Open Access
Add to BibSonomy
Abstract
Photoacoustic imaging (PAI) is a novel hybrid imaging technique that combines the advantages of optical and ultrasound imaging to produce hyperspectral images of the tissue. The feasibility of measuring oxygen saturation (sO2) with PAI has been demonstrated pre-clinically, but has limited use in humans under conditions of ischemia and reperfusion. As an important step towards making PAI clinically available, we present a study in which PAI was used to estimate the spatial distribution of sO2 in vivo during and after occlusion of the finger of eight healthy volunteers. The results were compared with a commercial oxygen saturation monitor based on diffuse reflectance spectroscopy. We here describe the capability of PAI to provide spatially resolved picture of the evolution of sO2 during ischemia following vascular occlusion of a finger, demonstrating the clinical viability of PAI as a non-invasive diagnostic tool for diseases indicated by impaired microvascularization.
© 2021 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
The measurement of oxygen saturation (sO2) is important in a wide range of medical fields, and is essential in the monitoring of the progression of cardiovascular and peripheral vascular disease, such as diabetes [1]. Oxygen saturation can also be used to predict the survival of flaps during plastic and reconstructive surgery [2]. Furthermore, tumor progression and malignancy are strongly dependent on tumor hypoxia [3], and cerebral sO2 is monitored in the diagnosis of cerebral desaturation in stroke patients [4].
Several techniques have been developed to monitor sO2, all of which have limitations. For instance, functional magnetic resonance imaging is used to monitor blood oxygen-level-dependent contrast, but is sensitive only to deoxygenated hemoglobin (HbR) [5]. Positron emission tomography requires the use of ionizing radioisotopes, which may cause damage to biological tissue [6]. The most well-known technique for measuring sO2 is pulse oximetry, which is based on the physiological activity of the cardiac pulse, in combination with the difference in spectroscopic reflectance at wavelengths of 660 nm and 940 nm to measure the concentrations of oxygenated hemoglobin (HbO2) and HbR [7]. Simultaneous plethysmography only allows the measurement of arterial blood saturation. Near-infrared spectroscopy and diffuse optical tomography make use of light in the wavelength range of 690–850 nm, and can be used to monitor HbO2 and HbR by analyzing distinct absorption peaks, for example, at 850 nm and 760 nm [8]. These techniques measure the combined oxygenation of venous and arterial blood. However, these spectroscopic techniques lack the spatial resolution required to measure the heterogeneous distribution of tissue oxygenation, and are often limited to a superficial depth of a millimeter or less [9].
Photoacoustic imaging (PAI) is a rapidly developing biomedical technique that has the capability of providing spatially resolved sO2 images of the tissue. It is a hybrid imaging modality, combining the high-contrast and spectroscopic-based specificity of optical imaging and imaging depth of ultrasound. Pulsed laser light is absorbed in the tissue, resulting in a thermoelastic response which generates acoustic waves detected by a high-frequency ultrasound transducer [10]. PAI is therefore capable of detecting the molecular composition of the tissue, at a depth almost an order of magnitude greater than conventional spectroscopic methods. PAI thus has the potential to map sO2 at specific locations in the tissue, non-invasively.
To date, PAI has mainly been developed for the measurement of sO2 in phantoms [11,12], and numerous pre-clinical studies on animals have exploited HbO2 and HbR to characterize tumor microenvironments, [8,13,14] as well as renal function [15], and myocardial ischema [16]. The feasibility of PAI has recently been demonstrated in the first clinical applications, showing promising potential in the diagnosis of breast cancer [17,18] and inflammatory diseases such as systemic sclerosis [19], Crohn’s disease [20], Raynaud’s phenomenon [21], as well as in healthy vasculature [22,23]. In these studies, HbO2 and HbR have proven to be valuable biomarkers for the differentiation of healthy and pathological tissue. The feasibility of using PAI to monitor spatially resolved ischemia-perfusion in a mouse model is well-established [14,24–26], whereas the number of studies performed on humans is still limited to determine clinical viability [21,27–29]. The aim of the present study is thus to monitor oxygenation using PAI and spectral unmixing during ischemia following vascular occlusion of a finger in eight healthy volunteers. The results are compared to those obtained with diffuse reflectance spectroscopy (DRS), performed using a commercially available instrument.
2. Methods
2.1 Ethics
The experimental protocol for this study was approved by the Ethics Committee of Lund University, Sweden. The research adhered to the tenets of the Declaration of Helsinki as amended in 2008. All the subjects were thoroughly informed about the study, and the voluntary nature of participation. All subjects gave their fully informed written consent.
2.2 Subjects
Eight healthy adult volunteers, four men and four women, were included in the study. The exclusion criterion was the presence of any advanced medical condition that could contraindicate strangulation of a finger, or microangiopathy, such as diabetes or smoking. Subjects were asked to refrain from caffeine-containing drinks and food for at least two hours, as well as strenuous exercise 24 hours, prior to the measurements. The subjects rested lying down for 10 minutes, after which their heart and lungs were auscultated. Blood pressure and heart rate were measured before and after the procedure. The median age of the subjects was 39 years (range 31–56), all were healthy and non-smokers. All participants had type II-III skin color on the Fitzpatrick scale [30]. Studies were performed in a temperature-controlled room, in which the temperature was maintained between 22 and 23°C.
