Noninvasive staging of pressure ulcers using photoacoustic imaging

Abstract

Ulcers including pressure ulcers and diabetic foot ulcers damage the skin and underlying tissue in people with compromised blood circulation. They are classified into four stages of severity and span from mild reddening of the skin to tissue damage and muscle/bone infections. Here, we used photoacoustic imaging as a noninvasive method for detecting early tissue damage that cannot be visually observed while also staging the disease using quantitative image analysis. We used a mouse model of pressure ulcers by implanting subdermal magnets in the dorsal flank and periodically applying an external magnet to the healed implant site. The magnet-induced pressure was applied in cycles, and the extent of ulceration was dictated by the number of cycles. We used both laser- and light-emitting diode (LED)-based photoacoustic imaging tools with 690 nm excitation to evaluate the change in photoacoustic signal and depth of injury. Using laser-based photoacoustic imaging system, we found a 4.4-fold increase in the photoacoustic intensity in stage I vs. baseline (no pressure). We also evaluated the depth of injury using photoacoustics. We measured a photoacoustic ulcer depth of 0.38 ± 0.09 mm, 0.74 ± 0.11 mm, 1.63 ± 0.4 mm, and 2.7 ± 0.31 mm (n = 4) for stages I–IV, respectively. The photoacoustic depth differences between each stage were significant (p < 0.05). We also used an LED-based photoacoustic imaging system to detect early stage (stage I) pressure ulcers and observed a 2.5-fold increase in photoacoustic signal. Importantly, we confirmed the capacity of this technique to detect dysregulated skin even before stage I ulcers have erupted. We also observed significant changes in photoacoustic intensity during healing suggesting that this approach can monitor therapy. These findings were confirmed with histology. These results suggest that this photoacoustic-based approach might have clinical value for monitoring skin diseases including pressure ulcers.

INTRODUCTION

Chronic wounds including diabetic ulcers and decubitus ulcers are a pervasive and expensive health-care challenge, but tools to diagnose these wounds before they have erupted or evaluate deep tissue response to therapy have remained elusive.^1,2 Chronic wounds cost the medical infrastructure US$25B annually with a single diabetic ulcer costing nearly US$50,000.³ These numbers are projected to grow as the population ages. A variety of treatments have been proposed including advanced wound dressings, skin substitutes, hyperbaric oxygen therapy, and growth factor-based therapies⁴; however, these have limited capacity for complete healing especially in advanced wounds. Several studies have shown that the prevention of wounds offers significant costs savings vs. wound treatment.^5,6

The Healthcare Research and Quality Agency recently estimated that pressure ulcers exert an annual burden of $9.1–$11.6 billion on the US health-care system.⁷ Pressure ulcers consist of necrosis and ulceration.^8,9 The location of ulceration is where tissues are compressed between bony prominences and hard surfaces. While pressure is the main cause, they are also affected by friction, shearing forces, and moisture.¹⁰ Common risk factors include age >65, impaired circulation/tissue perfusion, immobilization, undernutrition, decreased sensation, and incontinence.^11,12

Pressure ulcers are defined by both the depth of ulceration and types of tissue affected.¹³ In stage I, the epidermis appears reddened and is characterized by non-blanchable erythema under light pressure. Stage II is defined by ruptured skin and a loss of the epidermis and dermis, i.e., the formation of a visual ulcer. Stage III involves full skin loss—the lesion extends to subcutaneous tissue. Whole skin loss, muscle necrosis, and damage to tendons, and joints occur in stage IV.^13-16 Ulcer necrosis is attributed to the loss of blood flow under sustained pressure.^17,18

