Establishment of Normative Perfusion Values Using Hyperspectral Tissue Oxygenation Mapping Technology

ABSTRACT

Hyperspectral tissue oxygenation mapping (HTOM) is a new imaging modality that is able to evaluate tissue perfusion at the microcirculatory level by measuring oxyhemoglobin, deoxyhemoglobin, and oxygen saturation levels. This study establishes a database of normative values for HTOM to guide further utilization of this diagnostic modality. Methods. HTOM technology quantifies tissue oxygenation with quantitative and spatial analysis. To establish normative values, HTOM was used to scan 11 anatomical regions on 194 subjects without symptomatic vascular disease. Oxyhemoglobin, deoxyhemoglobin, and oxygen saturation values were obtained in all regions for all subjects. To evaluate reproducibility, a subgroup (n=74) returned 8 hours later for scans of the same anatomic sites. Another subgroup (n=19) underwent cuff ischemia testing to assess the performance of HTOM in a simulation of microvasculature ischemia. Results. One hundred and ninety-four subjects (93 male, 101 female, age range 18–80) completed anatomic site assessments. Normative values were obtained for anatomic regions on the lower and upper extremity. The values for oxyhemoglobin, deoxyhemoglobin, and oxygen saturation varied with anatomic location with plantar and palmar surfaces demonstrating highest baseline perfusion. No significant differences in HTOM values were noted at the 8-hour repeat evaluation. A decrease in perfusion occurred during cuff induced ischemia followed by reperfusion and a return to baseline. Conclusion. A database of normative ranges for oxyhemoglobin, deoxyhemoglobin, and oxygen saturation has been obtained to serve as reference values for future HTOM analysis. These data will provide a comparative construct to study hyperspectral imaging and its role in the diagnosis and treatment of vascular disease.

Introduction

Non-invasive vascular laboratory testing is important in detecting the severity of chronic limb ischemia (CLI). However, definitive methods are lacking to discern the functional state of perfusion at the microcirculatory level. Oxygen delivery, extraction, and saturation are key parameters to be considered when evaluating the state of the microvasculature in CLI, especially in regard to diabetes mellitus and wound healing. Hyperspectral imaging (HTOM) has already been used in several clinical scenarios, including discovering early changes in the microcirculation of the diabetic foot,¹ predicting clinical outcomes in diabetic foot ulcers,² maximizing limb preservation in amputation planning,³ assessing of shock,^4,5 and identifying residual tumor tissue during breast cancer surgery.^6,7 However, in order to take full clinical advantage of possible CLI applications, a normative database of values is needed to establish reference values of tissue perfusion, providing a comparative analysis for diseased states, thereby improving the technology’s clinical applicability. This study addresses the need to provide clinicians with a profile of normative value ranges for oxyhemoglobin and deoxyhemoglobin as measured by hyperspectral imaging. The study also examines anatomic variations and reproducibility in these normative values for oxyhemoglobin, deoxyhemoglobin, and oxygen saturation.

