The sound of blood: photoacoustic imaging in blood analysis

Abstract

Blood analysis is a ubiquitous and critical aspect of modern medicine. Analyzing blood samples requires invasive techniques, various testing systems, and samples are limited to relatively small volumes. Photoacoustic imaging (PAI) is a novel imaging modality that utilizes non-ionizing energy that shows promise as an alternative to current methods. This paper seeks to review current applications of PAI in blood analysis for clinical use. Furthermore, we discuss obstacles to implementation and future directions to overcome these challenges. Firstly, we discuss three applications to cellular analysis of blood: sickle cell, bacteria, and circulating tumor cell detection. We then discuss applications to the analysis of blood plasma, including glucose detection and anticoagulation quantification. As such, we hope this article will serve as inspiration for PAI’s potential application in blood analysis and prompt further studies to ultimately implement PAI into clinical practice.

1. Introduction

The human body contains up to 10% blood by weight [1]. Blood is a complex fluid composed of water, cells, metabolites, electrolytes, proteins, and hormones. Furthermore, it permeates, affects, and is affected by every organ of the body. Its functions range broadly from transportation of catabolic fuel and removal of metabolic waste to healing and microbial defense. Thus, testing of this omnipresent fluid can provide vast information regarding the body’s physiological and pathophysiologic status.

Obtaining and testing samples of blood provides invaluable information; however, it comes with drawbacks. Blood sampling requires invasive procedures which can be uncomfortable and result in adverse events [2]. In preclinical settings, test subject size (e.g. mouse models) can make blood collection difficult [3]. In addition, laboratory analysis of blood components involves chemical reactions and/or enzyme assays which can result in differences from in-vivo environments [4–6]. Furthermore, testing for specific blood components can require separate, complex, and expensive equipment [7]. Finally, the small sample volume used in blood tests make it challenging to detect rare events such as circulating tumor cells in early metastasis [8].

Early technological advancements have transitioned a few specific blood measurements to non-invasive methods but are rather limited in their sensitivity [9–11]. Beyond these few cases, many desired blood measurements remain inestimable to existing biosensors. Photoacoustic imaging (PAI) is a developing technology that may play a role in future applications of blood analysis in clinical and preclinical realms. Due to its seamless combination of optical contrast and acoustic detection, this technology may provide methods of non-invasive blood analysis that current methods cannot. Using its deep-imaging properties beyond the skin surface, this technology demonstrates promise in reducing patient/subject harm and improving in-vivo analysis.

PAI operates via the photoacoustic effect. A molecule absorbs an electromagnetic wave in the form of a laser pulse, resulting in a temperature rise. This temperature rise induces a thermoelastic expansion of the molecule of interest causing ultrasonic waves to propagate outwards [12]. The waves are detected by an ultrasound transducer and reconstructed into an image that maps the original optical energy deposition, i.e. optical absorption. These PA waves can originate from both endogenous (particularly hemoglobin) and exogenous contrast agents such as nanoparticles, making for a wide range of imaging targets [13,14]. Furthermore, PAI benefits from the tunability of the image contrast through the selection of excitation laser wavelength, allowing for many molecules to be independently imaged using the same imaging system [15].

Due to the unique combination of both optics and acoustics, PAI has inherited the advantages of pure optical and ultrasound imaging systems. Similar to optical imaging systems such as optical coherence tomography and confocal microscopy, PAI systems can achieve high-resolution, high frame-rate, and strong contrast and sensitivity, all using non-ionizing radiation [16]. Moreover, due to weak acoustic attenuation of the biological tissues, the imaging depth of PAI is usually greater than pure optical imaging systems [17]. Compared with pure ultrasound imaging, PAI has inherent functional and molecular sensitivity to the tissue status based on the rich optical absorption contrast. These properties make PAI a versatile technology with a wide range of preclinical and clinical applications. (See Table 1).

Table I.

Summary of example optical imaging methods with their lateral resolution and depth limitations. Photoacoustic Microscopy (PAM).

Imaging Modality	Resolution (Lateral)	Depth
Confocal Microscopy	180 nm [18]	100 μm [19]
Optical Coherence Tomography	~5 μm [20]	3 mm [21]
Raman Scattering Microscopy	300 nm – 2 μm [22]	20 μm–100 μm [23]
Optical Resolution PAM	2.6 μm [24]	1.2 mm [24]
Acoustic Resolution PAM	45 – 590 μm [25]	3 – 10.3 mm [25]
Photoacoustic Tomography (dependent on array used)	100 – 720 μm [26]	10 – 70 mm [26]

The specific performance of a PAI system depends on the system’s optical and acoustic configurations. Although there are more variations of imaging modalities than listed here, this paper will provide a brief overview of three broad groups of PAI systems: optical resolution photoacoustic microscopy (OR-PAM), acoustic resolution photoacoustic microscopy (AR-PAM) and photoacoustic computed tomography (PACT). As seen in Fig. 1, each modality relies on the photoacoustic effect but varies in the design based upon the desired imaging depth and spatial resolution. The readers are encouraged to refer to papers regarding the specific methods with more detailed descriptions [16,25,27–30].

