, at each location, the slope of the photoacoustic signal amplitude from 1210 nm to 1230 nm was calculated. The negative slopes within 30° and 60° were extracted as potential markers of lipid regions [36]. For display purposes, the spectroscopic photoacoustic image was combined with the ultrasound image to reflect the estimated location of lipid pools with respect to the vessel wall anatomy. The selected dynamic range for all images was 40 dB. No estimates were discarded after the data processing.
Fig. 3. Conventional B-mode image of the excised artery centered on a plaque located at both anterior and posterior wall. This and all other images cover area measuring 10 mm in depth and 25 mm laterally.
5. Results The conventional ultrasonic image, presented in Fig. 3, shows the anatomy of the excised artery. The ultrasound B-mode image indicates severe lumen narrowing, a traditional marker of plaque formation (highlighted by the inner yellow and red markers). Additionally, the ultrasound image shows inhomogeneous lateral echogenicity; the posterior wall appears hyperechogenic (fibrotic tissue) on the left side (0–13 mm) and predominantly hypoechogenic on the right side. Regions 1 and 2 were chosen in sections of the posterior and the anterior wall, which presented similar echogenicity (Fig. 3). The 2D map of axial strain rate, presented in Fig. 4a reveals the existence of tissue mechanical inhomogeneities. Note that in the strain rate map the intense red is associated with the maximum strain rate (8%/s) recorded for the artery. The shades of red indicate softer regions of the arterial wall (such as Region 2 in the anterior wall) and the shades of blue–black denote harder tissue. The strain
4.4. Histology studies At the completion of the imaging session, an arterial segment (2 mm wide 3 mm thick 25 mm length) corresponding to the posterior wall of the imaged vessel was selected for histology studies. The arterial sample was fixed in paraformaldhehyde (4% in phosphate-buffered saline). Then the central part of the arterial segment was sectioned through intima, media and adventitia in 5 mm wide 3 mm thick 15 mm long slices. The arterial segments were stained with hemalum and eosin Y (H&E) to detect the anatomical features of the vessel with atherosclerotic lesion [40]. This staining method involves application of hemalum, which colors the nuclei of cells (and a few other objects, such as keratohyalin granules) blue. The nuclear staining is followed by counterstaining with an aqueous or alcoholic solution of eosin Y, which colors eosinophilic structures (generally composed of intracellular or extracellular proteins) in pink.
Fig. 4. (a) Combined ultrasound and strain rate map corresponding to the peak systolic pressure, indicating a stiffer tissue in the center of the posterior wall (Region 1). (b) Temporal variation of the strain rate in Region 2 in the anterior wall.
I.M. Graf et al. / Ultrasonics 52 (2012) 435–441
rate around Region 2 in the anterior wall reached maximum in the intima-media region and then slowly decreased throughout the adventitia (Fig. 4a). The temporal variation of the strain rate in Region 2 of the anterior wall illustrates the compression of the tissue throughout the induced pulsations (Fig. 4b) and it demonstrates the robustness of the technique in displacement tracking as well as the absence of significant motion artifacts. The average strain rate of the anterior wall (4.7%/s) was higher than in the posterior wall (2.3%/s), indicating that the posterior wall was affected by atherosclerosis more than the anterior wall. Additionally the posterior wall presented large inhomogeneities, suggesting a variation in arterial mechanical properties. The strain rate was minimum (0.2%/s) in the central region of the posterior wall (Region 1), encompassing the hypoechogenic area of the ultrasound image (Fig. 3). The low-deformation region extends from the intima to the adventitia layer (Fig. 4a). Globally, the 2D map of axial strain rate illustrates inhomogeneous mechanics along the walls; the standard deviation of the strain rate was 1.36%/s and 1.27%/s in the anterior and posterior wall, respectively, quantitatively indicating large inhomogeneities, associated to the presence of plaques. The photoacoustic responses of the imaged arterial section at 1210 nm (wavelength corresponding to high optical absorption of lipids) and 1230 nm (wavelength corresponding to reduced optical absorption of lipids) are presented in Fig. 5a and b. The amplitude of the photoacoustic signal recorded at 1210 nm was relatively high in the Region 1 of the posterior wall. The photoacoustic signal amplitude detected in Region 1 of the posterior wall quasi-linearly decreased between 1210 and 1230 nm, suggesting the formation of intimal xanthoma in that region (Fig. 5c). The photoacoustic signal amplitude in Region 2 of the anterior wall remained almost constant, which suggested reduced lipid content (Fig. 5c).
