Custom-designed Verasonics-based multi-wavelength photoacoustic and ultrasound imaging: from technical advances to preclinical applications in cancer

Abstract

Real-time multi-wavelength photoacoustic (PA) imaging has emerged as a powerful modality for investigating tumor vascular dynamics, offering non-invasive, high-resolution visualization of hemodynamic changes that are critical for cancer diagnosis and therapeutic monitoring. In this study, we present a custom integration of the Verasonics Vantage ultrasound platform with an optical parametric oscillator (OPO) laser system. The first part of the study provides a comprehensive review of the current literature on PA imaging using the Verasonics system, emphasizing its capabilities, limitations, and adaptability for advanced imaging applications. Building upon this foundation, we introduce a robust framework that leverages the Verasonics system in conjunction with the fast-tuning capabilities of the OPO laser to enable synchronized, multi-wavelength PA imaging. In the second part of the study, we demonstrate selected applications of the developed system, such as multi-wavelength spectral imaging of tumor vasculature and near-real-time monitoring of therapeutic response. These case studies highlight the system’s capability to capture dynamic physiological changes and support functional, longitudinal assessments in complex pathological environments. Future directions include the exploration of alternative tunable laser sources and the integration of machine learning algorithms to enhance real-time image reconstruction and spectral unmixing. Collectively, this work establishes a versatile and scalable platform for advancing PA imaging in preclinical and translational cancer research.

1. Introduction

Hemodynamic changes are pivotal in cancer progression and in assessing the effectiveness of treatments [1,2]. These changes offer insights into blood flow, oxygenation, and overall vascular dynamics within tumors, which are critical for understanding tumor growth, metastasis, and response to therapies [3,4]. To capture these changes, there is an increasing demand for real-time imaging modalities that can provide immediate data and enable timely decisions [5–8]. Traditional imaging techniques, such as X-ray, computed tomography (CT), and positron emission tomography (PET), have long been used for cancer diagnosis and treatment planning, offering high-resolution structural and metabolic data to reveal abnormalities in tissue architecture and function. However, these methods often rely on exogenous contrast agents and ionizing radiation, which pose risks, especially with repeated use, and are generally not suitable for frequent monitoring. In contrast, non-ionizing imaging techniques, such as magnetic resonance imaging (MRI) and ultrasound (US), have become widely favored due to their safer profiles. MRI offers excellent soft tissue contrast for visualizing tumor boundaries and monitoring blood flow without radiation. However, its high cost, lengthy scan times, and stationary setup limit its broader application. Ultrasound, however, is widely accessible, affordable, and capable of delivering quick, real-time structural and functional information, which is essential in dynamic studies of tumor vasculature [9]. Photoacoustic (PA) imaging is a promising modality that further enhances functional imaging capabilities by combining ultrasound with a pulsed light source. This technique uses the photoacoustic effect, where high-energy pulsed light absorbed by a chromophore or other absorbing material causes rapid thermal expansion and contraction, generating acoustic waves. These waves are then detected by the ultrasound transducer, allowing for deeper, more detailed imaging that benefits from both optical and acoustic insights in real-time [10]. PA imaging is particularly promising for tumor imaging, where the ability to non-invasively and immediately visualize detailed vascular changes can enhance the assessment of treatment response [11–14]. Several types of PA imaging systems are in use for cancer research [15,16]. Common commercial systems include Vevo LAZR-X (VisualSonics, Fujifilm, Canada) [17–19], MSOT In-Vision (iThera Medical, Germany) [20–22], LOIS 3D (Tomowave Systems, USA) [23,24], Tritom (Photosound Technologies, USA) [25], and Acoustic X (Cyberdyne Inc., JAPAN) [26]. Cancer research encompasses both preclinical and clinical applications, requiring a diverse range of PA imaging systems tailored to specific experimental needs. The selection of an appropriate system depends on key factors such as imaging resolution, light penetration depth, and the range of wavelengths necessary for functional imaging. While several commercial PA imaging systems are available, they often lack the flexibility needed for advanced research applications, as they do not provide users with full control over system parameters. To overcome this limitation, a customizable research system is essential. Among the available ultrasound data acquisition platforms, the Vantage system by Verasonics, Inc. has been widely adapted and modified for diverse applications, including PA imaging [27]. This article mainly reviews the usability of the customizable Verasonics system for ultrasound and photoacoustic (USPA) imaging in various applications. We also demonstrate different uses of the system for multi-wavelength spectral imaging of tumor blood vessels and near-real-time monitoring of therapeutic responses.

2. Photoacoustics with Verasonics

The successful implementation of PA imaging requires precise integration of laser sources, ultrasound transducers, and data acquisition systems. The Verasonics Vantage 256 platform, widely used in preclinical and clinical research, provides a flexible framework for real-time USPA imaging. Recent studies have leveraged this system to enhance imaging capabilities, including real-time 3D imaging, multimodal integration, and functional imaging of disease pathology [28]. Figure 1 provides a comprehensive overview of the hardware and software components required for integrating the OPOTEK laser system with the Verasonics Vantage 256 platform for PA imaging. The imaging workflow begins with the configuration of multiple imaging parameters and acquisition sequences, which are programmed on the host controller within the MATLAB environment. The fundamental components of the setup script are well-documented in Verasonics’ technical resources [29]. Furthermore, recently, Kratkiewicz et al. outlined key technical considerations for implementing ultrasound and PA imaging using the Verasonics research system. This study served as a valuable reference in setting up our system, guiding the configuration of key parameters and ensuring optimal hardware-software integration [30].

Fig. 1.

