Open Access
Add to BibSonomy
Abstract
High-intensity focused ultrasound (HIFU) is a promising non-invasive therapeutic technique in clinical applications. Challenges in stimulation or ablation HIFU therapy are to accurately target the treatment spot, flexibly deliver or fast-move focus points in the treatment region, and monitor therapy progress in real-time. In this paper, we develop an array-based HIFU system integrated with real-time ultrasound (US) and photoacoustic (PA) imaging. The array-based HIFU transducer can be dynamically focused in a lateral range of ∼16 mm and an axial range of ∼40 mm via electronically adjusting the excitation phase map. To monitor the HIFU therapy progress in real-time, sequential HIFU transmission, PA imaging, PA thermometry, and US imaging are implemented to display the dual-modal images and record the local temperature changes. Co-registered dual-modal images show structural and functional information and thus can guide the HIFU therapy for precise positioning and dosage control. Besides therapy, the multi-element HIFU transducer can also be used to acquire US images to precisely align the imaging coordinates with the HIFU coordinates. Phantom experiments validate the precise and dynamic steering capability of HIFU ablation. We also show that dual-modal imaging can guide HIFU in the designated region and monitor the temperature in biological tissue in real-time.
© 2023 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
Minimal or non-invasive therapy in deep tissue is attracting great attention in the medical field. High-intensity focused ultrasound (HIFU) provides a promising approach to non-invasive surgery [1]. HIFU therapy uses thermal or mechanical energy to ablate or stimulate a designated region, for example, by treating tumors or other neural disorder diseases. Compared with other thermal-therapy methods, HIFU has the advantages of deep penetration, controllable ablation spot size, low hardware cost, and non-invasive therapy [2]. However, most HIFU transducers have fixed focus depths and focal spots smaller than the actual dimension of the treatment region [3–5]. Thus, the treatment time will be increased due to the mechanical scanning of the transducer. The second problem is the designated treatment spot and the actual HIFU focus may be misaligned in the long-time mechanical scanning. The third challenge for HIFU therapy is safety in surgery. Planning before treatment, monitoring the treatment in real-time, and evaluating the treatment outcome is important for an effective and safe treatment. Especially, precise and timely dose control is critical to minimize the damage to surrounding healthy tissues. Different imaging techniques have been developed to guide HIFU therapy. Magnetic resonance image-guided high-intensity focused ultrasound (MRgHIFU) has high-temperature monitoring sensitivity in deep tissue. But the long acquisition time, incompatible ferromagnetic materials, and high operating cost have limited its clinical application [6–10]. Ultrasound (US) imaging combined with HIFU therapy can achieve real-time monitoring and has a low hardware cost and high resolution. However, the US signal has low dependence on temperature and thus has insufficient sensitivity to accurately guide the HIFU treatment [11,12]. Conventional optical imaging may guide HIFU with a high resolution and high sensitivity but cannot penetrate deep tissue [13].
To address the challenges of long treatment time and inaccurate imaging guidance, we develop a novel array-based HIFU therapeutic system integrated with photoacoustic (PA) and US imaging for the first time, which offers complementary temperature and structural information to therapeutic guidance [4,14,15]. In the new system, the PA/US imaging transducer and the HIFU transducer are connected to one data acquisition system, and the HIFU transmission and the PA and US imaging are programmed synchronously. Different from the previously reported HIFU system integrated with PA imaging [4,5,15], the array-based HIFU transducer has a wide dynamic steering range in both the axial (40 mm) and the lateral (16 mm) directions. The treatment spots can be electronically moved by adjusting the excitation phase map. Therefore, fast multiply spots treatment can be realized in this integrated system. Additionally, the multi-element HIFU transducer can also be used to acquire US images and align with the imaging transducer in the axial direction, thus the tedious spatial calibration process between the therapeutic and diagnostic transducer can be avoided. According to the feedback from dual-modal images, the HIFU applicator enables automatically and selectively decides the ablation region. HIFU ablation, PA and US imaging, and real-time temperature monitoring are demonstrated in biological tissue.