Fig. 1.

Overview of CFM (confocal microscopy), TFM (two-photon microscopy), OR-PAM, AR-PAM, and PACT, produced by Wang and Yao [31].

OR-PAM provides the best imaging resolution and relies mostly on tight focusing of ballistic photons. Images by OR-PAM are limited to approximately 1 mm of depth (the optical diffusion limit in soft tissues), but can achieve an axial resolution of tens of microns and a lateral resolution of several microns or even sub-microns [29]. This imaging modality serves well for histology of tissue slices and can provide single-cell or organelle-level information

AR-PAM uses a wide-field optical excitation and relies on diffusive photons, which allows for greater penetration depths. Although the penetration depth is increased, the resolution becomes limited by the focused detection of high-frequency acoustic signals. Therefore, the resolution of AR-PAM systems depends on the focused transducer central-frequency, bandwidth, and focusing geometry [25]. For example, AR-PAM can use a 5 MHz transducer to image at a depth of 10.3 mm with a lateral resolution of 590 μm. Conversely, with a transducer at 40 MHz, AR-PAM is able to obtain a lateral resolution of 85 μm at a depth of 3.1 mm. Periyasamy et al. used a 30 MHz transducer and achieved a 57 μm axial resolution at a depth of 11 mm, whereas Moothanchery et al. used a 75 MHz transducer producing an axial resolution of 53 μm at a depth of 1.8 mm [27,32]. These studies provide examples of the wide range of resolutions AR-PAM can produce depending on the acoustic configurations.

PACT utilizes wide-field optical excitation in the diffusive regime, low-frequency array transducers, and inverse reconstruction algorithms to provide a tomographic image [28]. PACT systems can provide rapid 2D or 3D imaging with resolutions at the sub-millimeter level. A wide range of ultrasound transducer arrays have been explored by PACT for signal detection including linear arrays, ring arrays, and 2D planar or hemispherical arrays [33–35]. These imaging systems have various spatial resolution and imaging depths, which are jointly determined by the acoustic detection geometry and frequency. Because PACT often utilizes array transducers for parallel signal detection, PACT can acquire an image much faster than traditional PAM [36].

The optimal selection of PAI system subtype depends on the characteristics of the targets to be imaged. While depth and resolution are important factors to be considered, the target must also either have an endogenous chromophore or be labeled with an exogenous chromophore. Endogenous chromophores such as hemoglobin have an intrinsic PA contrast, providing label-free vascular imaging [37,38]. Conversely, some targets may require labeling with nanoparticles and/or organic dyes to produce contrast of the desired target [13,14].

As will be demonstrated throughout this paper, PACT and AR-PAM play a particularly important role in in-vivo blood analysis due to their depth advantage over traditional optical imaging systems that are usually limited to superficial depths of analysis (<2 mm) [39]. Skin anatomy reveals capillary vessel depth resides around 1–4 mm, whereas venules and arterioles lie below this depth, which are out of reach for optical imaging systems. Although capillaries play a significant role in certain pathophysiology, critical information may be lost at the venule/arteriole level. Photoacoustic imaging provides a solution to this limitation through utilization of diffusive photons and acoustic signals.

Technical innovations in photoacoustic imaging continue to improve resolution, decrease acquisition time, and produce novel molecular contrasts, opening new possibilities for biomedical applications [40]. Its non-ionizing, acoustic and optical capabilities make it well poised for use in in-vivo settings.

While not exhaustive, this paper seeks to provide a review of current techniques applicable to blood analysis using PAI. We have organized this paper to parallel that of blood composition as seen in Fig. 2. First, we provide examples of applications to the cellular component of blood including: red blood cell pathology, bacterial pathogen detection and malignant cell identification. We then transition to examining applications to the acellular (plasma) component including: pharmacologic analysis and electrolyte quantification. PAI’s potential is vast and accordingly we hope this paper helps establish the significance of PAI in blood analysis and provide inspiration for future work in developing this technology.

Fig. 2.