A) Schematic representation of the Verasonics and OPOTEK laser hardware and software components. The Verasonics system comprises a transducer connected to an acquisition module, with a host controller responsible for configuring hardware parameters and managing the acquisition sequence. The Verasonics system operates within a MATLAB environment. The OPOTEK laser system includes laser control software capable of executing various firing sequences, including single-wavelength emission, spectral scanning, and rapid wavelength tuning. B) The sequence of events is depicted with distinct signal traces: the blue line represents the US acquisition event, the red line indicates the Flashlamp OUT trigger from the OPOTEK laser, and the green line corresponds to the PA acquisition events. The dotted line marks the data processing event, during which image buffer data is saved continuously, without having to FREEZE the VSX routine. Each PA image consisted of an average of 2 laser pulse acquisitions.

First, we present below the studies where the Verasonics system was utilized to perform PA imaging using single-wavelength illumination (summarized in Table 1). A study by Jia et al. was the earliest where a USPA and a magnetomotive system were integrated to enhance PA contrast by accumulating targeted magnetic particles at regions of interest with a 5 ns, 532 nm pulsed laser (Surelite I-20, Continuum, Santa Clara, CA) and a Verasonics system with an L7–4 linear ultrasound array [27]. Building on such foundational work, Jo et al. investigated inflammation in arthritis patients using the Verasonics (Vantage V1) system. They imaged hand joints of human subjects with active synovitis and healthy volunteers at a wavelength of 580 nm. While this system was capable of real-time imaging as presented, the acquisition frame rate was not specified. Subsequently, the group has employed dual laser systems—a tunable dye laser (ND6000, Continuum, Santa Clara, CA) powered by an Nd: YAG laser (Powerlite, Continuum, Santa Clara, CA) and an optical parametric oscillator (SLOPO, Continuum, Santa Clara, CA)—to achieve dual-wavelength imaging. However, this configuration increased costs and required meticulous energy monitoring [31,32]. Similarly, Harrison et al. demonstrated the potential of co-registered USPA imaging for precise radiotherapy delivery in prostate cancer treatment. Their study used Verasonics (VDAS-I) alongside a Surelite III pump laser paired with a Surelite OPO Plus optical parametric oscillator, achieving real-time imaging at 5 frames per second by capturing interleaved photoacoustic and flash ultrasound data [33,34]. Expanding further, Yoon et al. customized the Verasonics (Vantage 256) system for photoacoustic and plane-wave ultrasound imaging. Their setup combined a wavelength-tunable nanosecond pulsed laser (Phocus Mobile, Opotek Inc., Carlsbad, CA, USA) with a linear array transducer (L11-4 v, Verasonics Inc.), operating at a center frequency of 9.6 MHz. Using a single optical wavelength of 1064 nm, they achieved laser pulse repetition frequencies of 10 Hz [35]. Gao et al. also followed similar setups to distinguish normal from ablated tissues based on acoustic frequency analysis, employing the Verasonics (Vantage 128) system with a Phocus Mobile laser [36]. Liu et al. [37], Leng et al. [38], and Gao et al. [39] further developed systems capable of acquiring real-time ultrasound and single-wavelength PA images with linear array transducers. These systems demonstrated procedures for obtaining 3D images, leveraging motorized scanning stages [37,38] and robotic systems like the da Vinci surgical platform [39]. Similarly, Wang et al. designed a 3D USPA system with an L11-5 v linear array transducer mounted on a three-axis motorized servo stage, using a 1064 nm single-wavelength pulsed laser operating at 10 Hz for monitoring magnetically propelled microrobots [40].

Table 1. Summary of studies utilizing Verasonics system form single or multi-wavelength PA imaging.

Studies	Verasonics system model	Laser (PRF)	Laser type	Wavelength(nm)		Application	Near real-time imaging in the study?
[35]	Vantage 256	OPOTEK Phocus (10 Hz)	Tunable	Single	1064	Contrast imaging	No
[36]	Vantage 128				700	Ablated tissue
[39]	Vantage 128	OPOTEK Phocus (20 Hz)			850	Oxygenated hemoglobin	Yes
[41]	Vantage 256	OPOTEK opolette (20 Hz)			532	Vascular
[37]	-	Innolabs			800	Vascular
[35]	Vantage 256	Beijing ZK laser (100 Hz)			720	Vascular
[60]	Vantage 256	Ekspla (20 Hz)			808	Nanoparticle contrast
[40]	Vantage	Dawa-300 (10 Hz)	Single		1064	Vascular
[44]	Vantage	Laser-Export Co. (100 Hz)			1053	Contrast agent
[56]	-	OPOTEK Vibrant B (10 Hz)	Tunable	Multi	532, 1220, 1310	Hemoglobin, lipid and collagen	No
[64]	Vantage 256				640, 790, 930	Ablated tissue profile
[67]	Vantage 128	OPOTEK Phocus (20 Hz)			700-850 @10 nm steps	Ablated tissue profile	Yes
[68]	Vantage 256	OPOTEK Phocus (10 Hz)			750, 850	Contrast agent and oxygen saturation
[71]	Vantage 256				690-950 @5 nm steps	Nanoparticle contrast
[54]	Vantage 64TM	Spectra-physics, Inc. (10 Hz)			750, 800	Oxygen saturation
[55]	V1	Surelight (10 Hz)			600, 565, 576, 584	Hemoglobin and Nanoparticle contrast
[61]	-	Innolabs (20 Hz)			780,1064	Contrast agent (dye)
[50]	Vantage 256				945, 965, 985, 1005	Contrast agent (dye)
[62]	Vantage 256	Continuum (10 Hz)			545, 560, 576, 584, 605, 625	Hemoglobin, K+ concentration
[69]	Vantage 256	Surelight (10 Hz), Qutantel (10 Hz)	Multiple lasers		700, 750 and 1064	Oxygen saturation, contrast agents	Yes