2. Methods
2.1 Integrated real-time HIFU/PA/US system
As depicted in Fig. 1(a), the integrated HIFU/PA/US system includes four main parts: a control unit, a data acquisition (DAQ), an optical system, and a HIFU/imaging probe. The control unit has a motion controller to translate the sample, a sequence controller to synchronize the HIFU transmission, the PA and US imaging, and the PA-based temperature monitoring. The DAQ (Vantage, Verasonics Inc, USA) has 256 independent Transmission/Receiving channels where 128 channels are used for HIFU therapy and the remaining ones are used for PA/US imaging. The optical system includes a Q-switched Nd: YAG laser (Quanta-Ray, INDI-40−20, SpectraPhysics, Santa Clara, California) and a tunable optical parametric oscillator laser (basiScan-M/120/HE, Spectral-Physics). The optical wavelength is tunable from 410 nm to 2500 nm. The laser pulse width is 8 nm, and the maximal pulse repetition frequency (PRF) is 20 Hz. The HIFU/imaging probe integrates a 128-channel HIFU transducer and a linear phased-array imaging transducer. The HIFU transducer (H-302, Sonic Concepts Inc, Seattle, WA) has a 2-MHz center frequency, 60% bandwidth, and 150-mm geometrical focus length. The linear imaging transducer (IP-105, Verasonics Inc, USA) has a 5.2-MHz center frequency and over 90% bandwidth and can fit in a central opening in the HIFU transducer. The used transmission power for the imaging transducer is always 3.2 W unless otherwise stated. As shown in Fig. 1(c), the HIFU elements are closely packed on a spherical surface in a single Archimedean spiral form [16]. In addition to the simple structure of the Archimedean spiral array, it has a large steering range. For example, the HIFU transducer has a spot size measured 0.8 mm in lateral and 7.34 mm in axial length, and can be steered 40 mm in the axial and 16 mm in the lateral directions.
Fig. 1. Set-up of the tri-modal system and the control sequence. (a) Schematic of the tri-modal system. (b) A 3D model of the water tank and the sample holder. (c) Arrangement of the HIFU elements and the PA and US imaging plane. (d) The time sequence of the HIFU, PA, and US events. $\Delta {T_1}$ is 50 milliseconds, $\Delta {t_0}$ is 2 milliseconds, $\Delta {t_1}$ is 0.3 milliseconds, and $\Delta {t_2}$ is 0.5 milliseconds. (e) Diagram of the synchronization between the DAQ and the laser firing for precise PA imaging. acq., acquisition; BS, beam splitter; DAQ, data acquisition system; HIFU, high intensity focused ultrasound; IUT, imaging ultrasound transducer; Mem, membrane; OPO, optical parametric oscillator; PA, photoacoustic; recon., reconstruction; Temp., Temperature; US, ultrasound; WT, water.
The HIFU transducer and the imaging transducer are aligned in the azimuthal direction via the marks provided by the manufacturer on the transducer surfaces. An image calibration approach is developed to align them in the axial direction. Firstly, the HIFU transducer is used to image a stable water-air interface in a pulse-echo mode. The imaging is in the X-Z and Y-Z planes. Then, the linear-array imaging transducer is used to image the same water-air interface. Finally, three positions in the two sets of images are compared in the depth direction to compensate for the misalignment between the HIFU and the imaging transducers.
The imaged or ablated sample is fixed on a holder that is partially immersed in water (Fig. 1(a) and 1(b)). To transmit US and isolate water, the bottom of the holder is made from a thin layer of polyethylene membrane for optical and acoustic transmission. The anechoic water tank is filled with degassed water whose temperature is set at ∼30°C in experiments. As shown in Fig. 1(d), the HIFU, PA, and US imaging modes are sequentially triggered to monitor the ablation in real time and avoid cross-talks. When a treatment region is manually selected from the dual-modal PA/US images, the HIFU transducer is activated. To synchronize with dual-modal imaging, the HIFU transmission repeats every 50 milliseconds (20 Hz), and in each cycle, the HIFU transmits for 40 milliseconds. The duty cycle, therefore, is 0.8. To guarantee precise synchronization of DAQ and laser firing and achieve high-quality PA imaging, the laser lamp operates at 20 Hz and sends a TTL trigger signal to the DAQ. After delaying by 200 µs, the DAQ sends a signal to trigger the laser Q-switch to output a laser pulse for PA imaging (Fig. 1(e)) [17,18]. In US imaging, we employed a diverging wave (DW) transmission. Twelve diverging beams spanning from -18° to 18° cover the entire medium. The time interval between two adjacent US transmissions is 500 µs. The coherent delay-and-sum (DAS) algorithm was employed for both US and PA image reconstruction [19].