Components of whole blood. Plasma is predominantly composed of water but contains many other active components. In addition to the listed components above, plasma contains diffused gasses, hormones, metabolites, waste products among others.

2. Applications to cellular components in blood analysis

Traditional hematological histopathology involves blood smear analysis with the use of light microscopy and hematoxylin and eosin (H&E) staining [41]. Although historically useful, traditional methods are limited in throughput and the staining can affect molecular processes [42]. Photoacoustic imaging shows promise for alternative methods for observing cellular components of the blood both in-vitro and in-vivo. This section of the paper reviews the current methods, advantages, and limitations regarding PAI’s application to imaging several cellular components of blood associated with sickle cell disease, pathogen detection, and circulating tumor cells.

2.1. Sickle cell disease

Sickle cell disease is a pathology of red blood cells (RBC), whereby a mutation in hemoglobin (Hb) results in deformation of cells in the form of a sickle. This deformation is thus known as sickling of the cells. This sickling is followed by two important pathophysiologic changes. First, a decrease in oxygen saturation of Hb leading to decreased oxygen delivery to the tissues [43]. Second, a clustering of cells impeding blood flow and thus tissue perfusion [44]. The clinical sequelae of sickle crises vary, but early diagnosis is critical for the prevention and possible reversal of this process.

One common complication of sickle cell disease is a pain crisis. These pain crises are a result of micro vaso-occlusions, the etiology of which varies. Currently, there are no objective detection methods of these vaso-occlusions. Thus, diagnosis is made on the basis of subjective patient information which makes episodes difficult to predict and treat [45]. Without objective evidence physicians can easily under- or over-treat symptoms [46]. Objective detection of vaso-occlusions could improve diagnosis and thus improve treatment regimens.

Photoacoustic imaging may provide a method for this detection. Initial studies utilized PAI to demonstrate the ability to detect erythrocyte changes using photoacoustic flow cytometry (PAFC). Biswas et al. determined differences in cell shape and size with in-vitro methods while Strohm et al. demonstrated the differences in RBC orientation effects on PA signals [47,48]. Fig. 3 demonstrates an example image using PAM of normal red blood cells (nRBCs) and sickled red blood cells (sRBCs). These studies were then validated by theoretical analyses providing mathematical foundations to understanding the PA signal changes, quantifying cell orientation and shape [47].

Fig. 3.

Photoacoustic images of normal RBCs (nRBC, left image) and photoacoustic images of sickled RBCs (sRBC, right image) produced by Cai et al. [49].

Due to the structural changes between sickle cell hemoglobin (HbSS) and natural Hb hemoglobin, Cai et al. demonstrated PA signal differences between RBCs (prior to sickling) containing these two Hb states (Hb and HbSS) [49]. They also showed differences in PA signals between cells that had sickled and those that contained HbSS but had not yet sickled. With this information, clinicians could detect an acute change in the ratio of sickled cells present within the body. Although a rapid change in the percentage of sickled cells may result in an acute pain crisis, further studies are needed for verification. While sickling percentage contributes to vaso-occlusions, measurement RBC movement through vessels may provide another method to quantifying vaso-occlusive crises. Galanzha et al. monitored various rheology parameters within microvascular using PAI. Their team was able to identify aggregations of RBCs and flow rate of RBC movement through microvasculature [50]. Ultimately, these observations may provide objective data thought to contribute to acute pain crises

Importantly, many of these studies also provided in-vivo demonstration for detecting the aforementioned information [48–50]. As previously mentioned, healthcare providers currently rely on subjective data to determine acute pain crises. PAI succeeds in providing an objective data source which can be acquired non-invasively. With the identification of an acute change in the quantity of sickle cells, scale of aggregation, and change in microvascular flow, physicians could implement more optimal pain treatment in sickle cell patients. Further pilot studies are needed to demonstrate the effectiveness of PAI in assessing the relationship of hemorheological changes during sickle cell pain crises.

2.2. Pathogen detection in the blood

More than 670 million communicable diseases occur each year and are the eighth leading cause of death worldwide [51–53]. A large portion of these diseases have treatments that are both effective and relatively low-cost, but they rely upon early and accurate diagnosis. There are two notable areas of communicable diseases in which PAI may provide clinical benefit with early diagnosis and treatment evaluation: malaria and bacteremia.

2.2.1. Malaria

Malaria is a parasite that is reliant on RBCs as a host. The parasite is transmitted via mosquito and is endemic to primarily tropical and subtropical regions of the world. The disease affects up to 230 million patients per year, 400,000 of which result in death [54]. Early diagnosis and complete disappearance of the infection are key to treatment [55]. Current diagnosis involves invasive blood sampling and high-resolution microscopy with an experienced diagnostician, which is not always available in the regions most affected by the disease. There have been many attempts at improving the diagnosis of malaria, but none have been entirely successful in reducing complexity and optimizing diagnosis rate [56].