439
The spectroscopic photoacoustic image illustrates the most probable location of intimal xanthoma (Fig. 5d). Thus the area highlighted by spectroscopic photoacoustics was a fraction of the area, which produced photoacoustic signals at 1210–1230 nm. Globally, the spectroscopic photoacoustic image suggests that lipids distribution is dominantly larger in the posterior wall (including Region 1) than in the Region 2 within the anterior wall (Fig. 5d). The microscopy image of the H&E stained diseased arterial sample shows interrupted inner elastic lamina and inhomogeneous intimal thickening, confirming the existence of plaques (Fig. 6). 6. Discussion The aim of the present study was to analyze the advantages of integrating ultrasound, strain rate and spectroscopic photoacoustic imaging, targeting superficial large arteries, such as the common carotid artery. The results show that ultrasound, strain rate and spectroscopic photoacoustic imaging of the same arterial segment is feasible. Additionally, the current results demonstrate that the three imaging modalities – ultrasound, strain rate and spectroscopic photoacoustics, are complementary and their combination provides an advanced characterization of arterial segments. Ultrasound images can assist vascular diagnosis by providing information on arterial anatomy such as the presence and the dimension of intimal xanthoma. A noninvasive ultrasound-based 2D map of axial strain rate illustrates tissue mechanical properties, for example it can indicate regions exposed to the highest stress. Spectroscopic photoacoustic imaging provides information about tissue composition such as the location of lipid-rich regions. The first part of our study derives a 2D map of axial strain rate using noninvasive ultrasound imaging. Clinical studies have previously demonstrated that decreased arterial elasticity and
Fig. 5. Combined ultrasound and photoacoustic image obtained at (a) 1210 nm and (b) 1230 nm wavelength. Per optical absorption spectra of lipid [25,26], 1210 nm wavelength corresponds to higher optical absorption of lipids compared to 1230 nm wavelength. (c) Normalized photoacoustic signal amplitude in 1210–1230 nm spectral range from the two regions of the vessel wall. For lipid-rich region (1), the photoacoustic signal quasi-linearly decreased from 1210 nm to 1230 nm, while the water-based arterial region (2) had a relatively constant photoacoustic response. (d) Combined ultrasound and spectroscopic photoacoustic image illustrating the most probable location of lipids.
440
I.M. Graf et al. / Ultrasonics 52 (2012) 435–441
Fig. 6. H&E staining of the arterial wall segment (20X magnification) underlines the inhomogeneous distribution of the plaque.
inhomogeneous arterial mechanical properties are indicative of generalized atherosclerosis [26,41]. The quality of the strain rate imaging could be further improved if the high frame rate (>200 Hz) required for accurate tracking of the arterial displacement could be achieved without decreasing the lateral sampling (i.e., the inter-distance between the echo lines), possible by using ultrafast or parallel beam imaging techniques [42]. Using such an acquisition protocol, the described strain rate method could be implemented to differentiate the variation of tissue mechanics within small regions in the order of hundreds of lm. The experimental conditions used here led to the measurement of significantly lower strain rate values (maximum 8%/s) than in vivo, which are due to the fact that the arterial tissue becomes stiffer postmortem [43–45]. Additionally, the absence of structures outside the arterial wall affected the propagation of the induced mechanical waves. Consequently in vivo studies are expected to provide a more precise assessment of lateral and axial mechanical inhomogeneities and will allow the use of cumulative strain, which is traditionally linked to the arterial elasticity. Previous in vivo studies of vascular diagnosis identified an association between mean arterial mechanics and the degree of atherosclerosis [41,46,47]. In these studies, the mechanical properties (i.e. compliance, distension, etc.) were derived from ultrasound data and averaged over the imaged arterial segments and statistically correlated to other biomarkers of atherosclerosis. Different from a global mechanical parameter, the strain rate analysis could assist in the delineation of a region affected by atherosclerosis and possibly in identifying the severity of the disease. Additionally, the ultrasound and strain rate imaging results were compared to and combined with spectroscopic photoacoustic images of the same arterial segment. For photoacoustic imaging, the artery was irradiated externally with a laser beam perpendicular to the imaging plane thus producing higher laser fluence at the periadventitia. Despite the exponential decline of laser fluence with imaging depth, the highest amplitude of the photoacoustic signals was detected in the thickened intima-media layer of the posterior wall. For in vivo studies the irradiation of the atherosclerotic artery should be parallel to