2.3 Photoacoustic imaging
PAI was performed using a Vevo LAZR-X multimodal imaging system (VisualSonics Inc., Toronto, ON, Canada). Details of the PAI setup for the examination of human subjects have been presented previously by Sheikh et al [31]. The PAI probe was mounted on an adjustable arm (Mounting Accessory, GCX Corporation, CA, USA) and placed in contact with the skin (Fig. 1(a)). Care was taken to ensure that the probes were held in place without causing pressure on the finger, in order to avoid pressure-induced hypoperfusion. The tissue was illuminated with pulsed laser light (repetition rate of 20 Hz and pulse duration of 7 ns) in the 680–970 nm wavelength range (in 5 nm steps) to generate hyperspectral PA images. The thermoelastic response caused by absorption of the excitation light was detected with a linear ultrasound probe (MX550D, VisualSonics Inc., Toronto, ON, Canada) with a central frequency of 30 MHz and bandwidth of 22–40 MHz, providing axial and lateral resolutions of 50 µm and 110 µm, respectively. Photoacoustic images were obtained to a depth of 6 mm. The signal-to-noise ratio below this depth was too low for reliable analysis. A complete hyperspectral photoacoustic image was generated every 15 seconds.
Fig. 1. Photographs showing monitoring with (a) the photoacoustic imaging device and (b) the commercial oxygen monitor using diffuse reflectance spectroscopy. Blood flow was occluded by a pressure cuff around the base of the finger, shown with its sphygmomanometer in (b).
2.4 Diffuse reflectance spectroscopy
A commercial oxygen monitor (moorVMS-oxy, Moor Instruments, Devon, UK) employing DRS was also used and the oxygen saturation obtained was compared to that from PAI (Fig. 1(b)). DRS employs an incoherent white light source providing illumination in the entire detection range between 500 and 650 nm, which is guided to the skin via a fiber. The diffuse reflectance is collected by a detector fiber that is laterally separated from the source fiber providing the sample illumination. Irrespective of what type of instrument is used, light penetration depth into tissue is wavelength dependent. Employing DRS with optical fibers, the source–detector fiber separation in addition governs the effective measurement depth; a larger separation resulting in a deeper measurement volume into the medium [32]. The detection depth of this commercial oxygen monitor is approximately 1 mm. DRS methods lack the spatial sensitivity of PAI, however, the sampling rate is high providing a measurement of sO2 every 100 milliseconds. The DRS probe was attached to the finger, using a double-sided adhesive O-ring, causing no pressure on the finger [33].
2.5 Experimental procedure
Finger occlusion is an established technique for inducing hypoxia [34]. The main advantage of finger occlusion over arm occlusion is that blood perfusion ceases momentarily, leading to the rapid onset of hypoxia. Furthermore, the digital arteries are easily accessible for imaging technology as they are located just under the skin. In the present study, occlusion was achieved using a finger cuff and applying a pressure >200 mmHg using a sphygmomanometer (DS-6501-300, Welch Allyn, IL, US). The arm was placed in anatomical position and the ventral surface of the index finger or middle finger was made available for measurements. The subject’s arm was stabilized using a vacuum pillow (GermaProtec, Germa AB, Kristianstad, Sweden), and the subjects were specifically asked not to move during the procedure. The lighting in the room was dimmed during the measurements to minimize background light. Measurements were performed with PAI and DRS before, during, and after occlusion. Measurements were made on one finger using PAI, and DRS measurements were made on the adjacent finger on the same hand of the individual. The order of measurements was randomized so as not to create a bias toward one of the measurements.
2.6 Data analysis
A linear spectral unmixing model was used for spectral analysis, according to Eq. (1):
where M is a vector representing the measured photoacoustic spectrum, a is the linear coefficient (also called the fractional abundance) for each endmember spectrum, and s is a matrix containing the endmember spectra representing the absorption spectra of the tissue constituents assumed to contribute to the measured photoacoustic spectrum, M. The only constraint on the linear model is that the coefficients must be positive, which means that negative absorption is not realistic. A non-negative least-squares approach was used with MATLAB (The MathWorks Inc.) to minimize the fitting discrepancy by varying the fractional abundance, ai, of each endmember spectrum in s. w is a vector accounting for spectral noise [35].In determining the endmember spectra, it was assumed that the main absorptive components in human tissue were HbO2 and HbR, melanin, fat, and water. The absorption spectra for each of these endmembers were taken from a previous report by Jacques et. al [36]. For each spectrum, the fractional abundance of HbO2 and HbR was used to calculate sO2 according to Eq. (2),
Spectral unmixing of the multispectral photoacoustic image was performed on a single spectrum extracted either on a pixel-by-pixel basis, or after averaging many photoacoustic spectra over a specific spatial region. In the former case, the calculated value of sO2 for each pixel was mapped onto the original ultrasound image with the same spatial coordinates as the photoacoustic image, in order to yield a spatially resolved image of the sO2 distribution. One drawback of spatially mapping sO2 on a pixel-by-pixel basis is that it takes 15 s to generate a full photoacoustic spectrum, and involuntary motion of the subject may result in artefacts in the extracted photoacoustic spectra. When sO2 is represented as a function of depth and time, the extracted values of sO2 are averaged across the entire width of the image and over a depth corresponding to 3 pixels, as this removes most lateral motion artifacts and minimizes axial motion artifacts.3. Results
3.1 Absorption spectra of the three skin layers
Photoacoustic imaging enabled non-invasive parallel acquisition of ultrasound and photoacoustic images, allowing anatomical co-localization of spectral and structural information. Figure 2(a) shows a cross-sectional high-frequency ultrasound image in which the three main skin layers have been identified (the epidermis, dermis, and hypodermis). The contrast between the epidermis and dermis is most visible, and the thickness of the epidermis determined from the ultrasound image is approximately 0.3 mm, which is comparable to 0.4 mm as previously reported in the fingertip [37,38]. The dermis and hypodermis are not as easily differentiated, as these skin layers are compositionally more similar. However, the thickness of the dermis in the fingertip has been estimated using laser Doppler flowmetry to be approximately 1.5 mm [38].