PAI provides a possible alternative with earlier detection rates and better ease-of-use. Malaria starts as a sporozoite entering the bloodstream via mosquito bite. These sporozoites infect hepatocytes and mature into merozoites, which in turn infect red blood cells. The parasite cycles between infecting red blood cells, reproducing, rupturing red blood cells and thus releasing and infecting other cells [57]. During its merozoite stage, the parasite breaks down hemoglobin producing a byproduct known as hemozoin [58].

Initial studies with PAI demonstrated hemozoin produces a strong PA signal, enabling its use as a surrogate marker for the presence of malaria [56,59]. Traditional optical of thick blood film is the gold standard for detection of malarial parasites but has a detection limit of approximately 50 parasites/microliter and only a sensitivity of around 85% [60,61]. Conversely, more advanced methods have reported a higher sensitivity, but detection limits have not greatly improved [61]. PA flow cytometry (PAFC) is a method by which cells move in a single file through a pulsed laser. Fig. 4 pictorially demonstrates in-vivo PAFC. Each cell that passes by a photoacoustic laser and sensor, which produces a corresponding PA signal. Each of these signals is analyzed to determine if the cell contains hemozoin and thus malaria.

Fig. 4.

Overview of photoacoustic imaging in application of malarial detection. System utilizes photoacoustic flow cytometry as demonstrated in a study performed by Cai et al. [56].

Cai et al. demonstrated the detection of malaria using PAFC occurs at lower rates (one parasite per 0.16 mL) than traditional malaria detection systems [56]. Due to this improved sensitivity, the detection time frame can be reduced from 4 days post-inoculation to 1–2 days post inoculation [56]. In addition to showing a lower threshold for detection, Menyaev et al. demonstrated a similar low-detection limit could be achieved in an in-vivo system, providing a non-invasive alternative to current diagnostic solutions [59].

Although PAI shows promise, its use may be limited due to the use of hemozoin as a surrogate marker. Hemozoin is only produced when the parasite has matured from the sporozoite stage into the merozoite. Patients are infected with malaria in their sporophyte stage and only develop into merozoites once in the liver. Ultimately there is an unavoidable detection lag between onset of infection and presence of hemozoin. Even so, PAI provides a promising methodology for high-quality, non-invasive, and automatable processes for malaria diagnosis and monitoring. Future studies are needed to demonstrate PAI’s effect on earlier detection and treatment outcomes in patients with malaria.

2.2.2. Bacteremia

More than 2.8 million bacterial infections occur annually in the US, costing the healthcare system more than 4.6 billion dollars per year [62]. One of the most deadly and costly complications of an infection is bacteremia. Bacteremia occurs when bacteria leaks into the blood which can lead to prolonged infection of the heart (endocarditis), sepsis, or seeding of distant organs [63]. As a result of its vast impact, bacteremia carries a 30-day mortality rate of 22% [64]. Early and accurate identification of the bacterial species is crucial to reducing antibiotic resistance, length of hospital stay, and morbidity/mortality [65]. Technology for both the identification and monitoring of pathophysiology of bacteria is critical for improving outcomes. The current gold standard for bacteremia identification relies on blood cultures [66]. Although culturing provides useful information for treatment, incubation can take tens of hours to days to develop [67]. This time-delay leads clinicians to choose broad spectrum antibiotics, which contributes to antibiotic resistance [68]. Advancements in technologies such as PCR and serology have enabled earlier detection, but faster methods are still needed.

While there are various methods of bacterial detection, current sensitivities for these technique are around 3.2 × 10⁸ CFU/mL (colony forming units/mL) [69]. Photoacoustic imaging shows promise in providing more efficient diagnosis of bacteria in the blood. Using PA-active labels two studies were able to detect bacteria at a lower rate than current methods. Edgar et al. used dye-marked bacteriophages mixed in a bacterial solution containing Salmonella or E. Coli, which was then run through their ex-vivo PAFC system seen in Fig. 5. A sample was introduced into the system and separated into finite segmentate using mineral oil to separate the segments. Each of the segments was passed by a photoacoustic detection system. Their team was able to detect bacteria at rates as low as 1 × 10⁵ CFU/mL [70]. Conversely, Galanzha et al. utilized golden nanorods (GNRs) to label S. Aureus, which were subsequently injected into rat models. Their in-vivo PAI system demonstrated a sensitivity as low as 0.5 CFU/mL which is substantially lower than current methods [71]. These studies provide proof of concept of in-vivo detection of bacteria, however further studies are needed to determine application to a broader range of bacteria. Ultimately, PAI could be performed at patient presentation, reducing the time for bacterial species identification significantly improving antibiotic selection, and mortality reduction.

