Visualization of Wound Healing Progression With Near Infrared Spectroscopy: A Retrospective Study

Objective. The aim of this retrospective study is to determine if near infrared spectroscopy (NIRS) can be used to evaluate wounds and adjacent soft tissues to identify patterns involved in tissue oxygenation and wound healing as well as predict which wounds may or may not heal. Materials and Methods. In this study, 25 patients with either diabetic foot ulcers or venous leg ulcers were examined retrospectively to determine if NIRS could be used to predict which wounds may or may not close. All patients had either diabetic or venous ulcers and were being actively treated in the clinic. Regardless of the treatment rendered, all wounds were tracked with NIRS at regular intervals. Retrospectively, the de-identified images were reviewed to determine any patterns that might exist. Wound bed and periwound oxygenation patterns were observed and classified, including correlation with both the clinical appearance and the NIRS images. Images of wounds that closed and those that did not were compared. Results. Four distinct patterns of tissue oxygenation that appeared to have some value for predicting which wounds would heal, and which would not, were identified among the 25 patients. A mechanism has also been proposed to try to explain the patterns of healing observed; Hyperperfusion, Imbibition, Neovascularization, and Trailing (HINT) describes various aspects of these patterns. Conclusions. As with any imaging technology, both qualitative and quantitative data are used to determine what is happening clinically. This study represents an early attempt to understand the role of NIRS and percent oxygenated hemoglobin in the wound healing process. It also lays the groundwork for identifying patterns associated with wound closure.

The closure of chronic wounds can be a complex and difficult process that is highly dependent on the optimization of conditions to facilitate healing. In order to understand the conditions that favor wound closure, the factors that are considered detrimental have to be identified and eliminated. This list would usually include reduction of bioburden, control of mechanical forces, and maintenance of adequate concentrations of blood-borne growth factors and cellular materials to drive the healing process.

Adequate tissue oxygenation is a critical step in achieving wound closure. The classic description of wound healing involves a 3-stage process in which debridement is followed by inflammation, proliferation, and maturation.¹Inflammation is the initial reaction to debridement in which bleeding occurs, followed by platelet infiltration, degranulation, and release of growth factors. Subsequently, the cells are attracted to the wound site and begin to proliferate to close the wound. Finally, the wound matures as the skin structure remodels to become well aligned and strengthened.

In its simplest form, oxygenation of the wound bed and periwound areas can be a strong reflection of the condition of a wound and its potential for healing. Segmental pressures and the ankle-brachial index (ABI) are simple tests that have been widely used to gauge the level of blood flow to the leg and foot.² Angiography and some ultrasonic devices can also demonstrate the arterial blood flow. However, when it comes to wounds, a system is also needed to allow one to measure oxygenation to the skin at the capillary level.

Radioisotope-labeled blood cells have been used to trace the flow in the soft tissues and are very helpful at identifying areas of inflammation or infection, but they have limited value for assessing the finer details of wound bed perfusion. Scanning devices using indocyanine green (ICG) dye have been used in the operating room to assess blood flow to areas where critical skin ischemia may be present, such as during plastic surgical procedures, but this requires expensive equipment, and the test itself can only be conducted a very limited number of times during a case due to the persistence of the dye.^3,4 

Transcutaneous oxygen pressure (TcPO₂) can be used to assess tissue oxygenation as a measure of perfusion over small regions without the need for dyes. However, this technology is very limited to small spots, typically on the order of 5 mm in diameter or less. Although this can be useful for assessing a specific location, it is difficult to determine the overall blood flow to a larger region of the surrounding skin.⁵

Near infrared spectroscopy (NIRS) is one of the newer options for evaluating oxygen delivery and usage in the microvasculature. This technology uses reflected light to calculate perfusion by taking advantage of subtle color changes that occur in hemoglobin when it is oxygenated. By recording the ratio of oxygenated to deoxygenated hemoglobin, one can accurately measure the percentage of oxygenated blood reaching the skin.⁶ Because the light can be generated by an array of diodes and detected over a larger area, it is possible to capture large regions of anatomy simultaneously, such as the entire plantar surface of the foot. 

The NIRS transmits very specific wavelengths of light between 600 nm and 1000 nm, and it measures the amount of reflected light to determine the ratio of deoxygenated to oxygenated hemoglobin. Unlike systems that utilize visible light, the near infrared spectrum can penetrate deeper into the skin while eliminating other potentially confounding wavelengths. Deeper light penetration is more accurate and reproducible, and it diminishes the effect of skin pigmentation color on the data.

There are some clear advantages to using this type of light-based system for measuring skin perfusion. With NIRS, it is a noncontact system. Unlike with TcPO₂, there are no disposable elements or items that have to be sterilized between patient measurements. The data can be captured and processed in as fast as 5 seconds and is highly reproducible. Images can include much larger surfaces (eg, the entire plantar surface of the foot) and can be taken before and after a treatment is performed in a single clinical or surgical visit. It also can be compared with prior visits with great consistency. Unlike the ICG-based systems, there are no injectable dyes used with NIRS.