Visual inspection is the standard-of-care for triaging pressure ulcers.¹⁷ In some cases, pressure-sensitive devices are used to alert medical staff when a certain pressure threshold is exceeded. However, this threshold value is patient dependent and varies with physiological metrics such as body mass index. The Hematron sensor can measure the thermal conductivity of skin, which is correlated with blood flow.¹⁹ However, this sensor cannot monitor the progressive tissue damage that characterizes ulcer development.²⁰ One of the most well-known methods for pathology detection and tissue characterization is ultrasound elastography.²¹ This method extends the typical information supplied by ultrasound—a real time, affordable, and non-invasive modality. Ultrasound elastography is considered an intrinsic factor evaluation to estimate the stiffness of tissue by measuring strain making it applicable to detection of pressure ulcers.²² This method is limited by superficial tissue assessment, artifacts from compression, and examiner dependency.²³ Nixon et al. utilized laser Doppler imaging to validate the clinical grading of erythema in 37 pressure ulcer patients and demonstrated that this method could monitor a range of blood flow values for normal skin and areas of erythema.¹³ However, the laser Doppler imaging method suffers from low penetration depth and therefore monitoring the injury on underlying tissues is impossible. Recently, Swisher et al. used flexible electrode arrays to measure the impedance correlated with tissue health and wound types in a rat model.²⁰ However, this device is not able to measure the effect of ulcers on subcutaneous fat tissue and muscle.

None of the above methods can monitor the subdermal extent of pressure damage in real-time. If damage to subdermal tissue could be identified before it is visible by eye, then ulcer-associated morbidity and costs could be prevented before they incur. Therefore, improved methods for detecting the extent of ulceration beneath the skin are of significant interest.

Photoacoustic imaging is a noninvasive and high-resolution technique that combines the contrast of optical imaging and the resolution of ultrasound technique.^24-26 This hybrid imaging modality offers higher penetration depth with less scatter than optical imaging.²⁷ In photoacoustic imaging, the tissue is illuminated by a nanosecond laser pulse. The wavelength is selected based on the maximum absorption coefficient of the target tissue. The photons absorbed by the tissue lead to a spatially confined temperature increases in turn leading to rapid volume expansion and photoacoustic pressure waves. The generated photoacoustic signals are detected by wideband ultrasound transducers.^25,28 Photoacoustic signals are received and stored after each laser pulse. Photoacoustic images are produced using various reconstruction algorithms such as delayand-sum (DAS)²⁹ or Fourier transform analysis (FTA).³⁰ Photoacoustic data overlaid on ultrasound images can report functional information and structural details.³¹

We hypothesized that the dysregulated vasculature associated with pressure ulcers could be imaged to study their development and predict eruption. Here, for the first time, we introduce the use of photoacoustic imaging to detect early stage pressure ulcers and monitor their development across different stages using an established murine model.¹⁶ We also utilized this imaging technique to monitor the healing and therapy associated with pressure ulcer treatment.

MATERIAL AND METHODS

Animal model and validation

Twenty-five nude mice (8–10 weeks, 25–35 g) were purchased from the University of California San Diego Animal Care and Use Program (ACP). They were kept in separate cages under a 12-hour light–dark cycle and sterile environment at constant temperature and humidity. All animal experiments were performed in accordance with NIH guidelines and approved by the Institutional Animal Care and Use Committee (IACUC) at the University of California, San Diego.

For the surgical implantation of the internal magnet (diameter of 7 mm and thickness of 1 mm; K&J Magnetic, Inc.), mice were anesthetized with 1–2% isoflurane and placed on a heating bed. The magnet was sterilized via autoclave. The skin was disinfected with povidone iodine at the surgical site. After creating a 5-mm incision, we placed the sterile magnet under the greater gluteus muscle¹⁶ and away from the incision. The incision was sealed using topical tissue adhesive (Abbott Laboratories, IL). We waited 10 days for healing to ensure that the surgical wound did not interfere with the pressure induction site.

A sterile external magnet (identical to the internal magnet) was then placed on the skin surface above the implant site at regular intervals. The external magnet was applied in cycles. Each cycle consisted of 2 hours of magnet-induced pressure followed by 1 hour of release. Wassermann et al. demonstrated in their model that 4, 6, 8, and 10 cycles were needed to create stages I–IV ulcers, respectively, and we followed this protocol.¹⁶ Imaging was performed either between cycles or between stages. Three mice were sacrificed at baseline conditions and after each stage for histology. The skin and underlying tissue were fixed in buffered 10% formaldehyde solution. Histological analysis (H&E staining) was performed to stage the ulcer as an independent method.