Fig. 5.

Example ex-vivo photoacoustic flow cytometry setup used by Viator et al. [70].

Although PA provides promise in reduction of bacterial detection time, the aforementioned studies utilized PA-active labels, which may have some limitations in in-vivo applications. While bacteriophages, like those used in the Edgar et al. study, are more stable and more cost effective than traditional antibody markers, they are bacteria specific [70]. Thus, in clinical applications where the bacteria to be identified is unknown, the bacteriophages cannot bind rendering the dye ineffective. Zlitini et al. attempted to overcome this issue by using the non-specific bacterial metabolite maltotriose. The maltotriose is labeled with a PA-active dye and is selectively and rapidly taken up by bacteria [72]. Their study successfully demonstrated the effectiveness of their maltotriose-dye in the detection of bacterially contaminated catheters.

Similar to the Galanzha et al. study, many have proposed the use of nanoparticles. These small PA-active contrast agents are of particular interest, as they have not only shown efficacy in bacterial identification, but may also be used in bacterial eradication [73]. Kim et al. used silver-covered nanoparticles, which adhered to the bacterial surface. The PA effect then resulted in release of silver from the nanoparticles, exerting their bactericidal effect and ultimately eradicating the covered bacteria [74]. While dyes are easily implemented in ex-vivo methods, hurdles exist for the transition to in-vivo applications. Further studies are needed for evaluation of toxicity and side-effect profiles that may render the technology impractical for clinical use.

Besides bacteria-labeled methods, Thompson et al. proposed a label-free alternative. Their study demonstrated bacterial speciation via analyzing bacterial spectral responses [75]. Further expansion of this technology may provide a non-invasive, label-free application of PAI for bacteremia detection and speciation.

In a pre-clinical setting, PAI has implications in studying microbial dynamics. Cai et al. used a choryll-peptide based agent to monitor macrophage chemotaxis and bacterial uptake [76]. Fig. 6 demonstrates the phototactically labeled peptide, which is engulfed by the macrophage. When the macrophage in turn engulfs bacteria, this produces a detectible PA signal, which can be used to study and understand timing of bacterial uptake with macrophages. Additionally, Galanzha et al. demonstrated bacterial movement within the microvasculature, demonstrating endothelial attachment and extravasation within the body [71]. Such studies provide indications that PAI could be utilized to get a better understanding of bacterial mechanisms which has implications to future treatment regimens.

Fig. 6.

Macrophage undergoing bacterial uptake inducing the production of a PA signal. Produced by Cai et al. [76].

PAI has imaging demonstrated abilities of rapid diagnosis, provide non-invasive detection, and the potential for therapeutic applications to bacterial infections superior to current methods. Future studies are needed to assess PA-active label safety profiles and demonstrate the quantifiable reduction in harm associated with earlier detection using PAI.

2.3. Circulating tumor cells (CTCs)

Cancer is the second leading cause of death in the United States [77]. Of cancer related deaths, 90% are attributed to metastasis [78]. This occurs when original tumor cells disseminate to distant sites within the body. These cells are known as circulating tumor cells (CTCs). Studies have shown that CTCs present in the blood result in higher rates of metastasis and thus poorer patient prognosis [79]. Consequently, detection of CTCs could provide improvement in cancer treatment and outcomes. Furthermore, trending CTC levels provide tumor recurrence monitoring and treatment responsiveness. Ultimately, preventing metastasis by CTC-targeted treatment could reduce the need for systemic treatment altogether and improve patient outcomes.

Previous studies have achieved CTC detection through a variety of methods, but various limitations have prevented their widespread use. Due to lack of reliable cellular markers and low concentration in whole blood (~0.0002–0.00075%), detection of CTCs has proved difficult [80]. Common detection methods include using cell surface markers, filtration and direct optical imaging systems (i.e. flow cytometry, raman scattering). Generally, ex-vivo methods suffer from low sample size which leads to a higher rate of false negatives [81,82]. Methods relying on cell surface markers also suffer from a high rate of false negatives due to CTC’s ability to shield said markers [81,82]. In vivo optical imaging systems such as wide field fluorescence imaging and photodiagnostic infrared spectroscopy showed promise to these limitations, but due to scattering effects, these systems are restricted to superficial depths, limiting its applicability [81,82].

