Hyperspectral Imaging and Ankle: Brachial Indices in Peripheral Arterial Disease
Article Outline
Background
To evaluate the correlation between ankle:brachial indices (ABI) and visible light reflectance spectroscopy hyperspectral imaging (HSI) determinations of oxygenated and deoxygenated hemoglobin (oxyHgb and deoxyHgb) levels in the skin of the distal lower extremity. This is a prospective, open, comparator trial which took place at the Vascular laboratory of a Veterans Administration Hospital in Long Beach, USA. Fifty-eight patients (85 limbs) were referred for routine vascular laboratory studies including ABI had concomitant HSI. Limbs with noncompressible pedal signals were excluded from the analysis.
Methods
ABI was determined with continuous wave Doppler ultrasound and leg blood pressure cuffs. A commercial HSI system (Oxu-Vu®, Hypermed, Inc.) was used to measure oxyHgb, deoxyHgb, and percent oxygenated hemoglobin (%oxyHgb) in the dorsum of the foot and ankle. HSI measurements of volar forearm skin were also obtained to normalize the lower extremity HSI measurements in a manner comparable with ABI.
Results
For purposes of comparison, data sets were divided into 3 groups: ABI > 0.9 (n = 53), 0.45 < ABI < 0.9 (n = 22), and ABI < 0.45 (n = 10). There were no significant differences between oxgyHgb, %oxyHgb, normalized oxyHgb, and normalized %oxyHgb when means values for these parameters were compared between the three groups based on unpaired t test statistics.
Conclusion
The lack of a correlation between HSI measurements and ABI is consistent with previous observations that in moderate peripheral arterial disease skin perfusion is maintained at normal levels and in critical ischemia paradoxical increased flow may be present. Although the current study failed to show a clinically useful correlation between HSI measurements of oxyHgb levels, further evaluation of this novel technology is warranted.
Introduction
Hyperspectral imaging (HSI) is a method for determining oxygenated and deoxygenated hemoglobin (deoxyHgb) levels in the superficial microcirculation of the skin using differential reflectance spectroscopy of light.1 A currently marketed imaging system (Oxy-Vu®, Hypermed, http://www.hypermed-inc.com) uses HSI to generate false color images of 10 × 10 cm areas of skin with spatial resolution of surface tissue oxygenated hemoglobin (oxyHgb), deoxyHgb, and percent oxygenated hemoglobin (%oxyHgb); this last parameter is comparable with hemoglobin oxygen saturation. Previous studies with the Oxy-Vu® and other HSI techniques using visible and near infrared spectroscopy have shown measurements to correlate with fluid resuscitation in hemorrhagic shock,2, 3 differences in the microcirculation of the foot between diabetics and normals,1, 4 depth of injury in burns,5 and tissue viability in plastic surgery.6, 7
The goal of the current study was to assess whether HSI might be useful for detecting arterial insufficiency in patients with peripheral arterial disease (PAD) by comparing measurements yielded with the Oxy-Vu® system with standard ankle:brachial indices (ABI) in subjects referred to a clinical vascular laboratory.
Materials and Methods
Patients referred to the vascular laboratory at the Veterans Administration Health Care System between March 2008 and September 2008 for evaluation of peripheral arterial disease underwent standard measurement of ankle:brachial pressure indices using blood pressure cuffs and continuous wave Doppler ultrasound (Model 1050-C, Parks Medical Electronics, Aloha, OR) as part of an Institutional Review Board-approved protocol. The higher systolic pedal pressure (posterior tibial or dorsalis pedis) in the lower extremity was divided by the ipsilateral brachial systolic pressure to determine ABI for that limb. Limbs with noncompressible ankle signals were excluded from the analysis. HSI was performed with ambient temperatures between 23 °C and 25 °C. Upper extremity images were obtained of the volar surface of the forearm. Three images were obtained for each lower extremity: a proximal image of the anterior leg with the distal border approximately at the transmalleolar level (D3), an image of the anterior ankle and dorsum of the proximal foot (D2), and a distal foot image including the dorsal aspect of the toes (D1). An example of the three lower extremity and one upper extremity images is depicted in Figure 1. Measurements of oxygenated and deoxyHgb levels were based on the average determined for the entire field of the skin imaged using the software provided with the Oxy-Vu® system. A total of 58 subjects and 85 limbs had both sets of measurements and were amenable to analysis after excluding limbs with noncompressible tibial vessels.