To address limitations in imaging speed and volumetric acquisition, researchers explored innovative approaches. For instance, Nagaoka et al. developed a custom hemispherical transducer connected to a Verasonics (Vantage 256) system integrated with a preamplifier and a 20 Hz tunable OPO laser (Opollete, OPOTEK, CA, USA). This configuration achieved volumetric imaging speeds of 10–20 volumes per second, covering a larger field of view compared to linear arrays for the same number of laser pulses [41]. In contrast, Xing et al. developed an automated, force-controlled PA imaging system that integrated a robotic arm with the Verasonics (Vantage 256) platform and a high pulse repetition frequency (100 Hz) OPO laser. This configuration enabled real-time 2D imaging at 100 frames per second and accelerated 3D volumetric acquisitions. However, ultrasound imaging was not performed in this setup due to the use of a preamplifier, which limited concurrent ultrasound acquisition [42]. Wei et al. advanced needle biopsy guidance using a high-PRF laser with a fixed wavelength and a rotating mirror to direct light onto different fields of view. Their system achieved a real-time imaging frame rate of 30 Hz but highlighted a tradeoff between signal-to-noise ratio (SNR) and frame rate, with higher frame rates reducing SNR [43,44]. Collectively, these studies demonstrate the versatility and potency of single-wavelength photoacoustic acquisitions in a wide range of biomedical applications, establishing a strong foundation for subsequent advancements in multi-wavelength imaging approaches.

3. Multi-wavelength photoacoustic imaging

While single-wavelength PA systems are effective for fundamental imaging tasks, they are inherently limited in their ability to separate signals originating from different tissue chromophores. Because the intensity of a PA signal at any given wavelength represents the cumulative absorption of all chromophores within the illuminated region, single-wavelength imaging cannot isolate individual absorbers. Multi-wavelength PA imaging addresses this limitation by leveraging the distinct optical absorption spectra of endogenous chromophores (such as oxyhemoglobin, deoxyhemoglobin, melanin, and lipids) or exogenous contrast agents [45,46]. For example, it can distinguish between oxygenated and deoxygenated hemoglobin, enabling precise assessments of tissue oxygen saturation levels [47,48]. This capability is particularly important for monitoring dynamic physiological processes, such as blood oxygenation, and other functional and molecular markers, making it invaluable for research and clinical diagnostics [12,49]. Moreover, exogenous contrast agents such as polymer dots [50], dye-based nanoparticles [51], and others [52]have been employed to study the localization and biodistribution of these agents within regions of interest. The use of such targeted contrast agents enhances the sensitivity and specificity of PA imaging, providing deeper insights into tissue composition and functional behavior.

3.1. Functional imaging and multi-wavelength Integration

Choi et al. developed a 3D PA imaging system with a custom hemispherical transducer, Verasonics (Vantage 256), and a high-energy tunable laser (PhotoSonus M-20). Capable of 20 Hz imaging at single-wavelength and rapid switching between 750 and 850 nm, the system enabled functional oxygen saturation imaging [53]. Similarly, Okawa et al. investigated fetal oxygenation in pregnant rabbits using dual-wavelength excitation (750 and 800 nm), comparing PA imaging’s sensitivity to near-infrared spectroscopy (NIRS) and found that PA images showed high sensitivity to fetal oxygenation [54]. These studies underscore the potential of PA imaging in functional diagnostics, offering improved SNR in PA images, albeit with trade-offs in acquisition speed. Yoon et al. expanded functional imaging to study dynamic transitions in perfluorohexane nanodroplets (PFHnDs). By integrating multi-wavelength PA imaging (700–940 nm) with ultrafast ultrasound, they achieved high temporal resolution for monitoring phase transitions [35]. These studies collectively highlight the promise of multi-wavelength PA imaging for functional diagnostics, offering improved SNR and dynamic monitoring capabilities, while also being limited by technical parameters, such as laser PRF, tuning speed, and wavelength availability, which remain key challenges for achieving fully real-time imaging.

3.2. Quantitative imaging for physiological and pathological insights

Photoacoustic imaging performed with Verasonics systems has been employed in various applications, including monitoring Physiological processes, monitoring nanoparticle uptake, distribution, and therapy response. Quantitative PA imaging applications, such as tissue oxygenation and pH mapping, are advancing our understanding of physiological and pathological processes. Jo et al. employed multi-wavelength imaging to map pH within tumors using nanosensors, while also analyzing tumor hemodynamics, including total hemoglobin concentration (THb) and oxygen saturation. This study demonstrated the value of spectral unmixing in functional imaging [55]. Lei et al. applied PA imaging to distinguish inflammation and fibrosis in Crohn’s disease, leveraging wavelengths targeting hemoglobin (532 nm) and collagen (1370 nm). While offering high chemical specificity, pulse averaging limited imaging frame rates, reflecting the balance between image quality and temporal resolution [56]. Wu et al. employed spectroscopic PA imaging to evaluate changes in cartilage composition associated with varying degrees of damage. By leveraging multi-wavelength imaging, they were able to quantitatively grade the extent of cartilage degeneration based on its optical absorption spectra [57]. Similarly, multi-wavelength PA imaging has been applied to carotid plaques to assess multiple tissue components with a limited set of wavelengths. This approach provides valuable insights into plaque composition and vulnerability, enabling a more comprehensive evaluation of disease progression and risk [58].

3.3. Diagnostic imaging and theranostics

PA’s ability to visualize nanoparticle distribution and therapeutic responses has catalyzed its application in oncology and has been reviewed extensively [15,59]. However, specifically utilizing the Verasonics-based PA imaging systems, Han et al., Sun et al, and Wang et al imaged tumoral uptake of bimetallic hyaluronic acid gold-platinum nanoparticles, indocyanine green-loaded liposomes, and semiconductor polymer dots, respectively [50,60,61]. Photoacoustic chemical imaging extends PA imaging’s capabilities by visualizing chemical compositions within biological tissues. Tan et al. introduced an ionophore-based nanosensor for potassium imaging, employing six distinct wavelengths (545–625 nm) for high-resolution imaging. Although acquisition rates were not detailed, these studies highlight the promise of PA imaging in molecular imaging [62].