Methods

The study enrolled 194 subjects with no symptoms of peripheral arterial or venous disease. Patients having diabetes mellitus, hypertension, or any known peripheral vascular disease were excluded from participation. The study was conducted at a single center with institutional IRB approval and written informed consent. Adverse events were monitored and reported. There were 93 males and 101 females studied. Subjects ranged in age from 18 to 80 years of age, with a mean of 42 ± 14. Racial distribution included 64 Caucasians (33.2%), 51 African Americans (26.4%), 43 Asians (22.3%), and 35 Hispanics (18.1%). Subjects included 125 non-smokers, 42 current smokers (21.8%), and 27 with a past history of smoking. Tissue oxygenation images were collected with a commercial hyperspectral imaging system (OxyVu, HyperMed, Inc., Burlington, Massachusetts). This hyperspectral imaging system obtains multiple images at discrete wavelengths, providing a diffuse reflectance spectrum for each pixel in the hyperspectral image. The system uses wavelengths between 500 and 660 nm to correspond to a region that includes two absorption peaks from oxyhemoglobin and one absorption peak from deoxyhemoglobin. Tissue oxygenation images or maps were constructed from oxyhemoglobin, and deoxyhemoglobin values determined from each pixel in the image. Prior to imaging each patient, the system was calibrated to a reflectance card (OxyVu CheckPad, HyperMed). Patients were imaged on a standard examination table or reclining chair. A fiducial target (OxyVu Target, HyperMed) was placed near the center on the image field of view to facilitate image realignment (image registration) and to correct for patient movement when collecting multiple images. Spectral decomposition was used to extract relative values of tissue oxyhemoglobin (Oxy) and deoxyhemoglobin (Deoxy) from the diffuse reflectance spectra by comparing to standard transmission spectra for oxyhemoglobin and deoxyhemoglobin solutions. Oxy and Deoxy units represent relative concentrations of Oxy and Deoxy found in the measured tissue volume (approximately the effective pixel size at the object, 0.1 mm x 0.1 mm, multiplied by the penetration depth of light into tissue, about 1 to 2 mm in this wavelength range). Tissue oxygen saturation (StO2), the fraction of oxygenated hemoglobin in superficial blood vessels, was calculated as the ratio of Oxy over the sum of Oxy and Deoxy. The study included three groups: a single measurement group (SM), repeatability group (RG), and cuff ischemia group (CI). All enrolled subjects were included in the SM group; a subset of 74 subjects was in the RG group; and 19 subjects between the ages of 18 and 39 were included in the CI group. For the SM and RG groups, hyperspectral scans were collected from 11 anatomical sites: palm, back of hand, anterior forearm, posterior forearm, lateral thigh, lateral leg above and below the knee, dorsal metatarsal, plantar metatarsal, heel and plantar arch. The dorsal metatarsal site was measured in the CI group. Subjects were instructed to rest in a seated position with legs horizontal to the floor for 10 minutes. Targets were then placed on each of the 11 hyperspectral scan locations and scans were performed at each site using the HTOM system (HyperMed). Data from scans were stored, and subjects in the SM group exited the study. Subjects in the RG group completed the initial set of 11 scans and then returned 8 hours following their initial scans to repeat all 11 scans at the same target locations. To ensure the same placement of targets and location of scans, these subjects were marked with a small “x” using a surgical marker at the scan sites. After the repeat scans, these subjects exited the study. Subjects in the CI group completed the initial set of 11 scans. This was followed by a baseline blood pressure measurement. The blood pressure cuff was then applied at thigh level, well proximal to the dorsal metatarsal target site. Baseline scans were acquired once per minute for 5 minutes at the dorsal metatarsal site with no cuff inflation. The cuff was then inflated to 200 mmHg, or 50 mmHg above systolic blood pressure, the lower of the two values being utilized. A registered nurse increased and monitored cuff pressure. Scans were acquired immediately after pressure increase and once per minute for 5 minutes following cuff inflation. Cuff pressure was then released with an initial scan and repeated scans once every minute for 5 minutes were obtained during reperfusion and recovery. There was no overlap in subject participation between the CI subgroup and RG groups. Multivariate analyses were carried out using analysis of variance (ANOVA) and standard extensions of ANOVA, such as stepwise linear regression models. In these analyses, the continuous outcomes were biometric indices: Oxy, Deoxy, and Sat. The independent variables of primary interest were anatomical site, gender, and smoking status. Categorical variables were anatomical site, gender, and smoking status. Analyses that report on pair-wise differences used the standard t-tests obtained after an ANOVA that set the standard error equal to the root of the mean squared error. Significance was reported when p Results Tissue Oxy, Deoxy, and oxygen saturation values were categorized by anatomical site and stratified by gender. Data are shown in units that correlate to the concentration of blood in the tissue (umol/L).⁸ Values were obtained for the lower extremity and upper extremity. Oxyhemoglobin and oxygen saturation were greatest on the plantar surface of the forefoot and hindfoot of the lower extremiy and the palmar surface of the upper extremity (Figure 2). Males consistently demonstrated higher oxyhemoglobin and oxygen saturation values at each anatomic region, although not to a degree that reached statistical significance. Females had higher Deoxy values that also did not reach statistical significance. Tobacco use did not have a statistically significant effect on Oxy, Deoxy, or oxygen saturation at any anatomic location. The spectrum of normative values for each measurement site was reconstructed as a normal distribution curve. Differences in tissue oxygenation values by anatomical site in the RG (reproducibility) subgroup are shown in Table 3. There was reproducibility of data without significant differences in those values obtained 8 hours after initial measurement. The reproducibility data revealed no significant differences in same-site, same-subject hyperspectral Oxy, Deoxy, and oxygen saturation measurements separated by 8 hours. Hyperspectral imaging in the CI group revealed a decrease in Oxy and increase in Deoxy after induction of the ischemic period. After release of the ischemic cuff, there was a period of rebound perfusion with a significant increase in Oxy. Deoxy levels returned to near baseline levels with reperfusion. Oxygen saturation levels followed the Oxy pattern of ischemia and reperfusion. There was then a subsequent return to baseline for all values. No adverse events were reported throughout the duration of the study.