Over the last decade, there has been a great deal of speculation about how to interpret images taken using NIRS. The earliest studies found that 40% oxygenated hemoglobin in the wound bed was needed for healing.⁷ Similarly, the 40% number has been useful for predicting the viability of skin flaps and grafts during surgery. Nonetheless, many flaps that have achieved 40% or more oxygenated hemoglobin have failed,⁸ while others have survived, so this number is not a hard and fast cutoff. A 2011 study demonstrated that the postoperative differences in tissue oxygen saturation within a skin flap correlated with the development of skin necrosis.⁹ It is important to consider that wound healing is not wholly dependent on vascular perfusion but rather oxygenation in the context of numerous other biologic influences, including mechanical forces to the wound bed, disease state, general health, infection, and the presence of bioactive materials (eg, collagen and growth factors).¹⁰ 

The purpose of the current study was to review a series of cases in which wounds were observed using NIRS to ascertain patterns of oxygen saturation. By examining patterns within the wound bed and the surrounding soft tissues, the author hoped better identify signs that might help clinicians to understand when a wound might reasonably be expected to close. It was the author’s hypothesis that there was a pattern of changes observed within the wound and the periwound regions that could be useful in predicting when wounds will progress toward healing. In this series, data from 25 wounds were observed to determine if a pattern emerged that might help one gauge where a wound is functioning in the collective healing process. Data were collected from wounds that closed and did not close. The vast majority of the wounds were diabetic foot ulcers (DFUs) and venous leg ulcers (VLUs), but a portion of wounds associated with other etiologies was also examined. 

This was a retrospective study of 25 wounds observed in the author's clinic in a metropolitan community hospital. Inclusion and exclusion criteria can be found in Table 1. Most patients were seen for DFUs and VLUs. Wound management varied, and it included numerous biologically active materials, negative pressure wound therapy, skin flaps, autografts, offloading, compression therapy, and management of infections. Several patients also underwent vascular procedures due to ischemia. Details regarding the etiology and treatment for these patients are reported in Table 2.

The NIRS images were collected using Snapshot_NIR (Kent Imaging Inc). Several images were captured at each visit, typically before and after debridement, and then again after application of any biologically active materials. The NIRS device used is a rechargeable handheld camera that captures both NIRS and standard visible light clinical images (Figure 1). By superimposing the images, one can appreciate the oxygen saturation levels within and around the wound bed across the entire field of view. 

Once the image has been captured, individual locations within the image can be selected to determine the percentage of oxygenated hemoglobin over an area as small as 5 mm x 5 mm. Calibration is performed each time the device is turned on, and images are always collected from a fixed distance using a laser focusing system built into the camera. An algorithm is utilized to compensate for skin pigmentation, depending on skin tone. Typically, patients with a score of Fitzpatrick 3 or higher in skin color will require melanin compensation that is built into the device.¹¹ Wound measurements can be recorded precisely using the associated device’s software package.

In each case, images were captured following calibration and then assessed for skin perfusion patterns. Melanin correction software was utilized in patients with Fitzpatrick 3 or higher skin tones. Oxygenation within the wound bed area and in the surrounding tissues was examined to determine if any trends existed that may coincide with wounds that either healed or did not progress toward healing. Both oxygenated and deoxygenated hemoglobin images could be viewed to see if there was any variation that could provide further insights into the healing process.

In this study, a variety of wounds, ranging from 1 cm² to over 30 cm², were observed. Over the 12-week observation period, about 60% of the wounds closed. Because the author was interested in the oxygenation and perfusion patterns surrounding the wounds, this closure rate was not a significant portion of the study, but it is fairly typical for most wound centers over a 12-week period,⁴ indicating that reasonable technologies were being used to achieve closure. Treatments such as cryopreserved human skin allografts, amniotic membrane, and decellularized collagen were utilized on patients in this study in order to attempt wound closure. Some of these treatments resulted in direct changes to the tissue oxygenation percentage. Some patients also had vascular intervention, including angioplasty (n = 2) and distal bypass surgery (n = 1). All patients were treated with regular debridement and wound cleaning. All of the wounds were treated with solutions designed to disrupt biofilms, such as hypochlorous acid at some point during their care as well.

The author’s observations focused on the tissue oxygenation percentage (S_tO₂) of the wound bed itself as well as the surrounding areas. In addition, the wounds were examined for patterns that may have given some early insights as to which wounds were progressing toward closure. Images showing the ratio of oxygenated to deoxygenated hemoglobin, along with the separate images showing only oxygenated or deoxygenated hemoglobin, were examined.

