1. Introduction

Atherosclerosis and its thrombotic complications, despite the significant advances in cardiology, continue to be the leading cause of morbidity and mortality worldwide. Progression of atherosclerotic lesions is a complex process influenced by mechanical and biological factors. Rather than degree of stenosis, the risk of acute coronary events is largely dependent on the mechanical properties and plaque morphology, therefore, both features should be considered in an integrated way for a more accurate assessment of atherosclerosis [1, 2]. Thus, a method that can integrate mechanical property with structural detail produces an attractive synergy of atherosclerosis characterization.

Photoacoustic (PA) imaging is a hybrid imaging technique based on a conversion from light to ultrasonic waves with a wide range of frequencies, which combines the advantages of high optical contrast and high ultrasonic resolution [3–7]. Conventional PA imaging is achieved by measuring the intensity of temporal PA signals generally in frequency of million hertz (MHz), which reflects the optical absorption contrast and can provide structural information of tissues with volumetric images [8–15]. Our recent study found that, the force-produced PA wave generally in frequency of kilo hertz (KHz) will produce a phase lag behind the laser excitation due to the damping effect of biological viscoelasticity [16, 17], the phase-resolved images reflect the viscoelasticity contrast and can provide mechanical characterization of tissues.

Previously, we proposed and validated simultaneous optical absorption and viscoelasticity imaging based on photoacoustic lock-in measurement [17]. However, due to the sine wave excitation via the continuous modulated laser (808 nm), this method could not provide depth information theoretically. Moreover, in in vivo or clinical application, the uneven surface of tissues would introduce a distance-dependence phase deviation for the PA phase detection, which will cause misleading viscoelasticity characterization. In this paper, to overcome the limitations, we utilized a quasi-continuous laser for pulsed excitation, and temporal PA signals was acquired to realize integration of en-face mechanical and in-depth structural information for PA imaging. Furthermore, an algorithm using a sound propagation model in conjunction with temporal PA intensity is developed to correct the phase deviation. These advantages of the integrated mechanical and structural PA imaging (IMS-PAI) enable a holistic evaluation of atherosclerosis by marrying viscoelastic property with structural morphology, which suggests a prospect in plaque vulnerability assessing.

2. Methods and materials

Generally, irradiate laser pulses are absorbed by tissue and partially converted into heat due to non-radiative transition, resulting in generated ultrasonic waves based on the thermo-elastic expansion. Meanwhile, the cyclical temperature variation in local region induces a thermal stress, generating a strain in the form of force-produced PA wave, which has the same frequency with laser excitation, while a phase lag behind it exists owing to the damping effect caused by biological viscoelasticity. In the rheological Kelvin-Voigt model, PA phase δ can be expressed by Eq. (1) [16],

The schematic setup of the IMS-PAI system was shown in Fig. 1 . A quasi-continuous laser (DS20HE-1064D/R, PHOTONICS) with pulse width of 22 ns and wavelength of 1064 nm operating at the repetition frequency of 25 KHz was used as the excitation source. The collimated laser was focused by a microscope objective to illuminate on the sample surface. The sample was fixed on a two-dimensional motor scanning platform between two ultrasound transducers (UT). UT1 holds hollow bowl-shape style and has center frequency of 25 KHz. UT2 has center frequency of 3.5 MHz (U8518056, Olympus) or 75 MHz (U8424009, Olympus). During data acquisition, the generated PA signals were acoustically coupled with distilled water and divided into two paths to obtain PA phase and temporal PA intensity respectively. PA signals detected by UT1 were firstly transferred to a low-pass low-noise preamplifier (SR552, Stanford Research Systems), then calculated by a lock-in detector (SR830, Stanford Research Systems) to resolve the PA phase. PA signals detected by UT2 were firstly transferred to a wide-bandwidth low-noise amplifier (Ha2, Precision Acoustics LTD), and acquired by a data acquisition system (NI PCI-5124, National Instruments). Both the PA phase and temporal intensity were recorded and analyzed on a computer, which controlled the motorized scanner with a custom program written by LABVIEW (National Instruments, USA) software simultaneously. At each step (50 μm) of the scan, time-averaged laser intensity on the tissue surface was limited well within the American National Standard Institute's safety limit (100 mJ/cm²) [19].

 

Fig. 1 Schematic of IMS-PAI system. CL: collimating lens, MO: microscope objective, UT: ultrasonic transducer, DAS: data acquisition system.

Download Full Size | PDF

3. Results

3.1 PA phase correction based on temporal PA intensity

The PA phase correction based on temporal PA intensity was firstly validated with a step-phantom. As shown in Fig. 2(a) , the phantom with a 1.14-mm-high step contained 4% agar and interfused with 2% ink. The temporal PA intensity (A-lines) at α and β were illustrated in Fig. 2(b), where the upper and lower boundary could be seen clearly. The PA phase correction was exhibited in Fig. 2(c). The phantom at the same agar concentration had the same viscoelasticity, and expected to obtain the same PA phase. But due to the existence of the step, the generated PA signal positioned at the lower side would introduce an extra distance-dependence phase deviation, as the blue dot shown in detected PA phase. Set the z coordinate value of upper boundary at x = 0 mm as the baseline of H = 0, where H was the depth relative to the baseline, so that H distribution could be determined by the temporal PA intensity, as shown in black triangle mark. The introduced phase deviation was corrected by an algorithm using a sound propagation model in conjunction with H distribution. The acoustic speed in the water was 1500 m/s and the acoustic frequency was 25 KHz, in theory, the distance of 1 mm would introduce 6° phase delay, then the actual PA phase could be obtained, as the red dot shown in corrected PA phase. The experiment demonstrated that the algorithm can reliably achieve accurate phase correction, and will benefit actual detection for tissues with uneven surface.