In interventional radiology, Lei et al. demonstrated the use of PA imaging for guiding laser ablation and biopsies with a tunable laser and interstitial optical fiber integrated with an external ultrasound transducer [63]. Meanwhile, Iskander-Rizk et al. developed a dual-wavelength fast-tuning laser technique, achieving real-time imaging comparable to single-wavelength methods, advancing therapeutic monitoring capabilities [64]. Building on prior advancements, Wu et al. and Gao et al. optimized wavelength selection and laser tuning for ablation therapy, demonstrating the value of integrating spectral and temporal features in therapeutic applications [65–67]. These advancements highlight the ongoing progress in PA imaging technology, reinforcing its potential for precision-guided interventions and therapeutic decision-making.

4. Multi-wavelength near real-time USPA imaging

Recent research has focused on integrating real-time multi-wavelength imaging with complementary imaging modalities such as high-frame-rate Doppler and Ultrasound Localized Microscopy (ULM). Notably, studies by Zhao et al. and Tang et al. have combined ULM with USPA to generate detailed vascular maps [68,69]. While these studies primarily focused on ULM, they also developed real-time multi-wavelength PA imaging schemes to capture transient changes in blood flow and tissue oxygenation. Specifically, Zhao et al. employed a tunable laser and performed sequential PA image acquisitions at 700 nm and 750 nm, while Tang et al. utilized multiple single-wavelength lasers with a 200 µs firing delay, synchronized with the Verasonics system for image acquisition.

A more recent study by Chen et al. introduced a multimodal imaging framework that integrates fast Doppler ultrasound with multi-wavelength PA imaging, further advancing the capabilities of real-time functional imaging [70]. Although this study successfully demonstrated real-time multimodal imaging, its applications have been largely limited to brain imaging. However, in pathological tissues, such as tumors, hemodynamics and tissue oxygenation are often impaired and highly heterogeneous. Our framework closely follows Chen et al. approach, leveraging the fast-tuning capability of the laser to minimize the transition time between wavelengths. This feature enabled consecutive acquisition of PA images at different wavelengths, facilitating the capture of rapid oxygenation changes in the tumor. A key advantage of multi-wavelength PA imaging is its ability to measure molecular-specific biomarkers and map complex biological processes in real-time. This capacity enables researchers and clinicians to observe and evaluate the interactions of different biomolecules within tissues, including those involved in disease progression, treatment response, and molecular signaling pathways. Real-time, multi-wavelength imaging allows for the continuous tracking of these processes, providing a more accurate, comprehensive understanding of the tissue's dynamic state. The need for real-time imaging is particularly critical when assessing functional parameters such as tissue oxygenation or metabolic activity. These parameters are highly sensitive and can fluctuate rapidly in response to various physiological stimuli or treatment interventions. Real-time multi-wavelength imaging ensures that these transient changes are captured before they alter, providing a true reflection of tissue conditions.

In the next part of this article, we focus on leveraging multi-wavelength PA imaging to investigate pre-clinical cancer models, aiming to capture subtle spatial and temporal hemodynamic variations in PA signals. Additionally, we demonstrate the application of real-time 3D and spectral imaging, highlighting its potential for tumor functional imaging in dynamic pathological environments.

5. Framework for combined ultrasound and multi-wavelength photoacoustic imaging

The integrated USPA imaging system consisted of a Vantage 256 platform (Verasonics, Redmond, WA, USA) and an optical parametric oscillator (OPO) laser (Phocus HE Mobile, OPOTEK, Carlsbad, CA, USA), operating at a PRF of 10 Hz with a tunable wavelength range of 690–950 nm. Figure 1(A) provides a comprehensive overview of the hardware and software components required for integrating the OPOTEK laser system with the Vantage 256 platform for PA imaging. The imaging workflow begins with the configuration of multiple imaging parameters and acquisition sequences, which are programmed on the host controller within the MATLAB environment. Once the VSX software sequence is initiated, it manages system events and data transfer as specified in the script. The Vantage system communicates with the host computer via a high-speed PCIe Express cable, ensuring rapid data transmission. On the system’s output end, a UTA-360 is mounted to facilitate the use of a custom transducer used in our study. A 256-element LZ250-S linear transducer (VisualSonics, FujiFilm; 21 MHz), designed with integrated optical fibers, delivers focused light approximately 10 mm from the transducer elements. For light delivery, the fiber bundle integrated with the transducer was directly coupled to the laser’s output port, ensuring precise and efficient illumination.