Fig. 2. (a) Cross-sectional high-frequency ultrasound image showing the epidermis, dermis and hypodermis separated by white dashed lines. (b) Photoacoustic spectra (solid lines) from the three layers to which the linear spectral unmixing model was applied assuming endmember absorption spectra from HbO2 (red), HbR (blue), melanin (yellow), and fat (green). The fit including all endmembers is shown as dashed lines. It can be seen the signal from melanin dominates in the epidermis, reflecting the presence of melanocytes, while in the dermis and hypodermis HbO2 and HbR dominate, reflecting the vascular plexus. The signal from the hypodermis is dominated by blood and subcutaneous fat, which is consistent with the anatomical structure of this skin layer.
Figure 2(b) shows the photoacoustic spectra obtained from the three superficial layers of the skin on a finger (epidermis, dermis, and hypodermis). Spectral unmixing was used to determine the relative composition using five endmember spectra: HbO2, HbR, melanin, fat and water. All five endmembers were used to unmix the photoacoustic spectra from the three layers. The photoacoustic spectrum for the epidermis is reproduced mainly by the endmember spectrum for melanin, with small contributions from HbO2 and HbR, reflecting the presence of melanin-containing melanocytes. In the dermis, the contribution from melanin disappears, and the photoacoustic spectrum consists of a combination of the endmember spectra for HbO2 and HbR, reflecting the presence of a superficial vascular plexus. The composition of the spectrum for the hypodermis is similar to that of the dermis, with the addition of the endmember spectrum for fat, reflecting the presence of subcutaneous fat. The contribution from water was marginal in all cases and was therefore omitted, although we stress the importance of including it in the analysis.
3.2 Spatially resolved sO2 with PAI
PAI was used to measure the spatially resolved evolution of sO2 during finger occlusion. Figure 3(a) shows the fractional abundance of HbO2, HbR, and melanin, superimposed on an ultrasound image. As expected, melanin dominates the upper layer, i.e., the epidermis, while HbO2 and HbR are found in the dermis. Figure 3(b) shows the evolution of sO2 over time in the epidermis, dermis, and hypodermis, in the form of a graph and a heat map. It can be seen that the absorption spectrum in the epidermis is not affected by finger occlusion, due to the absence of vascular supply. The greatest decrease in sO2 is seen in the dermis where the superficial vascular plexus is located.
Fig. 3. (a) Cross-section of a high frequency ultrasound image in which PA spectral information has been superimposed, showing the fractional abundance of HbO2 (red), HbR (blue), and melanin (yellow), in the skin before finger occlusion. Since the fractional abundance of all endmembers cannot be represented in each pixel in one image, only the endmember spectrum contributing most to each spectrum in every pixel is shown. The epidermis, dermis, and hypodermis are separated by white dashed lines. (b) Graph and 2D heat map that showing the evolution of sO2 over time, at different depths, during finger occlusion. Each horizontal line in the heat map represents the spatial average in the analyzed photoacoustic image, as described in the methods section. Note that the greatest decrease in sO2 is seen in the dermis, where the superficial vascular plexus is located, while a slightly smaller decrease is seen in the hypodermis. Vertical black dashed lines indicate the beginning and end of occlusion. (c) and (d) show the corresponding results for a cross-section containing a digital artery (indicated by the two drawn white lines) at the intersection between the dermis and hypodermis. The sO2 levels remain high in the digital artery throughout the measurement period. Moreover, the sO2 at depths below the artery are similar regardless of whether the artery is present or not, suggesting that the signal is not affected by spectral coloring.
Figure 3(c) and (d) show the corresponding results from PAI of a palmar finger artery. The presence of a blood vessel was confirmed in the sequence of ultrasound images showing blood flow. The fractional abundance of both HbO2 and HbR was high in the digital artery, compared to the surrounding tissue. sO2 in the artery decreases only slightly during occlusion compared to the surrounding tissue, most probably because arterial occlusion causes the blood in the vessels to stagnate in the artery, where there is minimal oxygen consumption. Detection of local variability in the sO2 response confirms the feasibility of using PAI to measure sO2 in human tissue with spatial resolution.