Discussion

Hyperspectral imaging technology provides a contact-free, near real-time method of quantifying tissue oxygenation. Medical hyperspectral imaging technology provides quantitative and spatial data on tissue oxygenation indicative of dermis perfusion. Alternative modalities such as transcutaneous oxygen monitoring measure oxygenation, but do not provide the same degree of anatomic localization and lack the capacity for collection of spatial information. Hyperspectral imaging gathers spectral and spatial information by systematically varying the wavelength of light emitted to collect images at multiple wavelengths between 450 and 650 nm. Specific chromophores in the blood, Oxy and Deoxy are identifiable by their signature patterns of light absorption at specific wavelengths. Based on the spectrum of reflected light acquired for each pixel, hyperspectral imaging technology generates a spatial gradient map that indicates local oxygen delivery, extraction, and saturation in the region of interest.⁹ This type of spatial distribution map is unique to this technology and crucial to revealing how the composition of Oxy and Deoxy varies anatomically. Early clinical observations using hyperspectral imaging have demonstrated possible clinical utility. Experience has been gained with hyperspectral imaging in the management of diabetic foot ulceration.² Changes in tissue oxygenation levels in the area surrounding foot ulcers observed with hyperspectral imaging predicted which ulcers had better potential for healing. Ulcer healing was predicted with high sensitivity (93%) and specificity (86%). Other diabetes research has used hyperspectral imaging to identify those patients at risk for foot ulceration.¹⁰ Determination of amputation level is another area of possible clinical utility. When evaluating ischemic lower limbs for amputation, hyperspectral imaging can assess tissue oxygenation at specific anatomic locations, such as the planned amputation site in order to determine healing probability and maximize limb preservation. Case reports have described more conservative amputations than initially planned through the use of hyperspectral imaging tissue oxyhemoglobin values; in one case, at the transmetatarsal joint instead of the ankle, and the other a Syme’s amputation rather than a below-knee amputation.³ The findings of this study establish a normative range of tissue oxygenation values for a diverse demographic population. The values reflect dependence of the observed Oxy, Deoxy, and oxygen saturation normative values on anatomic site and gender. The plantar surface of the foot demonstrated the highest levels of Oxy and oxygen saturation. This finding coincides with the vascular angiosomes of the foot in which the plantar surface involves the largest number of angiosomes contributing to its perfusion.¹¹ Functional utility would also support the finding of highest oxygenation at the plantar surface or the foot, the surface that supports our upright posture, and the palmar surface of the hand. The normative values were reproducible as demonstrated by the repeat group at a different time of day. The subgroup analysis of cuff-induced ischemia serves as validation that the hyperspectral imaging technology provides readings that accurately reflect the oxygen perfusion, extraction, and saturation occurring in the skin microvasculature in near-real time. Determination of normal ranges characterizes the general population in terms of hyperspectral imaging, so that clinicians can identify and interpret abnormalities when hyperspectral imaging results lie outside these established normative ranges. This allows for the development of a rubric of normative ranges that clinicians can apply to accurately analyze individual cases. Additional research has focused on patterns in hyperspectral imaging values. A study evaluating non-diabetic, diabetic, and diabetic neuropathic patients found a differential ability to distinguish these 3 groups of subjects, depending on the hyperspectral imaging target sites assessed.¹ Forearm oxygen saturation values revealed a difference between each group. Patients with both diabetes and neuropathy had the greatest reductions in oxygen saturation, followed by diabetics, with the non-diabetic patients having the highest oxygen saturation levels. Oxygen saturation measurements of the foot revealed higher oxygen saturation levels in both non-diabetic and non-neuropathic groups, but a discernibly lower amount of oxygen saturation in the neuropathic diabetics. Another study has presented further evidence that hyperspectral imaging values reveal differences between diabetic and normal subjects.² The researchers found statistically significant differences in hyperspectral tissue oxygenation values in the upper extremity for patients with diabetic foot ulcers, as compared to those for non-diabetic subjects. The study also indicated that hyperspectral measurements can reveal alterations in oxygenation that correlate with the severity of disease. The database of normative hyperspectral values generated by the current study, as well as the revealed variability of these normal values, will allow clinicians greater ability to interpret clinical implications of the oxygenation values observed in an individual patient. A low Oxy level, elevated Deoxy, and reduced oxygen saturation relative to the normative values would be interpreted as an ischemic site.