Photoacoustic imaging

We used both laser and light-emitting diode (LED)-based photoacoustic imaging systems to perform all in vivo procedures. The Vevo LAZR (VisualSonic Inc.) is a laser integrated high-frequency ultrasound system that utilizes a linear array transducer (LZ-550; Fc = 40 MHz) with optical fibers integrated to both sides of the transducer. For optical excitation, this system uses a Q-switched Nd:YAG laser (4–6 ns pulse width) with a repetition rate of 20 Hz (frame rate of 6 Hz) followed by an optical parametric oscillator (tunable wavelength 680–970 nm). The laser intensity on the surface of the skin was measured at 12.65 ± 0.65 mJ/cm² using a laser pyroelectric energy sensor (PE50BF-C; Ophir LLC).

We also used a LED-based photoacoustic imaging system (CYBERDYNE Inc.).³² The system is equipped with a 128-element linear array ultrasound transducer with a central frequency of 10 MHz and a bandwidth of 80.9% fitted with two 690 nm LED arrays. The repetition rate of these LEDs is tunable between 1, 2, 3, and 4 K Hz. The pulse width can be changed from 50 ns to 150 ns with a 5-ns step size. The LED intensity at 4 K Hz on the surface of skin was measured at 5.5 μJ/cm² using a photodiode sensor (S120C; Thorlabs Inc.). In both systems, the transducer can be scanned to generate three-dimensional (3D) images using a maximum intensity projection (MIP) algorithm.

Quantitative and statistical analysis

We collected B-mode photoacoustic/ultrasound images from five different positions above the target site in each animal before initiating the pressure ulcer model (baseline conditions) and after each stage. Each B-mode image contains 256 A-line scans (an A-line scan is simply a single line of collected data). We used 690 nm as the excitation wavelength for all scans. We analyzed our data using two different categories: photoacoustic intensity and photoacoustic depth. The intensity was quantified by first converting all photoacoustic data to 8-bit images; the mean values and standard deviations of the accumulative photoacoustic pixel intensities from 3 mm × 3 mm regions of interest (ROI; each ROI contained 50 A-line scans) were measured using ImageJ (Bethesda, MD).

To quantitate the extent of disease, photoacoustic depth was calculated via 10 A-line intensity profiles from each ROI (every fifth A-line) on all photoacoustic images in all four animals. For the photoacoustic images, we measured the average and standard deviation of ulcer depths from 10 different A-lines. The error bars in each figure represent the standard deviation from four different animals, and p values lower than 0.05 were considered significantly different.

RESULTS

Pressure ulcers and diabetic foot ulcers are known to dysregulate the vasculature of tissue, but this dysregulation is often difficult to detect until the lesion has advanced to stage II or III where it has erupted through the skin. Thus, we hypothesized that photoacoustic imaging could be used to detect early stage lesions because it quantitates tissue absorption including from hemoglobin and deoxyhemoglobin. This early stage detection of pressure ulcers could then be used to direct treatment and prevent the ulcer from progressing further and disrupting the epithelial barrier—these ruptured lesions are what cause the long-term complications.

We used a rodent model of human pressure ulcers to validate this imaging modality. There are several animal models of pressure ulcers^33-35 including swine and rats.^36-38 They usually apply pressure on the skin against the underlying bone; however, the use of anesthesia during this long procedure is the main limitation of this method.⁷ This model is limited to stage I ulcers due to the superficial position of the steel plate under the skin. Wassermann et al. introduced a chronic pressure model to induce all four stages by implanting a magnet under both the skin and deeper tissue layers, including muscle (Figure. 1). Their model was verified with histology.¹⁶

Figure 1.

Experimental procedures. All animals were anesthetized with 1–2% isoflurane and placed on a heating bed. The animals were allowed to heal for 10 days after implanting the internal magnet to ensure that the incision did not interfere with the pressure induction site. To create stages I–IV pressure ulcers, 4, 6, 8, and 10 cycles were applied, respectively.¹⁶ Each cycle included 2 hours of pressure followed by 1 hour of release. Photoacoustic/ultrasound images were acquired at baseline and at each stage with a 40 MHz transducer. Histology analysis (H&E staining) was used to confirm ulcer stage.