3.3 Comparison of sO2 using PAI and DRS
Figure 4 shows the results of PAI measurements of sO2 in the dermis and hypodermis, together with the results from the commercial DRS-based oxygen monitor. Finger occlusion resulted in a significant decrease in sO2 (median values) in the dermis (from 90% to 37%), and a slightly lesser decrease in the hypodermis (from 93% to 48%). The value of sO2 reached a constant level after 10 min, when monitored with PAI. When monitored with DRS, sO2 leveled off already after 2 min,, and a substantially lower level (10%), than when monitored with PAI. This may be because DRS only measures perfusion in the outer millimeter of the skin and therefore only targets the superficial vascular plexus.
Fig. 4. The decrease in sO2 during finger occlusion in eight subjects, expressed as median values (solid lines) and 95% confidence intervals (shaded areas). The vertical dashed lines indicate the beginning and end of occlusion. PAI enables spatially resolved measurements in the dermis (left) and hypodermis (middle). Note the greater decrease in sO2 when using DRS (right) compared to PAI.
4. Discussion
4.1 Benefits of multi-wavelength PAI
The results of the present study show the ability of PAI to provide spatially resolved information on the evolution of sO2 during ischemia following vascular occlusion. To the best of the authors’ knowledge, this is the first study of its kind, in which PAI has been used together with spectral unmixing over a broad spectral range (680–970 nm) during the occlusion of a human finger, providing a map of the absorption of the chromophores in tissue. PAI has been used in a few previous studies on oxygenation in humans. In these studies only a few wavelengths have been used to calculate sO2 which limits taking the absorption by other light absorbing constituents in the tissue into account. Eisenbrey et al. studied the effect of cold stimulus on oxygenation of the fingertips of patients with Raynaud’s phenomenon, using the built-in oxy-hemoglobin package of the PAI system (comparing the PAI signal at 750 nm and 850 nm) to calculate the percent hemoglobin oxygenation [21]. They found PAI to be able to identify patients with Raynaud’s phenomenon and also evaluate the response to a cold stimulus [21]. Yang et al. studied total hemoglobin (HbT) in the human forearm muscle during occlusion of the arm, using the isosbestic point of Hb (800 nm) and two other single wavelengths to obtain relative HbO2 and HbR concentrations, and found that the results were comparable to those obtained using near-infrared spectroscopy [27]. Karlas et al. also used PAI to study muscle oxygenation during occlusion of the arm, using three single wavelengths: 850 nm (HbO2), 800 nm (HbT), and 750 nm (HbR) [28]. Although arterial occlusion was expected to lead to a decrease in intramuscular oxygenation, the results showed an overall steady level of HbO2.
In the above-mentioned studies, the HbO2 and HbR content was extracted using only a few wavelengths without considering absorption by other tissue chromophores, which introduces some room for error. In the previous studies, HbO2 and HbR spectra were identified at depths where vascularization is expected, however, they were also found in parts of the body where vascularization is not expected (i.e. the fingernail and epidermis). In an ideal scenario, each absorbing chromophore has a unique and narrow spectral feature which thereby can be measured and evaluated independently from the others with one excitation wavelength per chromophore. However, since this is unrealistic in human tissue, it is necessary to use several excitation wavelengths in the PAI measurements and to employ spectral unmixing, including endmember spectra for as many chromophores as could potentially be present. The present study employs spectral unmixing accounting also for the absorption by chromophores that are not directly involved in the assessment of sO2 (i.e. melanin and fat). This allowed for a more reliable interpretation of the results in relation to the anatomic composition, establishing that the information gained from spectral unmixing with only a few excitation wavelengths is insufficient to identify the full range of anatomically expected chromophores.
4.2 Oxygen consumption in the skin
The PAI measurements of the present study showed that vascular occlusion led to a significant decrease in sO2 in the dermis, but a smaller decrease in the hypodermis. The reason for this depth-dependent oxygen consumption cannot be deduced from the present study, but may depend on metabolism as cells vary widely in their rate of oxygen consumption [39]. Oxygen consumption has been shown to be highest in the epidermis and in the lower layers of the dermis, where an increasing number of sweat glands lead to an increase in oxygen consumption [40]. Dermis contains many cells such as fibroblasts, mast cells, vessels and nerve endings, hair follicles and sweat glands [41], that consume oxygen at a high rate. Another explanation of the depth-dependent variation of sO2 decrease upon vascular occlusion may be differences in vascular anatomy between the various layers of the skin. The cutaneous microcirculation is organized in two vertical plexuses: one deeper at the junction between the dermis and the underlying subdermal fat, and one superficial in the outer dermis [42]. These two vascular plexus may react differently to occlusion using a finger cuff since the pressure is applied from the outside. It is well known that fluctuations in skin perfusion can change due to contraction and relaxation in smooth muscle cells and pericytes in the vessel walls reacting to nerve signals and local chemical mediators [43]. Arterioles of the deep dermis are larger than in the superficial plexus with a higher number of pericytes [41], which may also affect their response to occlusion. Our results underline the importance of obtaining spatially resolved information on the effect of medical interventions on oxygen saturation.