Based on these observations, the following patterns were noted.

• Wounds with an S_tO₂ level below 40% rarely closed if that was the average across the wound bed. However, many wounds that closed had isolated areas within the wound bed, where the S_tO₂ fell well below 30%, provided that the deprived area was no more than 50% of the total wound bed area (Figure 2).

Wounds surrounded by intense and uniform red areas (indicating high S_tO₂), with no gradation to the red areas, were generally observed to be clinically inflamed. These wounds did not show rapid or significant changes in wound size, despite the fact that the S_tO₂ may have hovered in the 65% to 75% range, both within the wound bed and in the periwound areas. In the case shown in Figure 3, the wound remained essentially unchanged for 9 months despite attempts to achieve closure with various biologics and negative pressure wound therapy. The NIRS analysis appeared to show excellent S_tO₂ across the wound bed, with an intense red ring surrounding the wound bed (>90% S_tO₂); the wound bed showed a lower level of oxygenation but still well above 65% across the entire surface. However, when considering the deoxygenated hemoglobin image (Figure 3C), it is apparent that the wound had a much higher concentration of deoxygenated hemoglobin than the surrounding tissues. This indicated that the oxygenated blood was reaching the wound margin, but not making it into the wound bed itself. The intense red ring surrounding the wound bed was indicative of arterial congestion leading from the margin of the wound to the wound bed itself. This may have been the result of basement membrane thickening and/or reduced capillary flow. Clinically, the wound bed had a dense, scar-like texture. The author hypothesized that this wound would probably have not benefited from vascular interventions, such as bypass, since there appeared to be good macroscopic flow but blockage at the capillary level.

• Most of the wounds that were progressing toward healing showed a wispy and gradual decrease in S_tO₂ as the distance away from the wound increased. This author described this as having an appearance resembling a ray of sunlight, with the central red region moving to a combination of yellow and red as the distance away from the wound increased (Figure 4). This was probably the most predictive sign that the wound was starting to heal.
• Necrotic wounds frequently had an S_tO₂ in the 25% or less range across the wound bed. Often the adjacent periwound areas varied tremendously in their S_tO₂ measures. Some of these wounds went on to gradually slough off the superficial necrotic areas, exposing darker yellow necrotic fatty tissue beneath.
• Once necrotic tissue was removed, if diffuse reddish areas of the wound bed were observed with an S_tO₂ above 60%, these wounds tended to gradually reduce in size and go on to full wound closure.
• In other cases, necrotic wounds showing almost zero S_tO₂ may have been bound by areas with reduced S_tO₂. In these cases, the wounds tended to increase in area and the necrosis spread outward clinically (Figure 5). This was thought to be indicative of the body unsuccessfully attempting to shunt blood to the areas of necrosis and/or areas of diminished S_tO₂. In these cases, the author hypothesized that the necrosis caused damage to the capillary beds, preventing ingress and egress of blood. If one considers the oxygenated and deoxygenated images, it can be seen that there was minimal blood flow reaching the wound. This differs from arterial congestion, where the blood is making it to the wound bed, but staying there so long that it becomes deoxygenated.
• Limbs showing a general decrease in S_tO₂ across large portions of the foot or leg, ranging from 65% or less, were generally associated with limb ischemia. This was particularly true for cases in which arterial pulses could not be detected and Doppler signals were monophasic or, in some cases, weakly biphasic. In those cases, patients who had vascular intervention showed significant improvement in overall S_tO₂, to greater than or equal to 70%, and went on to heal ischemic ulcers.

Classification
The goal of the current study was to establish a foundation to help identify and clarify the patterns of tissue oxygenation observed in order to use them to predict when wounds may or may have not been healing. The proposed classification system is based on observations made with the previously mentioned NIRS device (Table 3).

Near infrared spectroscopy is an imaging technology in its infancy. Although it is well supported that the S_tO₂ measured from location to location is accurate,⁵ the significance of this in the context of wound healing has been unclear. Other imaging technologies, such as radiographs for bone, have been around for many years, but the patterns that are used by physicians to understand different types of pathology are continually being developed. For example, an inexperienced doctor may have a difficult time differentiating between degenerative arthritis, infection, and trauma, while an expert would easily note the subtle differences in cortex continuity to make the distinction. Similarly, bony deformities and bone tumors are easily distinguished with experience. These observations are frequently qualitative rather than quantitative.

In this study, the author attempted to identify qualitative patterns that coincide with the physiology of wound healing. At its simplest level, there is a certain level of S_tO₂ needed for wounds to heal.¹¹ Although many clinicians frequently consider 40% to be the minimum wound bed S_tO₂ to support healing, the observations reported herein demonstrate that it is more than just the average across the wound bed that determines which wounds will close and which will not. Upon further examination, there does appear to be a pattern emerging that is based on the oxygenation of the wound bed, as well as the surrounding soft tissues. Measurements in the middle range from 40% to 60% oxygenated hemoglobin are unclear; outside of this range, there are patterns described in Table 3. 