PAI may provide an alternative to detection of CTCs, while improving upon the aforementioned limitations. O’Brien et al. demonstrated PAI’s ability to detect and separate CTCs using PAI with a two-phase microfluidic system [80]. Using an in-vitro system with melanoma CTCs in bovine blood, their system could detect the CTCs 100% of the time at a concentration of 1 CTC/μL. Importantly, they demonstrated PAI’s ability to image multicellular cross sections, unlike traditional flow cytometry systems which rely on cells in a single file for analysis. Multicellular cross section analysis resembles physiologic systems implying in-vivo imaging feasibility. Fig. 7 depicts an in-vivo version of CTC detection using the multicellular cross section technique for detection. The figure shows CTC release into the blood stream due to manipulation of a tumor followed by in-vivo PA detection in the blood stream.

Fig. 7.

In-vivo PAFC schematic for CTC detection, produced by Galanzha and Zharov [89].

Multiple studies have demonstrated photoacoustic imaging of breast, prostate, ovarian and skin cancer CTCs. Certain cell lineages require labeling with PA-active dyes, yet others, such as melanoma cells contain melanin making in vivo imaging readily available. CTC tagging is generally performed through the use of metallic nanoparticles similar to those used for bacterial detection [83–86]. Other types such as magnetic nanoparticles have been used, not only producing a PA signal, but have sequestration abilities which may have treatment implications [87]. Although possibilities are broad, PAI’s use may be limited due to invasive methods required for target labeling and possible side effects of such labels. Still, with the wide range of potential markers, photoacoustic imaging may be a powerful tool in CTC detection.

Despite these limitations, due to its acoustic detection, PAI’s imaging depth has a large advantage over other current CTC imaging methods. Due to the improved depth, PAI enables better in-vivo utilization through visualization of deeper and larger vessels. As such, these larger vessels with higher blood flow rate allow for an increased detection throughout. In typical blood analysis, a small blood sample is extracted from the patient to be tested. Here, PAI can monitor the circulating blood for an extended period, effectively increasing the sample volume [88]. Galanzha et al. demonstrated this power by monitoring the entire blood volume of a mouse under 1 min [89]. Others observed CTCs using PAI of blood vessels at 3–4 mm depth range in human studies as proof of concept [90]. Furthermore, PAI may be applied in real-time CTC detection for procedural purposes. Jurantli et al. studied at the effect of pressure secondary to resection or incision from a tumor [88,91]. These studies demonstrate PAI ability to image deeper, larger vessels which ultimately produces a more sensitive detection of CTCs.

Finally, while PAI has great diagnostic implications, its prospect for treatment implications sets it far beyond current methods. He et al. used a 8.8 mJ/cm2 laser fluence that allowed instantaneous rupture of melanoma CTCs once they were detected, but did not damage healthy cells [91]. Galanzha et al. used CTC-targeting nanoparticles that resulted in heating-induced nanobubbles and rupturing of the CTCs [89]. In addition to mechanical destruction, the lysing of the CTCs may reveal malignant markers such as cancer antigens, thereby increasing immune detection and further destruction of the cancerous cells [91]. While limitations exist and further research needs to be performed, PAI may have an impact on improving cancer outcomes through the early detection and possibly treatment of CTCs.

3. Applications to molecular components in blood analysis

The previous section described analysis of cellular components within the blood using PAI. Here, we highlight a few examples of PAI’s application for analysis of the acellular or plasma component of blood. Molecules within plasma are known as analytes, ranging in size from large proteins (e.g albumin, 15 nm) to electrolytes (eg. potassium, 0.266 nm) [92,93]. Quantification of these molecules provides an abundance of information about the physiological and pathological statuses within the body. Blood analysis has been and continues to be the cornerstone of modern medicine. Currently, blood analysis requires a wide range of technologies, making testing for these molecules complex, time consuming, and expensive. PAI’s application in blood analyte testing may improve the efficiency and lower the cost.

3.1. Non-invasive blood glucose monitoring

More than 34 million people in the US live with diabetes [94]. Complications from elevated blood glucose levels cost an estimated $320 million dollars to the US healthcare system annually and are rising [95]. Reliable and fast measurement of blood glucose levels is key to preventing devastating outcomes. Detection methods have progressed significantly over the years, but an accurate and non-invasive method has yet to be established.

Current detection methods often utilize disposable needles and microfluidics. While small, these invasive methods deter users from adherence to glucose monitoring [96]. A semi-implantable alternative that provides annual rather than daily needle exchange, but complexity of use and cost continue to be a barrier to use [97].