4.3 Comparison of techniques
Today, pulse oximetry is the most commonly used method for clinical monitoring of sO2. This technique uses the cardiac rhythm to monitor sO2 at each heartbeat using known absorption properties at two wavelength [7]. Normal recommended ranges for sO2 in adults are 94 to 98 percent [44]. There was a difference between the PAI measurements at baseline and those commonly obtained during clinical pulse oximetry measurements; the values obtained with PAI being clearly lower. The reason for this discrepancy cannot be deduced from the presents study, but it may be due to the fact that PAI measures oxygenation in the skin rather than in the arterial blood. The metabolism in the peripheral tissue is higher, and therefore the oxygenation of microvascular hemoglobin may be lower.
The decrease in sO2 appeared to be greater and more rapid when monitored with DRS than with PAI. The measurement depth of DRS is determined by the distance between the fibers guiding the light to and from the tissue, and the distance between the fibers in the moorVMS-oxy monitor enables the measurement of backscattered light from only the outer millimeter of the skin. The epidermis on human fingertip is approximately 0.40 mm thick [37], and the superficial vascular plexus is known to be located just under the basement membrane of the epidermis [42], i.e. approximately 1 mm below the skin surface. This is within the measurement range of DRS, explaining the strong reduction of sO2 measured by DRS. PAI has the advantage of spatially resolving sO2 to a depth of at least 6 millimeters in tissue, and the results confirmed that the greatest reduction in sO2 occurred at the location of the superficial vascular plexus, corroborating the results obtained by DRS.
4.4 Spectral unmixing for determination of sO2
Spectral unmixing was used to extract information on the chromophore content in the finger during occlusion. A previous report [45] has shown that applying linear spectral unmixing to PAI data to extract sO2 in tissue, only accounting for HbO2 and HbR as endmembers, can lead to large errors. Based on the expected anatomical composition of the measured tissue, we included endmember spectra representing absorption of HbO2, HbR, melanin, fat, and water when unmixing the PAI spectra. However, a difficulty remains in the determination of the prevalence of endmembers representing the chromophores in the measured volume. In the epidermis, where melanocytes are found, the photoacoustic spectrum should be reproduced mainly by the endmember spectrum of melanin. Interestingly, spectral unmixing showed some contributions from both HbO2 and HbR in the epidermis, which was not expected due to a lack of blood vessels. We attribute this to the fact that the rather featureless spectral shape of melanin can, to some extent, be reconstructed from a combination of the HbO2 and HbR spectra. This was indicated by the deviation between the total fit for all endmembers and the total photoacoustic spectrum around 760 nm, which is a characteristic absorption peak of HbR (Fig. 2(b), top panel). Furthermore, no change in the relative concentration between HbO2 and HbR was observed in the epidermis upon finger occlusion, supporting the hypothesis that the spectrum in the epidermis primarily represents melanin, and not HbO2 and HbR (Fig. 3) [46].
Accounting for endmembers that do not necessarily contribute to the PA absorption signal in the spectral unmixing analysis does not reduce the goodness-of-fit to the data. We propose that it is better to include too many endmembers rather than too few. An iterative approach to optimizing the number of endmembers has been proposed by Rogge et al., [47] which could certainly improve the overall results. In cases where a high hemoglobin content is expected, the improvement will probably be minimal as long as HbO2 and HbR are accounted for. However, when generating an sO2 map of a tissue segment several mm deep, where a pixel-by-pixel analysis is performed, there may be regions where the hemoglobin content is low and absorption is dominated by other tissue chromophores. Therefore, unless the relevant endmembers are assessed in the analysis of each PA spectrum on a pixel-by-pixel basis, it is better to include as many endmembers as possible.
4.5 Effects of spectral coloring
Spectral coloring occurs as incident photons successively penetrate deeper into tissue and experience a wavelength dependent absorption in different tissue layers. This not only results in an attenuation of the entire fluence spectrum with increasing depth, but also that the attenuation is wavelength dependent [48]. Thus, the spectra and values of sO2 obtained from deeper tissue layers should be interpreted with caution. Spectral coloring may be a contributing factor to why sO2 reduction becomes less prominent with depth, although we have some indications that spectral coloring may not be influencing our results. An artery was visualized at the junction between the dermis and the hypodermis (Fig. 3) and the high sO2 level in the vessel lumen contrasted with the lower sO2 levels in the surrounding tissue, suggests that spectral coloring did not obscure this anatomic feature. Since hemoglobin is a strong light absorber, one would expect that the passage of light through an artery should lead to significant changes in the fluence spectrum received by the tissue layers below. However, the lack of change in sO2 at similar depths in a nearby tissue segment of the same subject without an artery suggests that spectral coloring had little effect in this case.
Light scattering could explain why spectral coloring seemed to have little effect on our results. PAI measures absorption and is directly insensitive to scattering, although scattering may contribute indirectly to PAI absorption by saturating a large volume of the tissue with photons. If the volume of the blood vessel is much smaller than the volume populated with photons from the excitation source, the digital artery would have little effect on the PAI signal in tissue layers below it as photons could reach regions below the artery without having to pass through it. This could certainly be tested using phantoms and controlling their relative size compared to the generated excitation volume, which is a topic for a future study.
Although more complex analysis methods could be employed, as demonstrated by Tzoumas et al. [45], we have found that a relatively simple approach of applying a basic linear unmixing model to multispectral PA images can provide measures of sO2 that could be useful in the clinical setting. We emphasize, however, that the inclusion of endmember spectra in this analysis should be as extensive as possible for the particular tissue that is measured.