2.2 Photoacoustic thermometry
Because the efficiency of PA excitation is proportional to the Grueneisen parameter, and the local temperature changes can change the Grueneisen parameter, PA amplitude can be used to monitor the change of the local temperature in deep tissue [5,20]. With moderate temperature rises, the PA amplitude can be approximated as a linear function of the local temperature in degrees Celsius. If the temperature is above ∼50°C, the linear relationship will not be accurate due to the nonlinear temperature-dependency of the Grueneisen parameter [21]. In addition, the high temperature may change the tissue’s absorption coefficient and thus cause nonlinearity in the temperature-PA relationship. Nonetheless, PA amplitude can still be used to estimate the tissue temperature in a relatively linear range [22]. The relationship between the PA amplitude and the local temperature is described as
Where ${\bar{V}_p}({x,y} )$ is the mean PA amplitude, ${C_0}$ and ${C_1}$ are constant coefficients associated with the absorption coefficient, tissue properties, and optical fluence, T is the local temperature in the region of interest. If the initial pixel value is ${\bar{V}_{p0}}({x,y} )$, and the baseline temperature is ${T_0}$, then3. Experiments and results
3.1 System characterization
Figure 2 shows the system characterization results of acoustic pressure, depth alignment, and spatial resolutions. A hydrophone (Model HGL-0200, Onda Corporation, Sunnyvale, CA) was used to record the pulse repetition time (PRT) of HIFU exposure for 500 milliseconds. The transmitting power (3.2 W) of the HIFU transducer should not be set high to avoid hydrophone damage. The hydrophone was placed near the HIFU focus to measure the emitted US waves. As shown in Fig. 2(a), the PRT is 50 ms and the HIFU transmitting time is 40 ms, which conforms with the time sequence. Figure 2(b) shows the depth alignment process. First, the HIFU array transducer worked in pulse-echo mode to acquire the images of the water/air interface in the X-Z and X-Y planes. Then, we compared the two images with the US image of the same water/air interface acquired with the imaging transducer. The difference was used to compensate for the transducer misalignment in the axial direction. The spatial resolution of the imaging transducer was measured using a tungsten wire with a 100-µm diameter. The tungsten wire was placed at a depth of 150 mm and was perpendicular to the X-Z plane. Figure 2(c) and 2(d) show the lateral and axial resolutions for the US and PA imaging in the HIFU focal region. The lateral resolutions for US and PA imaging are 786 µm and 919 µm. The axial resolutions for US and PA imaging are 212 µm and 225 µm. Because the F-number (The ratio of focus depth and length of aperture) increases with the imaging depth, the lateral resolution is worse in deep tissue than the axial resolution.
Fig. 2. Results of system characterization and depth alignment. (a) Signal recorded by a hydrophone near the HIFU focus. (b) The depth alignment process. Two US images are acquired with the HIFU transducer in the X-Z and Y-Z planes. One US image in the X-Z plane is acquired with the imaging probe. The depths of the water/air interface in these images are used to align the two transducers in the depth direction. (c) US imaging of a tungsten wire. The lateral and axial profiles are plotted to quantify the US resolutions. (d) PA imaging of the tungsten wire. The lateral and axial profiles show the PA resolutions.