Early methods of non-invasive detection included urine glucose testing; however, reading levels were inaccurate due to the patient’s large variation in hydration level. More advanced methods such as impedance spectroscopy and sonophoresis were developed, but skin irritation, among other issues, prevented their full implementation [98]. An accurate and non-invasive glucose monitoring system is yet to be created.

In 2015, Pai et al. showed that photoacoustic spectroscopy (PAS) could be utilized to determine glucose levels of varying degrees [99]. Fig. 8 shows the Clark error grid (CEG) plot of glucose levels relative to the predicted concentration using photoacoustic spectroscopy (PAS) by Pai et al. Their results demonstrated PAS as a safe and effective method to measure glucose levels. Using the CEG criteria (areas A and B are clinically acceptable variation) 100% of their measurements fell within an acceptable range. Furthermore, Taknaka et al. reported a similar response and variation using differential continuous-wave photoacoustic spectroscopy and its correlation with glucose over 3 h [100].

Fig. 8.

Photoacoustic spectroscopy evaluation of glucose levels in solution produced in a Clark error grid. Produced by Pai et al. [99].

However, not surprisingly, the more physiologically relevant tests proved difficult. Studies demonstrated a significant overlap in signal generation between glucose and other physiological substances such as water resulting in larger error in glucose concentration results. Fig. 9 demonstrates the results from Sigrist et al. who tested the in-vivo correlation of PAS for glucose levels. Their results demonstrated a larger scattering of data resulting in an R² value of 0.8 secondary to background signal. As such, further studies are needed to investigate these variances and solutions to improve the PA-based glucose monitoring system.

Fig. 9.

Photoacoustic spectroscopy evaluation of glucose levels within the blood vs blood level extracted values. Produced by Kotmann et al. [101].

One technical obstacle in PA glucose measurement is the lack of ability to take repeatable and accurate measurements, due to the variances in skin content. Sim et al. reported concerns with signal variation due to eccrine glands. These glands produce PA-active substances with a similar spectrum as glucose [98]. Their study showed that because of these glands, there was a significant variation in optical absorption across the skin, affecting the glucose measurements. Yet, their team was able to identify skin areas with high and low optical absorption, and thus use areas between the eccrine glands to produce a more accurate measurement of blood glucose content. Furthermore, these studies review physiological glucose levels. In instances of pathological conditions such a diabetes where regular exogenous insulin is in use, these glucose variations may affect PAS concentrations. While PAI needs further tuning to optimize and overcome current and future obstacles, implementation of this technology may provide a revolutionary method to non-invasive glucose testing.

3.2. Alternative analytes

While there are a vast number of analytes that can be potentially measured by PAI, this paper discusses only a few pilot studies in PAI analyte testing. To demonstrate the range of PA’s applicability, we’ve selected analyte testing in the following categories: blood proteins, pharmaceuticals, and electrolytes. As previously mentioned, current analyte detection and measurement generally requires an individual test for each analyte of interest. Therefore, using a photoacoustic system for the testing of all analytes may provide a standardized method allowing for simplified and cost-effective blood analysis.

A few studies have shown efficacy in detection of blood protein concentrations. For example, carbon monoxide, an odorless, colorless gas causes a toxic conformational change to hemoglobin, forming the blood protein carboxyhemoglobin. Due to its rarity and subtle clinical characteristics, diagnosis can be difficult [102]. Chen and Peters et al. demonstrated PAI’s ability to detect carboxyhemoglobin’s blood level and characterize its kinetics [103,104]. Similarly, methemoglobin (metHb) is a toxic blood protein formed from the use of medications in the nitrite class. Like carboxyhemoglobin, testing for metHb can be arduous. Furthermore, diagnosis and treatment are time-sensitive. Multiple studies have shown PAI to be a viable option for expedited and in-vivo testing for metHb [105,106]. A third example of protein detection with PAI is bilirubin. Hyperbilirubinemia is a common finding in newborns. In most, it resolves naturally, but in certain instances it requires medical intervention. Thus, routine care involves regular bilirubin level checks. Due to the sensitivity of the pediatric population, noninvasive transcutaneous analysis is the preferred detection method. Current methods utilize optical imaging systems, which have proven effective for screening tools, but have depth limitations as mentioned previously [107]. Zhou et al. illustrated PAI’s effectiveness in the detection of bilirubin in a bovine blood phantom solution [108]. Fig. 10 demonstrates their results by showing a significant correlation between concentration of bilirubin and calculated concentration based on photoacoustic signal intensity. These studies show PAI’s potential to observe and quantify proteins within the blood which may have implications for biomarker detection and monitoring.