As illustrated in Fig. 1(A), the OPOTEK laser system supports external triggering of laser pulses through both flash lamp and Q-switch input ports. For single-wavelength imaging, these external triggering functionalities can be effectively utilized, as demonstrated by Kratkiewicz et al. However, for multi-wavelength imaging, a key limitation arises: the OPOTEK system does not offer an external control interface for wavelength switching. Consequently, the Verasonics system cannot inherently identify which wavelength is being used during each PA acquisition and responds to the flashlamp trigger signal. The OPOTEK laser system includes a user-friendly control software interface that enables fine wavelength tuning and laser parameter adjustments. This software is essential for executing multi-wavelength fast tuning and spectral scans, allowing users to program a preselected sequence of wavelengths. In addition, an important limitation to consider is that the fast-tuning functionality (i.e., rapidly switching between wavelengths) is exclusively accessible only through the OPOTEK software interface, preventing the use of external triggers for wavelength switching. To optimize synchronization and mitigate challenges associated with external laser triggering during fast wavelength tuning, we implemented a routine in which the laser’s flash lamp trigger output was directly interfaced with the Verasonics system’s trigger input, as shown in Fig. 1(B). At each programmed wavelength, PA images were generated by averaging signals from two consecutive laser pulses (Appendix 1, yellow highlight). This approach ensured consistent signal acquisition and minimized variability, thereby improving the SNR without compromising temporal resolution. For spectral scanning applications, the OPOTEK laser system was configured to sweep from 690 to 950 nm wavelengths in 10-nm increments, following a firing sequence such as 690, 690, 700, 700, …, 950, 950 nm, respectively. This systematic wavelength progression enabled comprehensive multi-wavelength imaging across the near-infrared spectrum. For functional imaging tasks, such as blood oxygen saturation mapping, the fast-tuning feature of the OPOTEK laser was utilized to rapidly alternate between two preselected wavelengths, specifically 750 nm and 850 nm. These wavelengths were chosen based on the differential absorption characteristics of oxyhemoglobin and deoxyhemoglobin, allowing for dynamic monitoring of tissue oxygenation. It is important to note that increasing the number of frame averages can further enhance image quality; however, this comes at the cost of longer acquisition times. Such adjustments may not be suitable for applications requiring real-time imaging of transient physiological changes, such as oxygen saturation dynamics. Therefore, frame averaging settings should be carefully optimized based on the specific imaging goals and temporal constraints of the study.

The imaging sequence begins with US image acquisitions (represented in blue), ensuring structural imaging is obtained before the laser pulse is initiated. Once the US acquisitions are completed, the system awaits the Flashlamp OUT trigger signal (red), which is generated by the laser to indicate pulse generation. The laser used in this study exhibited a flashlamp-to-Q-switch delay, a characteristic common to OPO-based lasers, which can vary depending on the specific laser and its operating parameters. As a result, users must adjust the timing sequence accordingly to ensure proper synchronization. Following a predefined delay, PA images are acquired (represented in green), capturing the tissue’s optical absorption characteristics. To enhance the SNR while maintaining real-time imaging capabilities, two PA frames/acquisitions of the same wavelength are averaged to generate a single image. After acquiring PA data at one wavelength, the laser transitions to the next predefined wavelength and emits two consecutive pulses, forming the subsequent PA images. Meanwhile, the data transfer process (represented by the dotted line) ensures real-time storage of acquired signals into the host computer memory.

Recent versions of the Verasonics system introduced an updated, user-friendly interface for ultrasound imaging, allowing simultaneous data acquisition and storage. However, this functionality is restricted to Verasonics transducers. When working with non-Verasonics transducers and when saving data from RF, IQ, or image buffers, the VSX system must be in freeze mode, which can limit acquisition rates, particularly for multi-wavelength imaging. To address this limitation, we utilized a recently introduced process structure feature known as the ‘storage’ class. By implementing the ‘storage’ process class, beamformed US and PA images were continuously saved to a predetermined location without requiring the system to freeze (see Process structure in Appendix 1). The maximum number of images or files saved was defined using the maxnumfiles variable (Appendix 1, blue highlight), which can be adjusted based on user or application-specific requirements. For instance, setting maxnumfiles  =  2000 allows continuous data saving for 400 seconds with a 2-pulse average PA image processing. It is important to note that the saved files do not contain PA wavelength information. Therefore, the image saving sequence and laser initialization sequence were manually timed to start simultaneously, enabling users to correlate the acquired images with the corresponding wavelengths during post-processing.

In the current version of the Verasonics system, stored files are restricted to the proprietary .vrs binary format, which encapsulates both metadata and imaging data. Verasonics provides dedicated functions to process .vrs files, enabling users to extract metadata and retrieve the corresponding USPA imaging data for further analysis. This approach allows real-time data storage and extends compatibility to non-Verasonics transducers, enhancing the system’s flexibility for various imaging applications.

6. Applications

6.1. 3D. PA spectroscopy phantom imaging

Using the imaging setup described above, along with the laser's PRF constraints, we achieved a 5 Hz frame rate for co-registered USPA imaging. This was achieved by averaging two laser pulses per PA image per wavelength. For spectral scanning experiments, a custom 3D-printed water bath was designed to facilitate the precise alignment of four transparent polyethylene tubes (inner diameter: 1.4 mm; outer diameter: 1.9 mm; PE200, Intramedic), as illustrated in Fig. 2(A). These tubes were filled with distinct chromophore solutions to evaluate their spectral responses. Bovine blood (HbO₂) was prepared by dissolving powdered hemoglobin (H2500, Sigma-Aldrich) in 1× phosphate-buffered saline (PBS) to a final concentration of 2.5 mM. To obtain deoxygenated blood (Hb) samples, 2.5 mg/mL sodium dithionite was added to facilitate oxygen depletion. A stock solution of synthetic melanin (1 µg/mL; Thermo Fisher Scientific) was prepared in deionized water and sonicated to ensure uniform particle dispersion. Additionally, a specialized photoacoustic naphthalocyanine dye, SiNC-1 (referred to hereafter as NC1 dye), provided by Saad et al., was diluted to a working concentration of 10 µM in dimethyl sulfoxide (DMSO). All samples were loaded into the aligned tubes within the water bath to ensure optimal acoustic coupling during spectral scans [51,72]. During spectral imaging, the maximum number of saved images was capped at 75 to accommodate blank acquisitions at the start and repeated cycles over the entire wavelength sweep. Each sample tube was positioned individually and scanned multiple times.

Fig. 2.