3.2 Simultaneous multi-spot ablation guided by a real-time US/PA image
We demonstrated the dynamic ablation using the array-based HIFU transducer with the US/PA imaging guiding. To validate the multi-spot ablation ability, a circle ablation pattern with nine spots was designated as shown in Fig. 3(a). The diameter of the circular pattern is 8 millimeters. According to the position information, the HIFU system automatically ablated a spot for 15 seconds (25 W) and then moved to the next spot. The total ablation time lasted 135 seconds. The right image in Fig. 3(a) shows the ablation result on a Polydimethylsiloxane (PDMS) sample, matching well with the designated pattern. Because the HIFU intensity is higher near the steering center than the edge, the middle region usually absorbs more acoustic energy with the same exposure time. To solve this problem, we can increase acoustic power dynamically when the HIFU focus is away from the center of the steering range. Figure 3(b) shows an ex-vivo experimental result. Two carbon rods with a diameter of 2 millimeters were inserted into a piece of fresh chicken breast tissue at two different depths. From the dual-modal US/PA images, we can determine the ablation spots and steer the HIFU focus to these locations. The power for biological tissue ablation is set at 52 W. The target lesions can be seen in the tissue after ablation. The optical wavelength used in PA imaging is 720 nm, and the laser fluence is about 5.4 mJ/cm2 on the tissue surface. The PA pressure is approximately proportional to the local optical fluence, optical absorption coefficient, and the Grueneisen parameter [23–26]. If the fluence and the Grueneisen are kept unchanged, the PA signal amplitude differences are mainly influenced by optical absorption. Therefore, PA imaging can guide HIFU to ablate the desired region according to their difference in signal amplitude. We also demonstrated the system can ablate designated regions according to the different PA images. As shown in Fig. 3(c), we prepared a phantom of chicken breast tissue and black-ink-filled polyethylene tubes to mimic the blood vessels. The ink concentrations in the three tubes were set at 20%, 50%, and 100%, respectively. The PA images of the ink tubes can be displayed in real-time. We can see the PA signal is high in the high-ink-concentration tube. The tube with the 50% ink concentration was selected for HIFU ablation. Finally, the tube was removed to clearly show the ablated tissue.
Fig. 3. Multi-spot HIFU ablation and US/PA guidance. (a) Pre-determined ablation pattern and the ablation results in a PDMS phantom. (b) Dual-modal US/PA-guided HIFU ablation. (c) Selected HIFU ablation under the guidance of PA imaging. A phantom of three black-ink-filled tubes buried in chicken breast tissue and the PA images. The tissue at the position of the 50% ink tube was ablated.
3.3 PA thermometry in HIFU ablation
Figure 4 shows the results of tissue experiments to validate the temperature monitor using PA thermometry. A black-ink-filled tube was inserted into a piece of chicken breast tissue. As shown in Fig. 4(a), a needle-shaped thermocouple (HYP0-33-1-T-G-60-SMPW-M, Omega, USA) was placed near the ink tube. Both the tube and the thermocouple tip were visible in the US and PA images. The tube position was measured from the PA image and used to guide the HIFU focus. Figure 4(c) shows the temperature rise recorded with the thermocouple in 3 experiments. At the same time, PA images were acquired at 20 Hz. We calculated the mean PA amplitude of 8 × 8 pixels at the ablation spot. Figure 4(d) shows the change in the average PA amplitude. The PA change has a similar trend to the thermocouple reading. The samples located in ascending stage were selected to calculate the average value of ${K_0}$ (Calibration coefficient in Eq. (2)). The result could be seen in Fig. 4(b). The mean value of ${K_0}$ is 26.92, 27.85, and 26.99, respectively. The fluctuation may be attributed to the noise of the measured PA values, and the thermocouple tip may not well accurately coincide with the measured region and so on.
Fig. 4. Results of PA thermometry. (a) US/PA imaging of sample and thermocouple in chicken tissue. (b) Calibration coefficient ${K_0}$. calculated from (c) the thermocouple results and (d) the PA results. (c) The temperature was recorded with the thermocouple in three ablation experiments. The ascending stage of the temperature curve is corresponding to the HIFU ablation period. (d) Normalized PA amplitude change in the three ablation experiments.
4. Conclusions
An array-based HIFU system integrated with PA and US imaging is developed. The HIFU beam can be dynamically focused in 3D space, which is flexible and fast in scanning. US and PA imaging provide complementary information and contrast to guide the HIFU localization. Besides, PA thermometry can monitor the temperature in the ablation progress.