the ultrasound beam, which requires a special imaging probe combining an ultrasound transducer and parallel aligned optical fibers. Previous intravascular ultrasound and multi-wavelength photoacoustic studies demonstrated that other ultrasound transducer and laser beam configurations can also be used to identify lipid-rich regions in atherosclerotic vessels [48]. Considering the absorption properties of lipids in 1210–1230 nm wavelength range, our current results suggest that spectroscopic vascular photoacoustic imaging based on slope analysis may be used to non-invasively identify the lipid-rich regions in atherosclerotic vessels. Quantitative comparisons between lipid-rich regions identified from spectroscopic photoacoustic images and Oil Red O histological sections of the tissue will be evaluated in future studies. Furthermore, the proposed spectroscopic photoacoustic imaging method should be tested on human atherosclerotic vessels
which have the potential to develop vulnerable atherosclerotic plaques as a difference to the New Zealand white rabbit models of atherosclerosis [49,50]. However, the validation of the system using samples excised from animal models of the studied disease is a standard scientific approach. Currently, the most frequently used animal models of atherosclerosis are New Zealand White rabbits, [35] Wantanabe Hereditary Hyperlipedimic (WHHL) rabbits [51] and ApoE deficient mice [52]. For this study we selected a rabbit atherosclerotic aorta, which is of the same order of magnitude as a human common carotid artery. The selection of the rabbit strain was dictated by the fact that we were more successful in developing large atherosclerotic plaques in New Zealand White rabbits, which coped better with long-term high cholesterol diet than the WHHL rabbits. In future studies, the performance of noninvasive spectroscopic photoacoustic imaging to identify lipid accumulation in an artery covered by a thick layer of fatty tissue will be investigated to prove the feasibility of the method for clinical patients [53,54]. In the current setting the laser beam crossed a 1.5 cm layer of water-based solution before reaching the artery. In the 1210–1230 nm range, the optical absorption of lipids is approximately 1.7 times the optical absorption of water-based tissue [36]. Consequently we expect that the proposed noninvasive spectroscopic photoacoustic imaging will be generally applicable to the majority of patients, but the precise limit of depth penetration must still be identified, especially in relation to tissue scattering. In conclusion, the information derived from the combination of the three complementary imaging methods could be used in the future to create a new parameter to assess the risk of plaque vulnerability. We will develop this new tool based on the clinical knowledge of the relationship between arterial characteristics (i.e. plaque’s geometry, mean and local mechanical properties, and volume of lipid accumulation) and risk of vascular events (i.e., transient ischemic attacks, stroke, myocardial infarction, etc.) [3,41,55] and [3,12,38,41,52,53]. Such an advanced risk score might prove to be superior to current estimates of cardiovascular events, such as IMT, which are derived from a more limited arterial characterization (i.e. only tissue morphology). Thus, the proposed system provides a variety of new noninvasive possibilities in diagnosis and monitoring of atherosclerosis. Acknowledgements We would like to thank James Amirian from the University of Texas Health Science Center at Houston for providing the excised rabbit aorta sample. This work was supported in part by the National Institutes of Health under Grant HL096981. References [1] A. Harloff, T. Zech, A. Frydrychowicz, M. Schumacher, J. Schollhorn, J. Hennig, C. Weiller, M. Markl, Carotid intima-media thickness and distensibility measured by MRI at 3 T versus high-resolution ultrasound, Eur. Radiol. 19 (2009) 1470– 1479. [2] T.Z. Naqvi, F. Mendoza, F. Rafii, H. Gransar, M. Guerra, N. Lepor, D.S. Berman, P.K. Shah, High prevalence of ultrasound detected carotid atherosclerosis in subjects with low Framingham risk score: potential implications for screening for subclinical atherosclerosis, J. Am. Soc. Echocardiogr. 23 (2010) 809–815. [3] I.M. Graf, F.H. Schreuder, J.M. Hameleers, W.H. Mess, R.S. Reneman, A.P. Hoeks, Wall irregularity rather than intima-media thickness is associated with nearby atherosclerosis, Ultrasound Med. Biol. 35 (2009) 955–961. [4] F.H. Schreuder, M. Graf, J.M. Hameleers, W.H. Mess, A.P. Hoeks, Measurement of common carotid artery intima-media thickness in clinical practice: comparison of B-mode and RF-based technique, Ultraschall. Med. 30 (2009) 459–465. [5] T.S. Hatsukami, R. Ross, N.L. Polissar, C. Yuan, Visualization of fibrous cap thickness and rupture in human atherosclerotic carotid plaque in vivo with high-resolution magnetic resonance imaging, Circulation 102 (2000) 959–964. [6] M. Hennerici, W. Steinke, C. Klotzsch, T. Els, Flow changes at the carotid bifurcation–an ultrasound study in the human, Vasa Suppl. 32 (1991) 66–71.