Fig. 10.

Photoacoustic imaging and quantification of bilirubin samples of varying concentration in phantoms. Produced by Zhou et al. [108].

Smaller molecules, such as pharmaceuticals, could also be monitored by PAI. Jeevarathinam et al. detected the drug heparin using a PAI system [109–111]. Similarly, Yim et al. utilized polydopamine nanocapsules (PNCs) with molecular dyes to enhance PA-mediated heparin detection [112]. Their studies also provided drug effectiveness data, by comparing rheology parameters extracted from the PA data with conventional activated coagulation time (ACT) measurements. They proposed PAI may play a role in areas such as invasive-intraoperative continuous clotting measures for cardiac surgeries. Fig. 11 shows measured and simulated photoacoustic intensities of heparin and their resulting ACTs as a function of heparin concentration.

Fig. 11.

Photoacoustic analysis of heparin using an in vitro (human blood) PAI system. The above graphs demonstrate a statistically significant correlation between PA measurements and ACT clotting measurements and heparin units. Produced by Zhou et al. [113].

In another study, Furdella et al. used PAI to detect and measure the drug Dil (1,1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanine perchlorate), a surrogate for rapamycin in DES (drug eluting stents) [114]. Their team monitored porcine coronary arteries after drug injection of the drug and found a concentration-dependent PA signal. These data may be used to confirm that the DES is providing a drug concentration within the desired therapeutic range. Lastly, Taruttis et al. performed pharmacokinetic studies using PAI with indocyanine green (ICG) dye [115]. They performed in-vivo PAI of the ischiatic vein of a mouse and characterized the circulation kinetics of ICG. These studies show PAI’s ability to identify and quantify pharmaceuticals, which may ultimately be used in pharmacokinetic studies.

Finally, there have been a few studies demonstrating ion concentration detection. Cash et al. used optode nanosensors in a photoacoustic system for detection of lithium [116]. Using complex formulated optode nanosensors, they showed a concentration-dependent photoacoustic signal using a chicken skin phantom. Lee et al. used a similar method with an ion-selective optode to perform an ex-vivo analysis of potassium concentration in a phantom [117]. While neither of these studies demonstrated in-vivo effectiveness, their studies provide a proof-of-concept for ion detection using PAI, lending to potential electrolyte concentration analysis in the future.

These studies demonstrated the wide range of existing applications for PAI in blood analysis and showed the potential for many new applications. Limitations from background signals, penetration depth, signal consistency and measurement accuracy require further work, but do not prevent the fast development and clinical implementation of this exciting technology in blood analysis.

4. Conclusion

As a ubiquitous fluid within the body, blood has broad implications in detection, monitoring and treatment of many pathophysiological processes. Due to PAI’s superior depth of penetration and non-invasive nature, it has a wide range of applications in blood analysis. Its versatility in detecting many targets with a single system provides advantage over the current olio of systems required to test for individual targets. Furthermore, PAI’s non-invasive nature lends to more accurate examination of pathophysiologic status within an undisturbed biological subject. The combination of these features makes PAI prime for enhanced utilization in preclinical settings and translation into more clinical settings.

Although improved over pure optical imaging systems, limitations still exist for PAI’s application to blood analysis. Background noises from intrinsic chromophores such as eccrine gland production and hemoglobin provide difficulty in identifying target signals in in-vivo settings, yet varying studies are under way to develop solutions [118]. Additionally, pure label-free detection of many blood analytes may prove difficult. The use of PA-active labels will require rigorous toxicologic testing, but ultimately may provide advantages in more sophisticated imaging of targets in in-vivo settings. Besides, some studies continue to investigate photoacoustic spectroscopy for label-free analysis for a broader range of analytes [119]. While challenges exist, studies have enlightened possible solutions that continue to show a positive future for PAI’s application to blood analysis.

While challenges exist, there are a significant number of studies demonstrating the impressive impact PAI could have on blood analysis. Its label free signal from hemoglobin lends itself well to further implementation in in-vivo monitoring of sickle cell disease. Its in-vivo monitoring of blood flow has implications for point of care detection in infectious diseases and its tunable wavelength and wide range of PA-active labels may enable for detection of a vast number of analytes.

Overall, photoacoustic imaging provides great promise in the application to blood analysis. As blood analysis is a cornerstone for modern medicine this technology could have significant impacts in the way blood is studied in pre-clinical labs and how blood is analyzed and monitored in the clinical setting. We expect a progression of PAI integration in blood analysis using photoacoustic imaging in future years as the technology evolves.