Although in vitro experiments have demonstrated the performance of the system, there are some technical problems needed to be addressed. First, the worse lateral resolution is degraded because of the long imaging distance, the imaging transducer suitable for long working distances will be adopted. Second, in vivo experiments will be done and analyzed to demonstrate the system’s performance in the future. Third, although the HIFU system provides 3D thermal ablation, the imaging transducer is performed in 2D. To implement 3D imaging and provide a comprehensive reference for the HIFU ablation, the imaging transducer in the central opening of the HIFU transducer could be rotated by 360 degrees in our follow-up studies and next-generation system. Finally, the accuracy of PA thermometry will be improved. For soft tissue like the skin, it has been observed a linear proportionality between the PA amplitude and the measured temperature if the local temperature is not less than 50°C [4]. The absorption spectrum of biological tissue may change due to protein denaturing if the local tissue temperature exceeds the critical point. The linear relationship will not be satisfied because of the nonlinear temperature-dependency of the Grueneisen parameter, the changed optical absorption coefficient, and the varying local optical fluence. In addition, because of the various combinations of fatty and aqueous in different tissues or organs, the temperature-dependent PA signal may be individual differences. To improve the PA thermometry accuracy, one possible method is to calibrate the linear amplitude-temperature relationship ex vivo in advance. Thus the thermal profile of the PA signal can be acquired for the different target applications. The critical temperature before protein denaturing can also be controlled. Another is quantified PA reconstruction by considering the acoustic- and optical- heterogeneity in biological tissue. The corrected PA amplitude will affect the PA thermometry accuracy.
In conclusion, because of the advantages of multi-modal imaging and non-invasive therapy, the tri-modal system still shows great prospects in clinical application.
Funding
National Natural Science Foundation of China (62135006, 81627805, 81930048); City University of Hong Kong (7020004); University Grants Committee of Hong Kong Special Administrative Region (11101618, 11103320, 11104922).
Disclosures
Lidai Wang has a financial interest in PATech Limited, which, however, did not support this work. All authors declare no competing interests.
Data availability
Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.
References
1. S. A. Quadri, M. Waqas, I. Khan, M. A. Khan, S. S. Suriya, M. Farooqui, and B. Fiani, “High-intensity focused ultrasound: past, present, and future in neurosurgery,” Neurosurg. Focus 44(2), E16 (2018). [CrossRef]
2. K. F. Chu and D. E. Dupuy, “Thermal ablation of tumours: biological mechanisms and advances in therapy,” Nat. Rev. Cancer 14(3), 199–208 (2014). [CrossRef]
3. Y. S. Kim, B. Keserci, A. Partanen, H. Rhim, H. K. Lim, M. J. Park, and M. O. Köhler, “Volumetric MR-HIFU ablation of uterine fibroids: role of treatment cell size in the improvement of energy efficiency,” Eur. J. Radiol. 81(11), 3652–3659 (2012). [CrossRef]
4. J. Kim, W. Choi, E. Y. Park, Y. Kang, K. J. Lee, H. H. Kim, W. J. Kim, and C. Kim, “Real-Time Photoacoustic Thermometry Combined with Clinical Ultrasound Imaging and High-Intensity Focused Ultrasound,” IEEE Trans. Biomed. Eng. 66(12), 3330–3338 (2019). [CrossRef]
5. X. Wu, J. L. Sanders, D. N. Stephens, and O. Oralkan, “Photoacoustic-imaging-based temperature monitoring for high-intensity focused ultrasound therapy,” 2016 38th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), 3235–3238 (2016). [CrossRef]
6. S. J. Graham, M. J. Bronskill, and R. M. Henkelman, “Time and temperature dependence of MR parameters during thermal coagulation of ex vivo rabbit muscle,” Magn. Reson. Med. 39(2), 198–203 (1998). [CrossRef]