Fig. 1
Composite figure showing false color images generated by HSI of the dorsum of the lower extremity and the volar aspect of the forearm. The scale on the right of the image denotes the relative proportions of oxygenated and deoxygenated hemoglobin (deoxyHgb). Violet and red correspond to high proportions of oxygenated hemoglobin (oxyHgb); yellow and green to high proportions of deoxyHgb.
An analysis of the relation between systolic pressures and HSI measurements was performed using both linear correlation analysis and t test statistics. Two measurements of skin perfusion based on HSI were chosen for comparison with ABI: oxyHgb and the proportion of oxygenated hemoglobin (%oxyHgb = 100 x oxyHgb/oxyHgb + deoxyHgb).
In some analyses, the average value of the D1, D2, and D3 images were compared with ABI, whereas in others just D2 measurements were compared. In other analyses, HSI values were normalized by dividing the values for lower extremity measurements by the ipsilateral forearm measurement of the same parameter. For example, the average oxyHgb measured in the combined D1, D2, and D3 images was divided by average ipsilateral forearm oxyHgb to yield a ankle:brachial oxyHgb ratio to compare with ABI for the same limb. Similarly, the ratios of lower to upper extremity %oxyHgb measurements were calculated.
Statistical analysis of data was performed with a microcomputer program: PRISM-2.01 (GraphPad InStat, La Jolla, CA).
Results
Fig. 2, Fig. 3, Fig. 4, Fig. 5 depict the results when average HSI measurements for the combined three lower extremity skin fields were used for analysis. Each HSI parameter is the mean of the hemoglobin values derived from averaging values over the entire field of each separate image. Figure 2 depicts the oxyHgb for limbs with ABI greater than or equal to 0.9 (ABI ≥0.9, n = 53), between 0.45 and 0.9 (ABI 0.45–0.9, n= 22), or less than or equal to 0.45 (ABI ≤0.45, n= 10). The mean oxyHgb levels for the ABI ≥0.9 group were 41.1 ± 2.4, for ABI 0.45–0.9 group 38.5 ± 3.9, and for ABI ≤0.45 group 44.3 ± 6.8. Comparison of the mean oxyHgb levels between groups showed no significant differences (unpaired t test, p ≥ 0.56). Figure 3 depicts a similar analysis using %oxyHgb values correlated with the same ABI ranges. Mean %oxyHgb values were 42.1 ± 1.2%, 39.2 ± 2.0%, and 43.2 ± 2.7% for ABI ≥0.9 group, ABI 0.45–0.9, and ABI ≤0.45, respectively. Again, comparison of mean %oxyHgb levels between groups showed no significant differences (p ≥ 0.21). Figure 4 shows the ratio of lower extremity oxyHgb to forearm oxyHgb (lower extremity oxyHgb/upper extremity oxyHgb) similarly segregated into ABI groups. Again, mean values for each ABI range were similar (1.0 ± 0.1, 0.9 ± 0.1, and 1.1 ± 0.2, respectively), and differences between groups were not significant (p ≥ 0.40). Figure 5 shows the calculated ratio of %oxyHgb between lower and upper extremities (lower extremity %oxyHgb/upper extremity %oxyHgb). Again, mean values were similar (1.0 ± 0.0, 0.9 ± 0.1, and 1.0 ± 0.1) and differences between groups were not significant (p ≥ 0.19).
Fig. 2
HSI measurements of oxyHgb levels in single limbs based on mean values for combined lower extremity images (D1 + D2 + D3) plotted against corresponding ABI for that limb.