5 Conclusions
In conclusion, the present study demonstrates the ability of PAI to provide spatially resolved information on the evolution of sO2 during ischemia following vascular occlusion. Multispectral PA images were obtained over a broad spectral range and analyzed with a linear spectral unmixing model taking into account absorption by melanin, fat and water in addition to HbO2 and HbR. This provided a physiological map depicting the oxygenation as a function of time after occlusion in human tissue down to a depth of 6 millimeters. These results provide the first visual evidence that vascular occlusion affects oxygenation differently in different layers of the skin. The unique ability of PAI to provide spatially resolved oxygenation, while still being non-invasive, promises unmatched clinical opportunities for monitoring local ischemia. The future development of PAI into a clinical diagnostic tool promises unmatched clinical opportunities for monitoring local ischemia and for discriminating between subtle changes in chromophore absorption by physiological and pathological tissue. Medical conditions that benefit from sO2 measurements with spatial resolution include diabetic wounds, peripheral vascular disease, and flaps used in reconstructive surgery.
Funding
Swedish Government for Clinical Research (ALF); Skånes universitetssjukhus; Region Kronoberg; Skåne County Council's Research and Development Foundation; Lund University Grant for Research Infrastructure; Swedish Cancer Foundation; Stiftelsen Kronprinsessan Margaretas Arbetsnämnd för Synskadade; Stiftelsen för Synskadade i f.d. Malmöhus län; Lund Laser Center; IngaBritt och Arne Lundbergs Forskningsstiftelse; Cronqvist Foundation; Sveriges Läkarförbund.
Disclosures
The authors declare no conflicts of interest
References
1. A. Kara, S. Akin, and C. Ince, “Monitoring microcirculation in critical illness,” Curr. Opin. Crit. Care 22(5), 444–452 (2016). [CrossRef]
2. J. Zötterman, M. Bergkvist, F. Iredahl, E. Tesselaar, and S. Farnebo, “Monitoring of partial and full venous outflow obstruction in a porcine flap model using laser speckle contrast imaging,” J. Plast. Reconstr. Aesthetic Surg. 69(7), 936–943 (2016). [CrossRef]
3. P. Vaupel, A. Mayer, and M. Höckel, “Tumor hypoxia and malignant progression,” Methods Enzymol. 381, 335–354 (2004). [CrossRef]
4. J. Xia, A. Danielli, Y. Liu, L. Wang, K. Maslov, and L. V. Wang, “Calibration-free quantification of absolute oxygen saturation based on the dynamics of photoacoustic signals,” Opt. Lett. 38(15), 2800 (2013). [CrossRef]
5. R. B. Buxton, “The physics of functional magnetic resonance imaging (fMRI),” Rep. Prog. Phys. 76(9), 096601 (2013). [CrossRef]
6. D. Rischin, R. J. Hicks, R. Fisher, D. Binns, J. Corry, S. Porceddu, and L. J. Peters, ““Prognostic significance of [18F]-misonidazole positron emission tomography-detected tumor hypoxia in patients with advanced head and neck cancer randomly assigned to chemoradiation with or without tirapazamine: A substudy of Trans-Tasman Radiation Oncology Group Study 98.02,” J. Clin. Oncol. 24(13), 2098–2104 (2006). [CrossRef]
7. M. W. Wukitsch, M. T. Petterson, D. R. Tobler, and J. A. Pologe, “Pulse oximetry: Analysis of theory, technology, and practice,” J. Clin. Monit. 4(4), 290–301 (1988). [CrossRef]
8. M. Kohl-Bareis, C. Leithner, H. Sellien, R. Guertler, D. Gerskin, B. Rohrer, G. Royl, U. Dirnagl, and U. Lindauer, “System for the measurement of blood flow and oxygenation in tissue applied to neurovascular coupling in brain,” Opt. InfoBase Conf. Pap. (October 2005) (2005).
9. P. Avci, A. Gupta, M. Sadasivam, D. Vecchio, Z. Pam, N. Pam, and M. R. Hamblin, “Low-level laser (light) therapy (LLLT) in skin: Stimulating, healing, restoring,” Semin. Cutan. Med. Surg. 32(1), 41–52 (2013).