I.M. Graf et al. / Ultrasonics 52 (2012) 435–441 [7] J.J. Wentzel, R. Corti, Z.A. Fayad, P. Wisdom, F. Macaluso, M.O. Winkelman, V. Fuster, J.J. Badimon, Does shear stress modulate both plaque progression and regression in the thoracic aorta? Human study using serial magnetic resonance imaging, J. Am. Coll. Cardiol. 45 (2005) 846–854. [8] H.C. Groen, F.J. Gijsen, A. van der Lugt, M.S. Ferguson, T.S. Hatsukami, A.F. van der Steen, C. Yuan, J.J. Wentzel, Plaque rupture in the carotid artery is localized at the high shear stress region: a case report, Stroke 38 (2007) 2379–2381. [9] M. Bardelli, Ultrasonographic investigation of the mechanics of vulnerable atherosclerotic plaques: significance of the volume strain, J. Hypertens. 27 (2009) 219–220. [10] J. Andersson, J. Sundstrom, L. Kurland, T. Gustavsson, J. Hulthe, A. Elmgren, K. Zilmer, M. Zilmer, L. Lind, The carotid artery plaque size and echogenicity are related to different cardiovascular risk factors in the elderly: the Prospective Investigation of the Vasculature in Uppsala Seniors (PIVUS) study, Lipids 44 (2009) 397–403. [11] J. Heliopoulos, K. Vadikolias, P. Mitsias, D. Artemis, G. Tripsianis, C. Piperidou, N. Artemis, A three-dimensional ultrasonographic quantitative analysis of non-ulcerated carotid plaque morphology in symptomatic and asymptomatic carotid stenosis, Atherosclerosis 198 (2008) 129–135. [12] E. Falk, Pathogenesis of atherosclerosis, J. Am. Coll. Cardiol. 47 (2006) C7–12. [13] D. Vela, L.M. Buja, M. Madjid, A. Burke, M. Naghavi, J.T. Willerson, S.W. Casscells, S. Litovsky, The role of periadventitial fat in atherosclerosis, Arch. Pathol. Lab. Med. 131 (2007) 481–487. [14] S.E. Elias-Smale, M. Kavousi, G.C. Verwoert, M.T. Koller, E.W. Steyerberg, F.U. Mattace-Raso, A. Hofman, A.P. Hoeks, R.S. Reneman, J.C. Witteman, Common carotid intima-media thickness in cardiovascular risk stratification of older people: the Rotterdam Study, Eur. J. Cardiovasc. Prev. Rehabil. (2011). [15] W. Xie, L. Liang, L. Zhao, P. Shi, Y. Yang, G. Xie, Y. Huo, Y. Wu, Combination of carotid intima-media thickness and plaque for better predicting risk of ischaemic cardiovascular events, Heart 97 (2011) 1326–1331. [16] J.H. Stein, Carotid intima-media thickness progression and cardiovascular disease risk, J. Am. Coll. Cardiol. 57 (2011) 2291 [author reply 2292]. [17] M. Sareen, K. Waters, A. Nair, D.G. Vince, Normalization and backscatter spectral analysis of human carotid arterial data acquired using a clinical linear array ultrasound imaging system, in: Conf Proc IEEE Eng Med Biol Soc, 2008, pp. 2968–2971. [18] S.K. Kakkos, J.M. Stevens, A.N. Nicolaides, E. Kyriacou, C.S. Pattichis, G. Geroulakos, D. Thomas, Texture analysis of ultrasonic images of symptomatic carotid plaques can identify those plaques associated with ipsilateral embolic brain infarction, Eur. J. Vasc. Endovasc. Surg. 33 (2007) 422–429. [19] S. Yoshida, M.S. Nieminen, T. Paananen, A. Kahri, Can high frequency ultrasound and MRI diagnose malignant atheromatous plaque in vitro?, Jpn Heart J. 36 (1995) 235–245. [20] B.S. Garra, E.I. Cespedes, J. Ophir, S.R. Spratt, R.A. Zuurbier, C.M. Magnant, M.F. Pennanen, Elastography of breast lesions: initial clinical results, Radiology 202 (1997) 79–86. [21] E.C. Godia, R. Madhok, J. Pittman, S. Trocio, R. Ramas, D. Cabral, R.L. Sacco, T. Rundek, Carotid artery distensibility: a reliability study, J. Ultrasound Med. 26 (2007) 