Fig. 3
HSI measurement of percent oxygenated hemoglobin (%oxyHgb = 100 x oxyHgb/[oxyHgb + deoxyHgb]) based on combined lower extremity images plotted against corresponding ABI for that limb.
Fig. 4
HSI measurements of oxyHgb normalized by corresponding ipsilateral volar forearm measurement plotted against ABI. OxyHgb levels determined as in Figure 2.
Fig. 5
HSI measurements of %oxyHgb based on combined lower extremity images normalized by corresponding ipsilateral volar forearm measurement plotted against ABI. %OxyHgb determined as in Figure 3.
Similar analyses of the data comparing his-derived hemoglobin measurements from only the proximal dorsum of the foot (D2 field only) yielded the same results; no significant differences between ABI groups were seen for oxyHgb, %oxyHgb, ratio of lower extremity to forearm oxyHgb, or ratio of lower extremity to upper extremity %oxyHgb (data not shown). Similarly, when linear correlation analysis was performed between HSI measurements and ABI, the correlations were not significant (data not shown). We also performed these same analyses comparing limbs with ABI < 0.9 to those with ABI ≥ 0.9. No statistically significant differences between groups were observed (data not shown).
Discussion
Our data show no significant correlation between the degree of arterial insufficiency measured with ABI methods and hyperspectral camera determinations of oxyHgb levels in the foot. Whether images of the entire foot or just the mid-dorsal area were used did not influence this finding. Similarly, substituting %oxyHgb for oxyHgb or normalizing the lower extremity values by comparison with forearm skin measurements (in a manner analogous to ABI determinations) did not lead to significant correlations between hyperspectral image data and ABI. As we excluded limbs with noncompressible tibial vessels, we doubt that failure to exclude limbs with unreliable ultrasound measurements or other methodological errors in determining ABI underly these results. Although we anticipated when embarking on this study that a correlation between ABI and hyperspectral parameters would be observed, this clearly was not the case.
Several explanations for our findings can be put forward. First, it is well appreciated that the lower extremities of patients with moderate degrees of peripheral arterial disease do not routinely differ in gross appearance from those of normal subjects. In our study, the preponderance of diseased limbs had ABIs between 0.45 and 0.9, and only 10 limbs had lower than 0.45 values. It is possible that if a greater number of more severely affected limbs had been assessed with HSI, more of a correspondence would have been observed. This explanation, however, would not materially alter our observation that with mild to moderate arterial disease hyperspectral skin hemoglobin measurements do not obviously correlate with ABI.
Regulation of the skin microcirculation and the influence of arterial disease on skin perfusion are very complex. Skin microcirculation is distributed between nutritive and thermoregulatory capillary blood flow.8 Previous investigations indicate that capillary vasomotion, the rhythmic oscillation in smooth muscle sphincters in capillary beds that controls the distribution of flow between microcirculatory beds, is sensitive to moderate ischemia,9 but overall skin perfusion is maintained at normal levels in patients with moderate PAD. Maintenance of skin perfusion in the presence of PAD appears to result in part from increased flow in the lower resistance thermoregulatory capillaries at the expense of nutritive blood flow. Although overall skin perfusion does tend to fall in critical ischemia, observations with plethysmography indicate that in many instances, total resting skin blood flow is often, paradoxically, increased with critical limb ischemia.10 This is clinically evident in some patients in whom obvious hyperemic skin is adjacent areas of ulceration or infarction. Of interest, we see in our own data several instances in limbs with ABI ≤0.45 where comparison of dorsal foot to volar forearm skin shows twofold or greater ratios for oxyHgb or %oxyHgb, consistent with hyperemic blood. Overall, previous studies of the effects of PAD on skin blood flow suggest that ischemia does not lead to consistent differences in total hemoglobin or hemoglobin saturation in superficial capillary beds when normal subjects are compared to those with moderate arterial disease, and probably skin blood flow measurement do not reliably detect severe disease because of hyperemic responses in some patients with critical limb ischemia.