A) Schematic of experimental setup with various tubes in a water bath. B) PA images (4.5 × 3.5 mm) with corresponding US images of various chromophores at different wavelength. The arrows indicate the change in signal intensity of oxygenated blood (HbO₂), deoxygenated blood (Hb), melanin and NC1 dye in red, blue, orange and green color maps respectively. C) A normalized absorption spectrum of the chromophores from literature. Scale bar: 1 mm

As depicted in Fig. 2(B), the oxygenated blood, visualized using a red colormap, demonstrated a gradual increase in PA signal intensity with increasing wavelengths, as highlighted by the red arrows. Conversely, deoxygenated blood, represented in a blue colormap, showed a decline in PA signal intensity with increasing wavelengths, as indicated by the blue arrows. These observations align with the well-characterized absorption spectra of hemoglobin, where deoxygenated blood exhibits higher absorption at 750 nm compared to oxygenated blood, while oxygenated blood demonstrates higher absorption at 850 nm.

Melanin, with its relatively linear absorption profile across the 690–950 nm range, exhibited no significant variation in PA signal intensity. However, a reduction in PA signal from the upper portion of the melanin-containing tube was observed, likely caused by particle sedimentation over the imaging period. Additionally, the PA contrast agent, NC1 dye, characterized to have a strong absorption peak around 870 nm, showed a pronounced increase in PA signal at 870 nm compared to 750 nm, as marked by the green arrows. Overall, the PA spectral data closely matched UV-VIS spectroscopy measurements and previously reported literature values, demonstrating the system's ability to perform spectral scans as seen in Fig. 2(C).

For 3D imaging, a linear ultrasound transducer was mounted on a precision motorized linear stage (X-LSM, Zaber) to enable controlled scanning. The stage was programmed to move at a constant speed of 2.5 mm/s while synchronized with the acquisition of dual-wavelength photoacoustic and ultrasound images. To assess oxygenation-dependent PA signal variations, oxygenated and deoxygenated blood samples, prepared as described earlier, were loaded into polyethylene tubes and arranged in an “X” configuration within a custom 3D-printed tube holder. The tubes were submerged in the water bath to ensure consistent acoustic propagation, and the transducer performed a linear scan across the sample, acquiring volumetric data. For the estimation of tissue oxygen saturation (StO₂) throughout the study, a custom MATLAB-based toolkit, PHANTOM, was employed for image segmentation. Following segmentation, StO₂ values were quantified using a simple linear unmixing algorithm, as described by Sweeney et al. This method enabled accurate decomposition of multispectral PA signals into their constituent chromophores, allowing for precise characterization of oxygenation dynamics in the imaging setup [72].

During dual-wavelength 3D imaging, a static homogeneous solution of oxygenated and deoxygenated blood was scanned at a linear stage speed of 2.5 mm/s. The maximum number of frames collected was set to 250, corresponding to a total scan distance of 10 mm. Data acquired at dual wavelengths were separated from the complete scan dataset for display and subsequent unmixing to estimate StO₂. As observed in previous experiments, deoxygenated blood exhibited a higher photoacoustic signal at 750 nm compared to 850 nm, while oxygenated blood showed the reverse trend, as demonstrated in Fig. 3. However, the PA intensity differences between oxygenated blood at 750 nm and 850 nm were not pronounced, potentially due to the spatial positioning of the tubes. Specifically, the oxygenated blood tube was positioned below the deoxygenated tube, potentially reducing its light illumination. Upon spectral unmixing of the 750 nm and 850 nm signals, the reconstructed StO₂ map corresponded to the expected blood oxygenation levels. Regions with high oxygenation were represented in red, while areas with lower oxygenation appeared in blue, hence exhibiting the application of blood oxygen imaging while performing a 3D scan.

Fig. 3.

3D photoacoustic (PA) images of oxygenated and deoxygenated blood in tubes arranged in an ‘X’ at 750 nm and 850 nm, along with the corresponding linear unmixed oxygen saturation (StO₂) map. The PA signal for deoxygenated blood is higher at 750 nm due to the greater optical absorption of deoxygenated hemoglobin compared to oxygenated hemoglobin at this wavelength.

6.2. Real-time hemodynamic changes

To investigate real-time dynamic changes in tissue oxygen saturation (StO₂), an air-oxygen breathing challenge was administered. All animal procedures performed in this study were approved by the Institutional Animal Care and Use Committee (IACUC) at Tufts University. A male homozygous Foxn1nu nude mouse (The Jackson Laboratory), aged 6–8 weeks, was subcutaneously injected with 5 × 10⁶ MIA PaCa-2 cells suspended in 100 µL of Matrigel (comprising 50 µL Matrigel and 50 µL PBS). When the tumor reached an approximate volume of 60 mm³, the mouse was anesthetized with 2% isoflurane in air and positioned on a heating pad under a transducer. To ensure proper acoustic coupling, bubble-free ultrasound transmission gel (Aquasonic 100 Ultrasonic Transmission Gel, Parker Laboratories, Inc.) was applied directly to the tumor site. The optimal center frame was identified for real-time imaging and monitoring.

During the imaging session, the mouse remained under isoflurane anesthesia. Initially, the mouse breathed air. At the 3-minute mark, the air supply was replaced with 100% oxygen while maintaining a constant isoflurane flow rate. After 5 minutes of oxygen exposure, the supply was switched back to air. This cycle of alternating between oxygen and air was repeated three times, with each phase lasting 5 minutes, to study tumor response under varying oxygenation conditions. All in vivo data processing was performed using MATLAB. Initially, raw images acquired at 750 nm and 850 nm were pre-processed by applying a threshold to eliminate background noise. Following this, the images were spectrally unmixed to separate distinct chromophore contributions. Tumor regions were identified, segmented, and subsequently overlaid on the unmixed images to visualize localized chromophore distributions effectively. A 5 × 5 median filter (medfilt2) was then applied to the data, ensuring noise reduction and spatial smoothing of the StO₂ map.