1157–1165. [22] L. Zhang, Y. Liu, P.F. Zhang, Y.X. Zhao, X.P. Ji, X.T. Lu, W.Q. Chen, C.X. Liu, C. Zhang, Y. Zhang, Peak radial and circumferential strain measured by velocity vector imaging is a novel index for detecting vulnerable plaques in a rabbit model of atherosclerosis, Atherosclerosis 211 (2011) 146–152. [23] P.F. Zhang, H.J. Su, M. Zhang, J.F. Li, C.X. Liu, S.F. Ding, Y. Miao, L. Chen, X.N. Li, X. Yi, Y. Zhang, Atherosclerotic plaque components characterization and macrophage infiltration identification by intravascular ultrasound elastography based on b-mode analysis: validation in vivo, Int. J. Cardiovasc. Imaging 27 (2011) 39–49. [24] A.P. Hoeks, K.D. Reesink, E. Hermeling, R.S. Reneman, Local blood pressure rather than shear stress should be blamed for plaque rupture, J. Am. Coll. Cardiol. 52 (2008) 1107–1108 [author reply 1108–1109]. [25] J. Folts, Arterial blood pressure gradient across vulnerable plaque might increase rupture, J. Am. Coll. Cardiol. 50 (2007) 2440. [26] H. Ribbers, R.G. Lopata, S. Holewijn, G. Pasterkamp, J.D. Blankensteijn, C.L. de Korte, Noninvasive two-dimensional strain imaging of arteries: validation in phantoms and preliminary experience in carotid arteries in vivo, Ultrasound Med. Biol. 33 (2007) 530–540. [27] A.P. Hoeks, J.M. Meinders, R. Dammers, Applicability and benefit of arterial transfer functions, J. Hypertens. 21 (2003) 1241–1243. [28] T.A. Whittingham, Medical diagnostic applications and sources, Prog. Biophys. Mol. Biol. 93 (2007) 84–110. [29] C.M. Wahlgren, W. Zheng, W. Shaalan, J. Tang, H.S. Bassiouny, Human carotid plaque calcification and vulnerability. Relationship between degree of plaque calcification, fibrous cap inflammatory gene expression and symptomatology, Cerebrovasc. Dis. 27 (2009) 193–200. [30] S.Y. Emelianov, S.R. Aglyamov, A.B. Karpiouk, S. Mallidi, S. Park, S. Sethuraman, J. Shah, R.W. Smalling, J.M. Rubin, W.G. Scott, Synergy and applications of
[31]
[32]
[33] [34]
[35]
[36] [37]
[38] [39] [40]
[41]
[42]
[43]
[44] [45] [46]
[47]
[48]
[49]
[50]
[51]
[52]
[53]
[54] [55]
441
ultrasound, elasticity, and photoacoustic imaging, in: Proceedings of IEEE Ultrasonics Symposium, 2006, pp. 405–415. P.C. Beard, T.N. Mills, Characterization of post mortem arterial tissue using time-resolved photoacoustic spectroscopy at 436, 461 and 532 nm, Phys. Med. Biol. 42 (1997) 177–198. A.A. Oraevsky, S.L. Jacques, G.H. Pettit, R.A. Sauerbrey, F.K. Tittel, J.H. Nguy, P.D. Henry, XeCl laser-induced fluorescence of atherosclerotic arteries. Spectral similarities between lipid-rich lesions and peroxidized lipoproteins, Circ. Res. 72 (1993) 84–90. T.J. Allen, P.C. Beard, A. Hall, A. Dhillon, J.S. Owen, Photoacoustic imaging of lipid rich plaques in human aorta, in: SPIE, 2010, pp. 75640C–75641C. S. Park, J. Shah, S.R. Aglyamov, A.B. Karpiouk, S. Mallidi, A. Gopal, H. Moon, X.J. Zhang, W.G. Scott, S.Y. Emelianov, Integrated system for ultrasonic, photoacoustic and elasticity imaging, in: S. Emelianov, W.F. Walker (Eds.) Proceedings of SPIE, 2006, pp. 61470H. F.D. Kolodgie, A.S. Katocs Jr., E.E. Largis, S.M. Wrenn, J.F. Cornhill, E.E. Herderick, S.J. Lee, R. Virmani, Hypercholesterolemia in the rabbit induced by feeding graded amounts of low-level cholesterol. Methodological considerations regarding individual variability in response to dietary cholesterol and development of lesion type, Arterioscler. Thromb. Vasc. Biol. 16 (1996) 1454–1464. O. Header, Opt. Absorp. Spectra (2001). P.J. Brands, A.P. Hoeks, L.A. Ledoux, R.S. Reneman, A radio frequency domain complex cross-correlation model to estimate blood flow velocity and tissue motion by means of ultrasound, Ultrasound Med. Biol. 23 (1997) 911–920. S.A. Prahl, Opt. Prop. Spectra (2001). C.L. Tsai, J.C. Chen, W.J. Wang, Near-infrared absorption property of biological soft tissue constituents, J. Med. Biol. Eng. 21 (2001) 7–14. E.M. Rubin, A.A. Martin, S.N. Thung, M.A. Gerber, Morphometric and immunohistochemical characterization of human liver regeneration, Am. J. Pathol. 