 

Fig. 2 (a) Agar phantom with a 1.14 mm high step (side view). (b) Temporal PA intensity at α and β, “U1” and “U2” correspond to upper boundary, “L” correspond to lower boundary. (c) PA phase correction based on temporal PA intensity.

Download Full Size | PDF

3.2 Phantom imaging of IMS-PAI

The IMS-PAI system was subsequently verified with a tissue-mimicking phantom. The structure of the 6.5 × 6.5 × 4 mm³ phantom was illustrated in Fig. 3(a) . In the 1.5-mm-thick top layer, a fan-shaped component contained 6% agar and 3% ink was buried in a square-shaped component contained 2% agar and 1% ink. The 2.5-mm-thick bottom layer was comprised of 3% agar and 4% ink. As is evident in Fig. 3(b), en-face PA viscoelasticity and absorption images matched well, and the fan-shaped component revealed lower viscoelasticity and higher optical absorption than the surrounding region. Integrated PA sections with en-face viscoelasticity and in-depth absorption distribution at different X positions were shown in Fig. 3(c). The arrow “a” and “b” were referred to upper and lower boundary of the top layer. These PA sections indicated a same en-face morphological change trend as shown in the dashed frame, and exhibited an integrated en-face viscoelasticity and in-depth absorption imaging modality. The experiment demonstrated the priority of the IMS-PAI for comprehensive PA characterization.

 

Fig. 3 (a) Phantom structure. (b) En-face PA viscoelasticity and absorption images. (c) Integrated PA sections.

Download Full Size | PDF

3.3 Imaging of atherosclerotic tissue by IMS-PAI system

To demonstrate the feasibility of IMS-PAI for biomedical application, an atherosclerotic tissue with a fatty streak harvested from a 15-weeks high-fat/high-cholesterol diet feeding rabbit was tested. Laser was illuminated on the inside surface of the sample. A high frequency UT (75 MHz) was used in this experiment to image the thin-layer structure of vascular wall. The photograph of the atherosclerotic tissue was shown in Fig. 4(a) , and tissue within the red dashed frame was used for ex vivo examination. The en-face PA viscoelasticity and absorption images shown in Fig. 4(b) distinguished the morphology of the fatty streak and corresponded well with the sample. The fatty tissue (main component is lipid) suggested higher viscoelasticity and slightly lower optical absorption than that of the surrounding normal tissue (main component is collagen). The result was consistent with the fact that the viscoelasticity of lipid is greater than collagen [20, 21], and the optical absorption of lipid is lesser than collagen at 1064 nm [22]. The inhomogeneity of PA viscoelasticity and absorption distribution was mainly attributed to the different degree of lipid accumulation. PA absorption within the fatty streak was relatively uniform, while PA viscoelasticity exhibited a high-phase area located in the top left corner, which may be caused by the inflammation of atherosclerosis. After PA experiment, the specimen was sliced and evaluated with cross-sectional Oil red O staining. The integrated PA sections and the corresponding histology at different Y positions were shown in Fig. 4(c). In the fatty streak, the intima thickening resulting from lipid accumulation exhibited high viscoelasticity and dense Oil red O staining. The most severe intima-media thickening at Y = 2.5 mm was about 0.7 mm, which was highly consistent with the histology. The imaging depth of the IMS-PAI system for atherosclerosis characterization was about 1.5 mm. The slices were counterstained with hematoxylin to visualize cell nuclei. Grainy staining in clusters beneath the endothelial layer showed a sign of foam cell infiltration. These integrated PA sections allowed complementary visualization of en-face viscoelasticity distribution and in-depth structural anatomy for the fatty streak, where the distortion of viscoelasticity distribution may be an early warning of plaque rupture, and the degree of intima thickening relates to the lesion extent. The experiment demonstrated the feasibility for accurate medical evaluation of atherosclerosis with the IMS-PAI.

 

Fig. 4 (a) Photograph of the atherosclerotic tissue with a fatty streak, region within the dashed frame is scanning area. (b) En-face PA viscoelasticity and absorption images. (c) Integrated PA sections and corresponding histology. The sections were stained with Oil red O (red) to evaluate lipid accumulation and counterstained with hematoxylin (blue) to visualize cell nuclei.

Download Full Size | PDF

4. Discussion and conclusions

By using the quasi-continuous laser instead of previously used continuous modulated laser as excitation source, depth information could be obtained to combine with viscoelastic information for comprehensive characterization. In addition, the temporal-intensity-based PA phase correction would facilitate practical application for tissues with uneven surface. Besides, unlike fusing with different technologies, mismatch in the analysis between the two parameters could be excluded since the PA phase and temporal intensity were acquired simultaneously in the same PA system. Despite the superiorities, there are still several further improvements should be made. First, the current used PA phase correction was some extent relying on the signal-to-noise ratio of detected PA signals, an independent and applicable algorithm need to be established for more accurate phase correction. Second, the scanning rate of the IMS-PAI system was 33 Points/Sec, which was mainly limited by the time constant of the lock-in detector. A focused ultrasound transducer will be employed in the PA phase detector mode to improve the detecting sensitivity and phase stability, thus the time constant can be reduced and the imaging speed can be accelerated. Third, although this IMS-PAI system was well applied to the ex vivo detection of atherosclerosis, a fully integrated system with a focused broadband transducer in reflection mode will be used in the future for in vivo and endoscopic detection, which indicates a potential in intravascular plaque vulnerability assessing.

In conclusion, we have developed an IMS-PAI for simultaneously en-face mechanical and in-depth structural characterization of atherosclerosis, which suggested a prospect in understanding the effect of local mechanical property on plaque progression and visualizing the coronary anatomy. The comprehensive information would contribute a lot to basic science research and clinical diagnosis of atherosclerosis.