Our first goal was to confirm that the model was created correctly. Figure 2A(i)-(iv) are photographs of the ulcer sites in mice at stages I–IV, respectively. These photographs show the effect of ulcers on the skin but not the underlying tissues. These same animals were then studied with ultrasound/photoacoustic imaging to evaluate the wound depth. Figure 2B represents the average histogram of five B-mode baseline fields-of-view for four animals. All average pixel intensities are lower than 20, and thus we used this number as the threshold for detecting the presence of ulcers using photoacoustic imaging. The inset in Figure 2B is a baseline photoacoustic/ultrasound image. Figure 2C, E, G, and I shows the photoacoustic/ultrasound image with progression through stages I–IV, respectively. Figure 2D, F, H, and J is representative photoacoustic A-line profiles (dotted line), which correspond to Figure 2C, E, G, and I, respectively. Note that the full analysis used 10 line profiles per image. We considered pixels with an 8-bit photoacoustic intensity >20 to represent ulceration of the underlying tissues.

Figure 2.

Evaluation of the photoacoustic response to various stages of pressure ulcers. (A) (i)–(iv) Photographs of stages IIV ulcers. (B) Average histogram of baseline photoacoustic pixel intensities. At baseline, all pixel intensities are lower than 20. The inset shows the photoacoustic/ultrasound image at baseline. (C) Ultrasound/photoacoustic image at stage I. (D) Photoacoustic A-line profile for the dotted line in panel C. We defined photoacoustic pixels with intensities higher than 20 as dysregulated tissue. This was repeated for stage II (E, F), stage III (G, H), and stage IV (I, J). The dotted line shows the A-line profile that was used to quantify the depth of pressure ulcer at different stages. The dotted rectangles show the ROIs.

Figure 3A shows that significant changes in photoacoustic intensity were seen between baseline (no pressure ulcer) and all four stages (p < 0.05). We found a 4.4-fold increase in photoacoustic intensity at stage I in comparison to the baseline. We saw no any significant changes between stages I–IV in photoacoustic intensity. Figure 3B quantifies the photoacoustic depth effect of the pressure ulcer on the skin and underlying tissues. We measured photoacoustic ulcer depths of 0.38 ± 0.09 mm, 0.74 ± 0.11 mm, 1.63 ± 0.4 mm, and 2.7 ± 0.11 mm for stages I–IV, respectively. There were significant differences between each stage (p < 0.05). We also monitored the animal 30, 60, and 90 minutes after stage I to simulate healing/therapy. Figure 3C shows photoacoustic/ultrasound images at 0, 30, 60, and 90 minutes healing after stage I. Figure 3D quantifies the photoacoustic intensity effect of healing procedure. We observed significant decrease in photoacoustic intensity after 60 minutes (p < 0.05).

Figure 3.

Quantitative and statistical analysis of photoacoustic data at each pressure ulcer stage and during healing. (A) Photoacoustic intensity as a function of ulcer stage. There was a significant difference in the photoacoustic intensity at baseline vs. stages I–IV. (B) Quantitative analysis for the depth effect at different ulcer stages. Photoacoustic ulcer penetration is significantly different between stages. (C) Ultrasound/photoacoustic image at 0, 30, 60, and 90 minutes after stage I to simulate healing/therapy. The dotted rectangles show the ROIs. (D) Photoacoustic intensity as function of healing time. There was significant change in the photoacoustic intensity after 60 minutes of healing. Error bars represent the standard deviations among four (A and B) and three (D) different animals. * Indicates p < 0.05.

These initial data were collected with a laser-based scanner; however, these systems are bulky, delicate, and expensive. More recently, LED-based photoacoustic imaging system has offered important improvements in size, cost, and stability.^32,39,40 Thus, we also evaluated these ulcers with LED-based photoacoustics. Figure 4 shows baseline (top) and stage I ulcers (bottom) via ultrasound, photoacoustic, and overlay ultrasound/photoacoustic using LED excitation. The insets in Figure 4B and C are photographs of mice without and with pressure ulcers, respectively. We observed a 2.5-fold increase in the photoacoustic signal using the LED-based photoacoustic imaging system at stage I.

Figure 4.