Conclusion

The clinical utility of data generated from HTOM will be enhanced by creating a database of normal ranges of Oxy, Deoxy, and oxygen saturation values. Knowing the values and their variations will assist clinicians in accurately interpreting tissue oxygenation values. This will allow non-invasive evaluation of disease progression, prognosis for healing, effectiveness of endovascular revascularization, and determination amputation level. With increasingly frequent clinical use of the technology and as additional studies are performed, data will be gathered on an ever- broadening demographic spectrum of patients. The compilation of such additional data will contribute to further refinement of the established normative ranges and better define the clinical scenarios that will most benefit from this promising technology.

References

1. Greenman RL, Panasyuk S, Wang X, et al. Early changes in the skin microcirculation and muscle metabolism of the diabetic foot. Lancet 2005;366:1711–1717.

2. Khaodhiar L, Dinh T, Schomacker KT, et al. The use of medical hyperspectral technology to evaluate microcirculatory changes in diabetic foot ulcers and to predict clinical outcomes. Diabetes Care 2007;4:903–910.

3. Frykberg R, Tierney E, Tallis A. Applying hyperspectral imaging technology to limb preservation (Abstr). Presented at the New Cardiovascular Horizons Meeting, September 2008, New Orleans, Louisiana.

4. Gillies R, Freeman JE, Cancio LC, et al. Systemic effects of shock and resuscitation monitored by visible hyperspectral imaging. Diabetes Technol Ther 2003;5:847–855.

5. Cancio LC, Batchinski AI, Mansfield JR, et al. Hyperspectral imaging: A new approach to the diagnosis of hemorrhagic shock. J Trauma 2006;60:1087–1095.

6. Panasyuk SV, Yang S, Fallar DV, et al. Medical hyperspectral imaging to facilitate residual tumor identification during surgery. Canc Biol Ther 2007;6:439–446.

7. Freeman JE, Yang S, Panasyuk SV, et al. In situ evaluation of residual breast tumor and tumor grade using medical hyperspectral imaging (MHSI). Abstract #10677, American Society of Clinical Oncology Annual Meeting, 2006 and J Clin Oncol 24(18S, Part 1) June 2006: 582S.

8. Blood concentration data. On file, HyperMed, Burlington, Massachusetts.

9. Colarusso P, Kidder LH, Levin IW, et al. Infrared spectroscopic imaging: From planetary to cellular systems. Appl Spectrosc 1998;52:106A–120A.

10. Dinh T, Panasyuk SV, Tracey BH, et al. The use of medical hyperspectral imaging (MHSI) to identify patients at risk for developing diabetic foot ulcers. Poster #1106-P, American Diabetes Association 65th Annual Sessons, June 2005, and Diabetes, June 2005, Vol 54, Supp 1: A270.

11. Neville RF, Attinger C, Bulan E, et al. Revascularization of a specific angiosome for limb salvage: Does the target artery matter? Ann Vasc Surg 2009;23:367–373.

From *Georgetown University Hospital, Washington, D.C. and §State University of New York, Syracuse, New York.

Manuscript submitted June 24, 2009 and accepted September 8, 2009.

Correspondence: Richard Neville, MD, Georgetown University Hospital, Division of Vascular Surgery, 3800 Reservoir Rd, 4PHC, Washington, DC 20007. E-mail: neviller@gunet.georgetown.edu

Disclosure: The authors report no conflicts of interest regarding the content herein.