10. K. S. Valluru and J. K. Willmann, “Clinical photoacoustic imaging of cancer,” Ultrasonography 35(4), 267–280 (2016). [CrossRef]
11. C. Liu, Y. Liang, and L. Wang, “Single-shot photoacoustic microscopy of hemoglobin concentration, oxygen saturation, and blood flow in sub-microseconds,” Photoacoustics 17, 100156 (2020). [CrossRef]
12. L. J. Rich and M. Seshadri, “Photoacoustic imaging of vascular hemodynamics: Validation with blood oxygenation level-dependent MR imaging,” Radiology 275(1), 110–118 (2015). [CrossRef]
13. M. Martinho Costa, A. Shah, I. Rivens, C. Box, T. O’Shea, E. Papaevangelou, J. Bamber, and G. ter Haar, “Quantitative photoacoustic imaging study of tumours in vivo: Baseline variations in quantitative measurements,” Photoacoustics 13, 53–65 (2019). [CrossRef]
14. M. Chen, H. J. Knox, Y. Tang, W. Liu, L. Nie, J. Chan, and J. Yao, “Simultaneous photoacoustic imaging of intravascular and tissue oxygenation,” Opt. Lett. 44(15), 3773 (2019). [CrossRef]
15. K. Okumura, J. Matsumoto, Y. Iwata, K. Yoshida, N. Yoneda, T. Ogi, A. Kitao, K. Kozaka, W. Koda, S. Kobayashi, D. Inoue, N. Sakai, K. Furuichi, T. Wada, and T. Gabata, “Evaluation of renal oxygen saturation using photoacoustic imaging for the early prediction of chronic renal function in a model of ischemia-induced acute kidney injury,” PLoS One 13(12), e0206461 (2018). [CrossRef]
16. Z. Li and H. Chen, “In vivo determination of acute myocardial ischemia based on photoacoustic imaging with a focused transducer,” J. Biomed. Opt. 16(7), 076011 (2011). [CrossRef]
17. A. Becker, M. Masthoff, J. Claussen, S. J. Ford, W. Roll, M. Burg, P. J. Barth, W. Heindel, M. Schäfers, M. Eisenblätter, and M. Wildgruber, “Multispectral optoacoustic tomography of the human breast: characterisation of healthy tissue and malignant lesions using a hybrid ultrasound-optoacoustic approach,” Eur. Radiol. 28(2), 602–609 (2018). [CrossRef]
18. G. Diot, S. Metz, A. Noske, E. Liapis, B. Schroeder, S. V. Ovsepian, R. Meier, E. Rummeny, and V. Ntziachristos, “Multispectral Optoacoustic Tomography (MSOT) of human breast cancer,” Clin. Cancer Res. 23(22), 6912–6922 (2017). [CrossRef]
19. M. Masthoff, A. Helfen, J. Claussen, W. Roll, A. Karlas, H. Becker, G. Gabriëls, J. Riess, W. Heindel, M. Schäfers, V. Ntziachristos, M. Eisenblätter, U. Gerth, and M. Wildgruber, “Multispectral optoacoustic tomography of systemic sclerosis,” J. Biophotonics 11(11), 1–7 (2018). [CrossRef]
20. F. Knieling, C. Neufert, A. Hartmann, J. Claussen, A. Urich, C. Egger, M. Vetter, S. Fischer, L. Pfeifer, A. Hagel, C. Kielisch, R. S. Görtz, D. Wildner, M. Engel, J. Röther, W. uter, J. Siebler, R. Atreya, W. Rascher, D. Strobel, M. F. Neurath, and M. J. Waldner, “Multispectral Optoacoustic Tomography for Assessment of Crohn’s Disease Activity,” J. Infect. Dis. 376(13), 1292–1294 (2017). [CrossRef]
21. J. R. Eisenbrey, M. Stanczak, F. Forsberg, F. A. Mendoza-Ballesteros, and A. Lyshcik, “Photoacoustic oxygenation quantification in patients with Raynaud’s: first-in-human results,” Ultrasound Med. Biol. 44(10), 2081–2088 (2018). [CrossRef]
22. A. Taruttis, A. C. Timmermans, P. C. Wouters, M. Kacprowicz, G. M. van Dam, and V. Ntziachristos, “Optoacoustic imaging of human vasculature: Feasibility by using a handheld probe,” Radiology 281(1), 256–263 (2016). [CrossRef]
23. A. Helfen, M. Masthoff, J. Claussen, M. Gerwing, W. Heindel, V. Ntziachristos, M. Eisenblätter, M. Köhler, and M. Wildgruber, “Multispectral optoacoustic tomography: intra- and interobserver variability using a clinical hybrid approach,” J. Clin. Med. 8(1), 63 (2019). [CrossRef]
24. R. Bi, U. S. Dinish, C. C. Goh, T. Imai, M. Moothanchery, X. Li, J. Y. Kim, S. Jeon, Y. Pu, C. Kim, L. G. Ng, L. V. Wang, and M. Olivo, “In vivo label-free functional photoacoustic monitoring of ischemic reperfusion,” J. Biophotonics 12, e201800454 (2019). [CrossRef]
25. F. Duan, H. Ma, J. Zhang, S. Li, H. Li, Z. Wu, F. Hong, Z. Lüming, and L. Nie, “Optical-resolution photoacoustic microscopy continually monitors macrophages activities of acute inflammation in vivo,” Chin. Opt. Lett. 18(12), 121701 (2020). [CrossRef]
26. J. Lv, S. Li, J. Zhang, F. Duan, Z. Wu, R. Chen, M. Chen, S. Huang, H. Ma, and L. Nie, “In vivo photoacoustic imaging dynamically monitors the structural and functional changes of ischemic stroke at a very early stage,” Theranostics 10(2), 816–828 (2020). [CrossRef]