147 (1995) 397–404. A. Paini, P. Boutouyrie, D. Calvet, M. Zidi, E. Agabiti-Rosei, S. Laurent, Multiaxial mechanical characteristics of carotid plaque: analysis by multiarray echotracking system, Stroke 38 (2007) 117–123. S. Park, S.R. Aglyamov, W.G. Scott, S.Y. Emelianov, Strain imaging using conventional and ultrafast ultrasound imaging: numerical analysis, IEEE Trans. Ultrason. Ferroelectr. Freq. Control 54 (2007) 987–995. S.L. Wilens, The postmortem elasticity of the adult human aorta. its relation to age and to the distribution of intimal atheromas, Am. J. Pathol. 13 (1937) 811– 834. B.S. Gow, C.D. Hadfield, The elasticity of canine and human coronary arteries with reference to postmortem changes, Circ. Res. 45 (1979) 588–594. R.T. Yen, S.S. Sobin, Elasticity of arterioles and venules in postmortem human lungs, J. Appl. Physiol. 64 (1988) 611–619. I.M. Graf, F.H. Schreuder, W.H. Mess, R.S. Reneman, A.P. Hoeks, Spatial distension variations are associated with focal atherosclerotic plaques, Cerebrovasc. Dis. 29 (2009) 199–205. T. Montalcini, G. Gorgone, C. Gazzaruso, A. Pujia, Carotid distension and distensibility impairment in individuals affected by familial combined hyperlipidemia, Atherosclerosis 212 (2010) 177–180. B. Wang, J.L. Su, J. Amirian, S.H. Litovsky, R. Smalling, S. Emelianov, Detection of lipid in atherosclerotic vessels using ultrasound-guided spectroscopic intravascular photoacoustic imaging, Opt. Express 18 (2010) 4889–4897. X. Niu, V. Zammit, J.M. Upston, R.T. Dean, R. Stocker, Coexistence of oxidized lipids and alpha-tocopherol in all lipoprotein density fractions isolated from advanced human atherosclerotic plaques, Arterioscler. Thromb. Vasc. Biol. 19 (1999) 1708–1718. J.H. Rapp, W.E. Connor, D.S. Lin, T. Inahara, J.M. Porter, Lipids of human atherosclerotic plaques and xanthomas: clues to the mechanism of plaque progression, J. Lipid Res. 24 (1983) 1329–1335. S.P. Roux, H. Kuhn, H. Lengsfeld, O.H. Morand, Effects of chronic aortic coarctation on atherosclerosis and arterial lipid accumulation in the Watanabe hereditary hyperlipidemic (WHHL) rabbit, Atherosclerosis 93 (1992) 123–132. Z. Ma, A. Choudhury, S.A. Kang, M. Monestier, P.L. Cohen, R.A. Eisenberg, Accelerated atherosclerosis in ApoE deficient lupus mouse models, Clin. Immunol. 127 (2008) 168–175. P. Babusik, I. Duris, Comparison of obesity and its relationship to some metabolic risk factors of atherosclerosis in Arabs and South Asians in Kuwait, Med. Princ. Pract. 19 (2010) 275–280. P. Sucharda, Obesity and atherosclerosis – what’s the link? Vnitr. Lek. vol. 56, 2010, pp. 289–291. J. Cai, T.S. Hatsukami, M.S. Ferguson, W.S. Kerwin, T. Saam, B. Chu, N. Takaya, N.L. Polissar, C. Yuan, In vivo quantitative measurement of intact fibrous cap and lipid-rich necrotic core size in atherosclerotic carotid plaque: comparison of high-resolution, contrast-enhanced magnetic resonance imaging and histology, Circulation 112 (2005) 3437–3444.