We do not think our failure to observe significant differences between normal limbs and those affected by arterial disease reflects a deficiency in the HSI algorithm for measuring hemoglobin levels or oxygen saturation. In other experiments, we have seen marked changes in skin perfusion induced by postural changes or topical vasodilators measured by HSI in agreement with the results predicted by theory (see Fig. 6, Fig. 7).
Fig. 6
Effect of posture on HSI images. A Dorsal (D2) skin in normal subject with the lower extremity elevated at 45°: oxyHgb = 23, deoxy Hgb = 37, total Hgb = 60, %oxyHgb hemoglobin = 38%. B Same subject, standing: oxyHgb = 22, deoxyHgb = 76, total hemoglobin = 98, %oxyHgb = 22%. C Dorsal skin of patient with severe PAD with the limb elevated 45°: oxy Hgb = 31, deoxy Hgb = 40, total hemoglobin = 71, %oxyHgb = 44%. D Same patient, standing: oxy Hgb = 37, deoxy Hgb = 105, total hemoglobin = 142, %oxyHgb = 26%. The greater increase in total hemoglobin with dependency in the patient compared with the normal subject is consistent with the rubor often seen with dependency in PAD. The % oxyHgb shows a similar decrease in both subjects comparing elevated and dependent positions. Measurements for each skin field are based on the entire field imaged.
Fig. 7
HSI before and after applying a thin film of 0.5% nitroglycerin topically. Measurements were taken from just the area of skin treated with nitroglycerin. A Prior to application oxyHgb = 41 and %oxyHgb = 46%. B Fifteen minutes after application oxyHgb = 77 and %oxyHgb = 68%.
Although in our study, HSI measurements do not show sensitivity for PAD in a manner analogous to Doppler ultrasound measurements of tibial blood pressure, possibly responses to arterial interventions could be reliably assessed by comparing images obtained before and after surgical or medical therapy. Other ways of using HSI might make the technique useful for detecting or measuring arterial disease. For example, measurements of skin oxyHgb levels before and after releasing a proximal blood pressure cuff demonstrate delayed skin reperfusion in subjects with PAD as compared with normals (http://www.hypermed-inc.com/casestudy_2.html). A protocol based on measuring reperfusion kinetics might be a practical adjunct to noninvasive ultrasound, particularly when inability to compress tibial vessels interferes with determination of ankle systolic pressure. Future investigations of HSI and vascular disease are clearly warranted.
References
- The use of medical hyperspectral technology to evaluate microcirculatory changes in diabetic foot ulcers and predict clinical outcomes. Diabetes Care. 2007;30:903–910
- Systemic effect of shock and resuscitation monitored by hyperspectral imaging. Diabetes. Technol Ther. 2003;5:847–855
- Hyperspectral imaging: a new approach to the diagnosis of hemorrhagic shock. J Trauma. 2006;60:1087–1095
- Early changes in the skin microcirculation and muscle metabolism of the diabetic foot. Lancet. 2005;366:1711–1717
- Clinical utilization of near-infrared spectroscopy devices for burn depth assessment. Wound Repair Regen. 2007;15:332–340
- . Assessment of tissue viability using near-infrared spectroscopy. Br J Plast Surg. 1998;51:210–217
- . Investigation of TRAM flap oxygenation and perfusion by near-infrared reflection spectroscopy and color-coded duplex sonography. Plast Reconstr Surg. 2004;113:141–152
- . Skin microcirculation in peripheral arterial obliterative disease. Biomed. Pharmacother. 2004;58:427–431
- . Discrepancy between total and nutritional skin microcirculation in patients with peripheral arterial occlusive disease (PAOD). Vasa. 1990;2:105–111
- . Blood flow characteristics and tissue nutrition in apparently ischemic feet. Br Med J. 1971;3:220–224
PII: S0890-5096(10)00136-6
doi:10.1016/j.avsg.2010.03.005
Published by Elsevier Inc.