LED-based photoacoustic evaluation of pressure ulcers at stage I. (A) B-mode ultrasound image at baseline conditions when no pressure has been applied. (B) B-mode photoacoustic image at baseline at the same position as panel A. Minor photoacoustic signal is observed from the epidermis. The photographic inset shows the mouse in absence of ulcer. (C) B-mode photoacoustic/ultrasound overlay at baseline conditions. (D) B-mode ultrasound image at stage I. (E) B-mode photoacoustic image at stage I at the same position as panel D. We observed a 2.5-fold increase in photoacoustic intensity compared to baseline. The photographic inset shows the stage I ulcer. (F) B-mode photoacoustic/ultrasound overlay at stage I. The image depth is 1 cm and the scale bars are 2 mm.

These preclinical results suggest that LED-based photoacoustics imaging system (Figure 4) has value for monitoring and staging these ulcers. Photoacoustic imaging is able to measure the status of underlying tissues without performing any invasive measurements such as histology (Figures 1, 4, and 5).

Figure 5.

Pressure ulcer detection pre-stage I. (A) The experimental procedure. Imaging was done with laser-based photoacoustic system before starting and after each cycle. (B) Photoacoustic data at each cycle. We observed a significant difference (p < 0.05) in photoacoustic intensity between second vs. third cycle and third versus fourth cycle within stage I pressure ulcers. The insets show the photograph images of ulcer on animal body after each cycle.

Our final goal was to monitor pre-stage I ulcers. Thus, we imaged after each of the cycles between baseline and stage I (Figure 5A). There was no difference in photoacoustic intensity among the baseline, first, and second cycles. However, there was a significant increase in photoacoustic intensity between baseline/first cycle/second cycle and third/fourth cycle (p <0.05) (Figure 5B). Therefore, photoacoustics can detect ulcers prior to the typical stage I classification. This imaging technique is sensitive enough to detect the mild physiological changes (third and fourth cycles) that are not visible to the naked eye (Figure 5B).

Finally, we validated the model with histology. Figure 6A and B is histological images of the skin and underlying tissue, respectively, for the baseline condition (no pressure ulcer). Figure 6C shows histology image of the skin at stage I. The red arrow shows the superficial and epidermal skin loss in stage I. Figure 6D shows the histology of muscle at stage I—this is normal at stage I. Figure 6E and F shows histological images of skin and muscle, respectively, at stage II. The structure of dermis and epidermis is now disrupted with mild necrosis on underlying tissues. Stage III leads to full-thickness skin (Figure 6G) loss (red arrows) and necrotic areas in subcutaneous tissue layers (arrows; Figure 6H). By stage IV, all skin was removed (red arrow; Figure 6I) with large necrotic regions of muscle (arrows, Figure 6J). This confirmed that we the imaging data reflect dysregulated biology via the four-stage pressure ulcer from the Wassermann model.¹⁶

Figure 6.

Histological evaluation of pressure ulcer induction model. (A) Histology of skin samples for control animals. Typical structure of epidermis and dermis is shown. (B) Histology of muscle sample for control animals. (C) Histology of skin sample for the animals with stage I ulcers. The superficial and epidermal skin loss in stage I is shown using an arrow. (D) Histology of muscle sample for the animals with stage I ulcers. No sign of ulcer was found on muscle tissue in stage I of the pressure ulcer. (E) Histological images from skin at stage II. The arrow shows the disruption in the structure of dermis and epidermis. (F) Histological image from the muscle. Light necrosis on muscle histology image is observed. (G) Histology image of skin in stage III of pressure ulcer. Full loss of epidermis and dermis is the consequences of this stage of ulcer on the skin. (H) Muscle histology at stage III of the pressure ulcer. The arrows show the increase of necrosis on underlying tissues. (I) Histological image of skin in stage IV of ulcer. Full skin was removed as the arrow represents. (J) Histology of muscle sample for stage IV animals. Large necrotic regions of the muscle are shown using the arrows.