As shown in Fig. 4(A), a noticeable difference in StO₂ was observed when the mouse was breathing air (yellow background) versus 100% oxygen (blue background). The plot illustrates the average StO₂ value within the segmented tumor region (in grey), with a Savitzky-Golay filter applied to smooth the data over time. As depicted in Fig. 4(B), a two-tailed t-test was performed on the average StO₂ values recorded during air and oxygen administration, revealing a statistically significant difference (p < 0.05) in tumor StO₂. Figure 4(C) presents the corresponding StO₂ maps at selected time points. During air breathing (minutes 1, 10, and 20), lower StO₂ levels were evident within the tumor region. Conversely, during 100% oxygen breathing (minutes 6, 15, and 25), there was a marked increase in oxygenation (indicated by a rise in red intensity on the StO₂ maps). Visualization 1 ^{(29.6MB, mp4)} shows the real-time changes in oxygen saturation with time. These results demonstrate the capability of the imaging system to monitor real-time oxygen saturation dynamics and record the response of tumor tissues to altered oxygenation conditions.

Fig. 4.

A) Change in StO₂ was observed when the mouse was breathing air (yellow background) versus 100% oxygen (blue background). The plot illustrates the average StO₂ value within the segmented tumor region (in grey), with a Savitzky-Golay filter applied to smoothen the data over time. B) A simple t-test shows a statistically significant difference (p < 0.05) in tumor StO₂ during different cycles. C) The corresponding StO₂ maps at selected time points while breathing air (minutes 1, 10, and 20) and 100% oxygen (minutes 6, 15, and 25). Visualization 1 ^{(29.6MB, mp4)} shows the dynamic oxygen saturation in tumor cross-section over the 3 cycles of air and oxygen inhalation. Scale bar – 1mm

6.3. Monitoring treatment response

A six-week-old male Foxn1nu nude mouse (The Jackson Laboratory) was used for tumor model development. A subcutaneous injection of 3 × 10⁶ U87 cells in a 100 µL suspension (1:1 mixture of Matrigel and phosphate-buffered saline) was performed using a 1-mL tuberculin syringe with a 28-gauge hypodermic needle. The U87 cells were cultured in suspension with media supplemented with 10% fetal bovine serum and 1% penicillin-streptomycin (100 U/mL) and maintained in a 37°C incubator under 5% CO₂. When the tumor volume reached approximately 120 mm³, the mouse was anesthetized using isoflurane gas (2–3% for induction and 1.5% for maintenance) and placed on a temperature-controlled heating pad to maintain body temperature. Throughout the imaging session, the mouse's respiration was continuously monitored to ensure physiological stability. To facilitate effective acoustic coupling between the ultrasound transducer and the tumor, a layer of bubble-free ultrasound transmission gel was applied directly to the tumor site. Real-time monitoring with PA imaging began 5 minutes before treatment to obtain a baseline and continued for an additional 15 minutes after the cessation of the light dose. Photodynamic therapy (PDT) was performed as described by Langley et al. [73]. Liposomal benzoporphyrin derivative (BPD) at a dose of 0.5 mg/kg BPD eq. was administered intravenously into the mouse via the tail vein, and the mouse was kept in the dark for a 90-minute drug-light interval before irradiation. The light dose was administered with a 690 nm continuous-wave laser (Modulight Corporation) with a custom four-way split optical fiber bundle (Thorlabs, Inc.). Each of the four fibers was mounted to the transducer with a custom-designed 3D printed attachment, angled 10 degrees from the mouse’s frontal plane. The vertical suspension height of the PDT fibers was adjusted to irradiate tissue in the photoacoustic focal region of the transducer through modification of the fiber holder from our previous work. The tumor was irradiated for 1,000 seconds at a fluence rate of 400 mW/cm² for a total light dose of 400 J/cm². Power measurements were taken with a 12 × 12 mm square optical sensor (Ophir Starbright) to achieve the desired PDT fluence rate. An analogous 3D printed fiber holder that allowed the optical sensor to be rotated about a central axis was used to acquire the output of each of the four fibers, and the laser power was adjusted until the total output of all fibers summed to 400 mW/cm².

The transducer, coupled with a custom 3D-printed attachment, was securely mounted onto a linear stage, as illustrated in Fig. 5(A). Ninety minutes after the injection of liposomal BPD into the mouse, a 3D ultrasound scan was performed to identify the tumor's central plane for real-time monitoring, as shown in Fig. 5(B). Once the center frame was selected, a continuous acquisition script was initiated, enabling the saving of up to 2000 frames for subsequent analysis. Visualization 2 ^{(33.5MB, mp4)} presents StO₂ maps overlaid on ultrasound images, capturing dynamic changes throughout the entire monitoring period.

Fig. 5.

A, B) Illustrate the experimental setup and tumor orientation during imaging. C) Displays images acquired at 750 nm and 850 nm, along with the corresponding oxygen saturation (StO₂) maps at specific time points: minute 3 (pre-PDT), minute 5 (onset of PDT laser activation), minute 8 (minimum StO₂ during treatment), and minute 22 (immediately after PDT laser deactivation) during real-time monitoring. D) Presents a quantitative analysis of the Savitzky-Golay smoothed tumor StO₂ over the real-time monitoring period, overlaid on the averaged tumor StO₂. Visualization 2 ^{(33.5MB, mp4)} shows the real-time oxygen saturation changes with PDT. Scale bar – 1mm

Figure 5(C) showcases images acquired at 750 nm, 850 nm, and corresponding oxygen saturation (StO₂) maps from specific time points during the real-time monitoring session. At minute 3 (pre-PDT), the tumor exhibits baseline oxygen saturation. At minute 5, the PDT laser was activated, initiating photodynamic therapy, followed by a significant drop in StO₂ at Minute 8, marking the point of minimum oxygen saturation. By Minute 22 (post-PDT), partial recovery in oxygenation was observed. The StO₂ maps clearly reveal dynamic changes, with areas represented in red showing a pronounced decrease in oxygen saturation as treatment progressed. These findings align with observations reported by Langley et al., where a sharp decline in StO₂ was followed by a gradual recovery post-treatment [73].