7. D. Germain, E. Vahala, G. J. Ehnholm, T. Vaara, M. Ylihautala, M. Savart, A. Laurent, J. Tanttu, and H. Saint-Jalmes, “MR temperature measurement in liver tissue at 0.23 T with a steady-state free precession sequence,” Magn. Reson. Med. 47(5), 940–947 (2002). [CrossRef]
8. C. Weidensteiner, B. Quesson, B. Caire-Gana, N. Kerioui, A. Rullier, H. Trillaud, and C. T. W. Moonen, “Real-time MR temperature mapping of rabbit liver in vivo during thermal ablation,” Magn. Reson. Med. 50(2), 322–330 (2003). [CrossRef]
9. F. Siedek, S. Y. Yeo, E. Heijman, O. Grinstein, G. Bratke, C. Heneweer, M. Puesken, T. Persigehl, D. Maintz, and H. Grull, “Magnetic Resonance-Guided High-Intensity Focused Ultrasound (MR-HIFU): Technical Background and Overview of Current Clinical Applications (Part 1),” Rofo 191(6), 522–530 (2019). [CrossRef]
10. M. A. Solovchuk, S. C. Hwang, H. Chang, M. Thiriet, and T. W. H. Sheu, “Temperature elevation by HIFU in ex vivo porcine muscle: MRI measurement and simulation study,” Med. Phys. 41(5), 052903 (2014). [CrossRef]
11. E. S. Ebbini and G. Ter Haar, “Ultrasound-guided therapeutic focused ultrasound: Current status and future directions,” Int. J. Hyperthermia 31(2), 77–89 (2015). [CrossRef]
12. M. A. Lewis, R. M. Staruch, and R. Chopra, “Thermometry and Ablation Monitoring with Ultrasound,” Int. J. Hyperthermia 31(2), 163–181 (2015). [CrossRef]
13. H. Yang, S. Aleman, and H. Jiang, “Optical Property Measurement and Temperature Monitoring in High-Intensity Focused Ultrasound Therapy by Diffuse Optical Tomography: A Correlation Study,” Appl. Sci. 12(14), 7093 (2022). [CrossRef]
14. F. J. O. Landa, X. L. Deán-Ben, R. Sroka, and D. Razansky, “Volumetric Optoacoustic Temperature Mapping in Photothermal Therapy,” Sci. Rep. 7(1), 9695 (2017). [CrossRef]
15. V. P. Nguyen, J. Kim, K. Ha, J. Oh, and H. W. Kang, “Feasibility study on photoacoustic guidance for high-intensity focused ultrasound-induced hemostasis,” J. Biomed. Opt. 19(10), 105010 (2014). [CrossRef]
16. K. P. Morrison, G. W. Keilman, and P. J. Kaczkowski, “Single archimedean spiral close packed phased array HIFU,” IEEE Int. Ultrason. Symp. IUS400–404 (2014). [CrossRef]
17. Y. Zhang and L. Wang, “Video-Rate Ring-Array Ultrasound and Photoacoustic Tomography,” IEEE Trans. Med. Imaging 39(12), 4369–4375 (2020). [CrossRef]
18. Y. Zhang, L. Wang, and L. Wang, “Video-rate full-ring ultrasound and photoacoustic computed tomography with real-time sound speed optimization,” Biomed. Opt. Express 13(8), 4398–4413 (2022). [CrossRef]
19. Y. Zhang and L. Wang, “Adaptive dual-speed ultrasound and photoacoustic computed tomography,” Photoacoustics 27, 100380 (2022). [CrossRef]
20. L. Wang, J. Xia, J. Yao, K. I. Maslov, and L. V. Wang, “Ultrasonically encoded photoacoustic flowgraphy in biological tissue,” Phys. Rev. Lett. 111(20), 204301 (2013). [CrossRef]
21. I. V. Larina, K. V. Larin, and R. O. Esenaliev, “Real-time optoacoustic monitoring of temperature in tissues,” J. Phys. D. Appl. Phys. 38(15), 2633–2639 (2005). [CrossRef]
22. . Y. Zhou, J. Yao, and L. V. Wang, “Tutorial on photoacoustic tomography,” J. Biomed. Opt. 21(6), 061007 (2016). [CrossRef]
23. C. Liu, J. Chen, Y. Zhang, J. Zhu, and L. Wang, “Five-wavelength optical-resolution photoacoustic microscopy of blood and lymphatic vessels,” Adv. Photonics 3(01), 016002 (2021). [CrossRef]
24. Z. Qu, C. Liu, J. Zhu, Y. Zhang, Y. Zhou, and L. Wang, “Two-step proximal gradient descent algorithm for photoacoustic signal unmixing,” Photoacoustics 27, 100379 (2022). [CrossRef]
25. Y. Zhang, Y. Wang, P. Lai, and L. Wang, “Video-rate dual-modal wide-beam harmonic ultrasound and photoacoustic computed tomography,” IEEE Trans. Med. Imaging 41(3), 727–736 (2022). [CrossRef]
26. J. Zhu, C. Liu, Y. Liu, J. Chen, Y. Zhang, K. Yao, and L. Wang, “Self-fluence-compensated functional photoacoustic microscopy,” IEEE Trans. Med. Imaging 40(12), 3856–3866 (2021). [CrossRef]