DISCUSSION

Pressure ulcers are debilitating and can significantly impair quality of life.¹ They are associated with loss of pain sensation and disordered circulation.^41,42 The gold standard to preventing pressure ulcers include regular patient turning/repositioning.⁴³ However, there are relatively few tools for molecular-level insight into when to reposition and who to reposition. This work describes a noninvasive, high-resolution imaging technique that utilizes photoacoustic signal to detect pressure-induced tissue damage in a nude mouse model in vivo and may have utility in predicting the timing of repositioning.

We utilized both the intensity of photoacoustic signal and the depth of the photoacoustic signal to characterize ulceration (Figures 2 and 3). We and others have shown that the photoacoustic signal changes can report the presence of inflammation and erythema.^44,45 The photoacoustic depth penetration can distinguish the degree of injury analogous to conventional staging of pressure ulcers. Surprisingly, the signal intensity had less utility in differentiating ulcer stage; however, we can quantify the level of injury by calculating the position of high-intensity (bit depth >20 [8-bit images]).

We used both laser and LED-based equipment. While the laser-based system has more power and higher resolution, LED-based equipment offers significant cost advantages that might aid in clinical translation.^32,39 The results of LED-based photoacoustic imaging (Figure 4) could easily discriminate between different stages. More importantly, this equipment had sufficient sensitivity to detect the mild physiological changes associated with pre-stage 1 lesions that are not visible to the naked eye (Figure 5B). The value of imaging is that it offers a 3D map of the underlying tissue without invasive measurements such as histology (Figures 1, 4, and 5).

Several imaging techniques have been used for monitoring the pressure ulcers: laser Doppler imaging,¹³ fluorescence imaging,⁴⁶ ultrasound imaging,⁴⁷ magnetic resonance imaging,⁴⁸ and hyperspectral imaging.⁴⁹ They all have their own specific limitations. Laser Doppler, fluorescence, and hyperspectral imaging are optical techniques and suffer from low penetration depth. Thus, these approaches cannot monitor the progress of injury and healing in underlying tissues and muscles.^13,46,49 Aoi et al. used conventional ultrasound to monitor the pressure ulcers longitudinally. They concluded that imaging of the discontinuous fascia and heterogeneous hypoechoic area could predict future progression of pressure ulcers.⁴⁷ However, this technique lacks contrast and cannot monitor the dysregulated vasculature associated with pressure ulcers. Magnetic resonance imaging has been used to evaluate pressure ulcer in making clinical decisions,⁵⁰ but it is costly and time-consuming. Photoacoustic imaging provides high spatial and temporal resolution as well as sufficient penetration depth for monitoring ulcers. It is also well suited for multimodal imaging via several other techniques. We used photoacoustic in combination with ultrasound to add contrast to the data.

Photoacoustic approach does have some limitations. First, it is difficult to image through bones, and bones can cause reflections.⁵¹ Second, the amplitude of the photoacoustic signal from deep tissues might not represent the severity of ulcer (Figure 3A). This is because photoacoustic signal is directly related to the intensity of incident light on tissue: Deep tissues receive less light due to optical absorption and scattering.^52,53 This could potentially be solved with image-processing tools than compensate for this reduced fluence.^54-56 Fortunately, we can also circumvent this by quantitating photoacoustic signal depth (Figure 3B), which does correspond to stage. Third, questions always exist about how well murine models recapitulate human disease; however, the model used here was confirmed histologically to present a phenotype consistent with human ulcers. We also used spectral photoacoustic imaging to measure the oxygen saturation among the different ulcer stages (data not shown). Somewhat surprisingly, there was no trend across stages perhaps because of spectral changes in the epidermis due to scabbing and scarring. This shifting baseline might affect the accuracy of oxygen saturation measurements in vivo.

CONCLUSIONS

There is a clinical need for a noninvasive technique that can detect early tissue damage that would otherwise go unnoticed. For the first time, we introduce photoacoustic imaging as a noninvasive tool that can be utilized to detect pressure ulcers before stage I. Here, we used a published protocol to produce pressure ulcers in nude mice. We observed significant changes in photoacoustic intensity even before stage I of pressure ulcer. We also demonstrated that the stage of ulceration can be determined by quantifying the depth of photoacoustic signal from the injury. We also observed that photoacoustic imaging can monitor ulcer healing suggesting potential clinical value in monitoring therapeutic response.