Figure 5(D) provides a quantitative representation of the average tumor StO₂ over the real-time monitoring period. The results highlight the significant and rapid drop in tumor oxygenation following PDT activation, followed by a gradual increase, demonstrating the temporal dynamics of oxygen saturation during and after PDT treatment. Furthermore, real-time monitoring of PDT treatment is critical, as it enables the ability to predict treatment response and optimize therapeutic outcomes.

7. Discussion and conclusion

Multi-wavelength photoacoustic imaging has emerged as a powerful modality with extensive applications in biomedical imaging. Recent clinical investigations, particularly in breast cancer screening [74], melanoma detection [75], and thyroid cancer assessment [75], highlight the potential of advancing this technology toward routine clinical use [76,77]. However, achieving accurate and reproducible imaging outcomes is highly dependent on standardized protocols, careful image acquisition, and robust processing methodologies. Alongside the expansion of PA imaging applications, significant progress has been made in overcoming its intrinsic limitations, such as spectral coloring, limited spatial resolution, and the complexity of chromophore-specific unmixing. Addressing these challenges is critical for improving quantitative accuracy and clinical reliability.

One promising direction is the integration of advanced noise reduction algorithms, which can enhance the signal-to-noise ratio (SNR) without compromising imaging speed. For instance, Paul et al. demonstrated a four-fold improvement in SNR using a U-Net-based framework for photoacoustic image denoising [78–80]. Similarly, other machine learning based algorithms [81] and deep learning architectures, including GAN-based convolutional neural networks (CNNs) [82], Noise2Noise models [83], and modified U-Net architectures [84], have shown effectiveness in both denoising and artifact removal in PA imaging. In addition to noise suppression, spectral unmixing and correction of spectral coloring remain active areas of research. Although many current unmixing strategies were originally developed for hyperspectral optical imaging [85–87], these algorithms that are combined with deep learning frameworks could be adapted for PA imaging to improve chromophore-specific accuracy and quantitative mapping [88–90].

On the other hand of hardware, one of the primary challenges in real-time multi-wavelength PA imaging is the inherent limitations of current laser systems, particularly concerning wavelength tuning speed, pulse stability, and energy output [91]. Recent advancements, such as the introduction of the Verasonics Vantage NXT system, have improved digital acquisition rates, thereby reducing pulse jitter noise in PA images. Additionally, the integration of high-performance pre-amplifiers, such as those developed by Photosound, can significantly enhance the SNR without necessitating image averaging, thus preserving real-time imaging capabilities. These developments mark important steps toward more robust and efficient multi-wavelength PA imaging, facilitating broader clinical and research applications.

Through this review, we have examined relevant literature on tunable-wavelength lasers integrated with Verasonics systems, highlighting the need for a practical approach to multi-wavelength PA imaging that can facilitate near real-time imaging without extensive hardware modifications. While several approaches have been proposed, many require complex system adaptations that may limit feasibility. To illustrate a potential solution, we present a framework that demonstrates how multi-wavelength PA imaging can be implemented in a streamlined manner. This framework is intended to showcase possibilities rather than to serve as a formal validation, providing examples of capturing dynamic tissue oxygenation changes and other functional imaging applications. It is important to acknowledge that multiple approaches exist to achieve real-time multi-wavelength PA imaging, each with its own advantages and constraints. Our work integrates a tunable OPO-based laser system, specifically the Opotek Phocus Mobile laser, to illustrate the feasibility of real-time multi-wavelength imaging. Additionally, to facilitate reproducibility and adaptability across different experimental setups, we provide a pseudo-script in the appendix. This script outlines the ‘event’ setup and the continuous data-saving process, allowing researchers to tailor the framework according to their specific applications and the available laser system features. While this approach offers a streamlined solution, future studies may explore the integration of alternative tunable laser systems, such as fiber lasers or LED-based systems, to further optimize real-time imaging performance [92].

In summary, while real-time multi-wavelength PA imaging remains an evolving field with ongoing technological advancements, our work presents a viable framework for its implementation using commercially available laser systems. Future research efforts could focus on refining tuning mechanisms, enhancing synchronization protocols, and integrating machine learning algorithms for optimized spectral image processing. By addressing these challenges, the field can move closer to achieving near-real-time multi-wavelength imaging across pre-clinical and clinical applications.

Appendix I

The script was adapted from the example provided by Verasonics and recommendations given by Kratkiewicz et al. [30]. As mentioned previously, this script was configured for Vantage 256 and Phocus mobile laser. Other pulsed lasers with rapid wavelength-tuning capabilities, such as OPO-based systems from Quantel, Ekspla, or Innolas, may require alternative frameworks due to differences in their communication protocols, synchronization methods, or triggering mechanisms. The framework described here has been specifically developed and tested for the OPOTEK Phocus laser system. Below are important pseudo-codes that are essential for continuous saving. Do note that this is not a full running script. These sections need to be added to a functioning PA script to achieve continuous data saving.

Funding

National Institutes of Health 10.13039/100000002 ( R01CA231606, R01CA266701, S10OD026844); Tufts University 10.13039/100008090 ( UL1TR002544, Tiampo family professorship award).

Disclosures

The authors declare no conflict of interest.

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.