In this study, several patterns within the wound bed and in the surrounding areas were described, which coincide with the author’s theory on the progression of wound healing. This theory, which the author has called HINT, is based on this hypothesized natural progression.

H: Hyperperfusion. In this stage, S_tO₂ buildup is seen around the wound bed. There should be a gradual darkening as the oxygenated blood moves toward the wound bed. The NIRS will show a rim of bright red surrounding the wound bed.

I: Imbibition. This is the start of the exchange between blood surrounding the wound site and blood within the wound bed. It is most pronounced when a graft or biologic material is introduced to the wound surface, and the graft and wound bed begin to exchange fluids. In this stage, improved blood flow within the wound bed would be expected to see, as demonstrated by improved oxygenation in this area.

N: Neovascularization. As the interface between the wound bed and surrounding tissues becomes better perfused, the flow toward the wound bed from the far periphery begins to thin and the distinction between the wound bed and surrounding tissues becomes less defined. This creates a uniform field of oxygenation in the periwound areas that becomes less pronounced as the distance from the wound increases.

T: Trailing. This is the most obvious stage to observe the transition to a wound that will go on to heal. At this stage, the surrounding tissues appear to be normalizing, leaving behind wispy-appearing streaks and peaks surrounding the wound. These are expected to thin to virtually no bright red spots around 3 cm away from the wound in all directions.

This entire pattern is analogous to a wave moving towards and then away from the shore. Initially, it brings with it a flood of nutrients, fish, and shells. The wave makes contact with the sand, transferring these materials to the shore. Eventually, the wave dissipates and retracts toward the ocean. 

In the case of wounds, the perfusion of the surrounding tissues leads to an influx of cytokines and growth factors between the wound bed and the surrounding tissues. This early exchange (imbibition), leads to a further exchange between the wound bed (and any biologics or grafts that may be present) and the surrounding tissues to stimulate the neovascularization phase. Eventually, the influx of blood becomes more evenly dispersed as the wound starts to progress through the healing stages. Hyperperfusion recedes and more uniform, normalized blood flow to the surrounding tissues is restored.

With NIRS, it is this appearance of increased oxygenation followed by a wispy retraction of flow that seems to signal the wound is rapidly moving toward closure. The wound that remains erythematous, and is perpetually surrounded by hyperoxygenated tissue, is more characteristic of a nonprogressive wound that is still communicating with the surrounding tissues but not necessarily moving toward closure. And, finally, the ischemic wound, which demonstrates large areas with less than 40% oxygenated hemoglobin, is lacking sufficient blood flow to heal in many cases.

Like radiographs or an ultrasound, NIRS is a tool that can be used to understand and even predict pathologies particularly related to the condition of the skin and wound healing. It also provides a combination of qualitative and quantitative data that is subject to interpretation by clinicians. With time, these patterns will become more recognizable to physicians who utilize them regularly. The current study is the author’s attempt to understand these relationships.

There are many shortcomings to this study, including the relatively small sample size. Furthermore, the author’s interpretation of the patterns observed on the skin and wound surface corresponding to microscopic vascular activity are a hypothesis based on the author’s knowledge of wound healing. Definitive microscopic studies would offer the support needed to fully understand the patterns observed.

In this study, the potential for using NIRS to predict which wounds may heal was explored. The method used was a retrospective chart review in which 25 patients with either a DFU or VLU were followed from baseline for 12 weeks. The NIRS data were reviewed at baseline and at subsequent visits to determine if patterns exist that suggested the fate of the wound. Based on this sample of observations, 4 basic patterns were identified. In addition, a mechanism has been proposed (HINT) that suggests what might be taking place as a wound progresses through the healing process.

The process of pattern identification is only in its infancy. Similar to physicians who utilize ultrasound technology to examine soft tissues, the author believes that it will take time and additional studies to truly understand what NIRS images are suggesting. These patterns should be studied with larger pools of data and may even include the use of artificial intelligence to select the specifics of the various patterns examined here. As more physicians adopt this technology, more patterns will become evident and predictive value of this technology will improve. 

Author: Adam Landsman, DPM, PhD, FACFAS

Affiliation: Harvard Medical School, and Department of Orthopaedic Surgery, Massachusetts General Hospital, Boston, MA

Correspondence: Adam Landsman, DPM, PhD, FACFAS, Department of Orthopaedic Surgery, Massachusetts General Hospital, 52 Second Avenue, Suite 1150, Waltham, MA 02451; alandsman@mgh.harvard.edu

Disclosure: The author discloses no financial or other conflicts of interest.