27. J. Yang, G. Zhang, W. Chang, Z. Chi, Q. Shang, M. Wu, T. Pan, L. Huang, and H. Jiang, “Photoacoustic imaging of hemodynamic changes in forearm skeletal muscle during cuff occlusion,” Biomed. Opt. Express 11(8), 4560 (2020). [CrossRef]
28. A. Karlas, M. Kallmayer, N.-A. Fasoula, E. Liapis, M. Bariotakis, M. Krönke, M. Anastasopoulou, J. Reber, H.-H. Eckstein, and V. Ntziachristos, “Multispectral optoacoustic tomography of muscle perfusion and oxygenation under arterial and venous occlusion: A human pilot study,” J. Biophotonics 13(6), 1–9 (2020). [CrossRef]
29. J. Reber, M. Willershäuser, A. Karlas, K. Paul-Yuan, G. Diot, D. Franz, T. Fromme, S. V. Ovsepian, N. Bézière, E. Dubikovskaya, D. C. Karampinos, C. Holzapfel, H. Hauner, M. Klingenspor, and V. Ntziachristos, “Non-invasive measurement of brown fat metabolism based on optoacoustic imaging of hemoglobin gradients,” Cell Metab. 27(3), 689–701.e4 (2018). [CrossRef]
30. T. B. Fitzpatrick, “The validity and practicality of sun-reactive skin types i through VI,” Arch. Dermatol. 124(6), 869–871 (1988). [CrossRef]
31. R. Sheikh, M. Cinthio, U. Dahlstrand, T. Erlöv, M. Naumovska, B. Hammar, S. Zackrisson, T. Jansson, N. Reistad, and M. Malmsjö, “Clinical translation of a novel photoacoustic imaging system for examining the temporal artery,” IEEE Trans. Ultrason. Ferroelectr. Freq. Control 66(3), 472–480 (2019). [CrossRef]
32. Y.-W. Ri, S.-H. Jong, and S.-J. Im, “Theoretical prediction of the source-detector separation distance suited to the application of the spatially resolved spectroscopy from the near-infrared attenuation data cube of tissues,” arXiv:1409.4246v1, 1–9 (2014).
33. N. Reistad, M. Mayjonade, A. Ahadi, and S. Andersson-Engels, “Characterization of probe contact effects on diffuse reflectance spectroscopy measurements,” Biophotonics South Am. 9531, 953143 (2015). [CrossRef]
34. Z. Marcinkevics, U. Rubins, A. Grabovskis, J. Cimurs, and A. Caica, “Hyperspectral evaluation of skin blood oxygen saturation at baseline and during arterial occlusion,” Proc. SPIE 10685, 106851A (2018). [CrossRef]
35. N. Keshava and J. F. Mustard, “Spectral unmixing,” IEEE Signal Process. Mag. 19(1), 44–57 (2002). [CrossRef]
36. S. L. Jacques, “Optical properties of biological tissues: A review,” Phys. Med. Biol. 58(14), 5007–5008 (2013). [CrossRef]
37. T. W. Judi and J. D. Everall, “The thickness of the epidermis,” Br. J. Dermatol. 89(5), 467–476 (1973). [CrossRef]
38. I. Fredriksson, M. Larsson, and T. Strömberg, “Optical microcirculatory skin model: assessed by Monte Carlo simulations paired with in vivo laser Doppler flowmetry,” J. Biomed. Opt. 13(1), 014015 (2008). [CrossRef]
39. B. A. Wagner, S. Venkataraman, and G. R. Buettner, “The rate of oxygen utilization by cells,” Free Radical Biol. Med. 51(3), 700–712 (2011). [CrossRef]
40. K. Ohara, “Studies on the oxygen consumption of human skin tissues, with special reference to that of sweat glands,” Jpn. J. Physiol. 2, 1–8 (1951). [CrossRef]
41. J. Kanitakis, “Anatomy, histology and immunohistochemistry of normal human skin,” Eur. J. Dermatol. 12(4), 390–401 (2002).
42. I. M. Braverman, “The cutaneous microcirculation,” J. Invest. Dermatol. Symp. Proc. 5(1), 3–9 (2000). [CrossRef]
43. G. Bergers and S. Song, “The role of pericytes in blood-vessel formation and maintenance,” Neuro. Oncol. 7(4), 452–464 (2005). [CrossRef]
44. B. Ceylan, L. Khorshid, Ü Yapucu Günes, and A. Zaybak, “Evaluation of oxygen saturation values in different body positions in healthy individuals,” J. Clin. Nurs. 25(7-8), 1095–1100 (2016). [CrossRef]
45. S. Tzoumas, A. Nunes, I. Olefir, S. Stangl, P. Symvoulidis, S. Glasl, C. Bayer, G. Multhoff, and V. Ntziachristos, “Eigenspectra optoacoustic tomography achieves quantitative blood oxygenation imaging deep in tissues,” Nat. Commun. 7(1), 12121 (2016). [CrossRef]
46. I. V. Meglinski and S. J. Matcher, “Computer simulation of the skin reflectance spectra,” Comput. Methods Programs Biomed. 70(2), 179–186 (2003). [CrossRef]
47. D. M. Rogge, B. Rivard, Z. Zhang, and J. Feng, “Iterative spectral unmixing for optimizing per-pixel endmember sets,” IEEE Trans. Geosci. Remote Sens. 44(12), 3725–3736 (2006). [CrossRef]
48. B. Cox, J. G. Laufer, S. R. Arridge, and P. C. Beard, “Quantitative spectroscopic photoacoustic imaging: a review,” J. Biomed. Opt. 17(6), 061202 (2012). [CrossRef]
