Evaluation of peripheral artery disease with the TIVITA ® Tissue hyperspectral imaging camera system

Abstract

BACKGROUND:

Objective, reliable and easy screening for peripheral artery disease (PAD) is essential to confirm the diagnosis and initiate the respective treatment. Therefore, a new non-invasive hyperspectral camera (TIVITA^® Tissue) was tested in patients with and without PAD.

OBJECTIVE:

It was hypothesized that the oxygenation parameters of the TIVITA^® Tissue correlate to established modalities for detection of PAD and allow differentiation between individuals with and without PAD.

METHODS:

Evaluation of tissue oxygenation was performed in the angiosome of the medial plantar artery in 25 healthy young people and in 24 patients with and 25 patients without PAD in comparable age. Thereby, superficial oxygenation (StO₂) and near-infrared (NIR) perfusion index were measured with the TIVITA^® Tissue. Additionally, the ankle-brachial-index (ABI), the complaint free walking distance and the vascular quality of life were assessed and demographic data were obtained from all participants.

RESULTS:

TIVITA^® Tissue analysis revealed significantly reduced StO₂ and NIR perfusion index in PAD compared to healthy young participants and patients without PAD. StO₂ and NIR perfusion index positively correlated with ABI, the complaint free walking distance and the vascular quality of life score.

CONCLUSIONS:

In summary, this new hyperspectral imaging camera bears great potential for PAD screening as well as for follow up.

Keywords

Hyperspectral imaging vascular surgery screening perfusion imaging skin microcirculation

1 Introduction

Hyperspectral imaging (HSI) is a non-invasive, non-contact imaging technique that enables spectroscopical measurement of tissue oxygenation over a larger area without the need for contrast agents or radiation [1]. An overview of hyperspectral solutions in clinical applications is given by Lu and Fei [2]. Experimental studies showed that this technique is feasible for detection of impaired tissue perfusion following microvascular arterial anastomosis in the rat hind limb [3], for analysis of wound healing in a 3D in vitro model [4] and that HSI bears the potential to predict the region and extend of flap necrosis in random pattern dorsal skin flaps [5]. HSI has already been proven to be feasible for prediction of wound healing [6] and wound perfusion [7], of tissue regeneration in diabetic foot ulcers [8 –11] and for early detection of oxygenation and perfusion changes in irradiated skin [12, 13]. HSI has further been successfully tested for intraoperative brain cancer detection [14] and allows determining the ideal site for gastrointestinal anastomoses based on intestinal perfusion [15]. The tissue perfusion is analyzed in real-time due to the detection of oxygenated and deoxygenated hemoglobin. The correlation of HSI with perfusion of the skin further allows to evaluate burn wound depth in vivo [16].

Although the feasibility of HSI in detection of impaired perfusion in patients with peripheral artery disease (PAD) has already been shown [17 –19], none of the existing HSI system gained significance for clinical routine use. The camera systems that have been tested in the past are different to the new TIVITA^® Tissue (Diaspective Vision, Pepelow, Germany).

The TIVITA^® Tissue is designed to detect and measure the chromophores hemoglobin with its derivatives oxyhemoglobin O₂Hb and deoxyhemoglobin HHb and water to provide a two-dimensional map of the cutaneous and subcutaneous oxygenation pattern (StO₂ and NIR perfusion index), the relative hemoglobin content in the tissue (THI – tissue hemoglobin index) and the water content (TWI – tissue water index). Compared to near infrared spectrophotometric technique [20], this hyperspectral imaging camera therefore not only assesses oxygen saturation in the NIR part of the electromagnetic spectrum but gives additional information on the superficial oxygen saturation, i.e. StO₂. In comparison to these chemical-based assessments of tissue oxygenation other systems for evaluation of tissue perfusion like thermography or indocyanine angiography indirectly measure tissue perfusion by emission of infrared radiation or excitation of the exogenously supplied fluorescent dye respectively [21, 22]. There are also techniques to measure deeper partial tissue oxygen pressure using intra-muscular probe [23]. However, these are invasive and therefore less suitable for PAD screening. Optical imaging technologies like laser Doppler or laser speckle imaging (LDI, LSI) measure the blood flow in superficial skin layers [24 –26]. In several situations it is possible, that the blood flow and therefore the ‘perfusion’ is high but the oxygenation of the arterial blood is reduced or that the oxygen consumption in the target tissue is increased leading to an insufficient oxygen supply. Also, the presence of dyshemoglobins like methemoglobin or carboxyhemoglobin is concealed if the measurement is only based on the blood flow. Due to its chemical-based assessment of tissue oxygenation and hemoglobin content TIVITA^® Tissue allows a more profound analysis of tissue perfusion than it is possible with methods displaying the blood cell velocity or pulsation strength.

Previously discussed solutions are associated with many problems that preclude extensive use in clinical practice: either the devices and setups are too big, too complicated or too expensive as they include optical fibers with low image resolution and area size - or the scan was performed by moving the camera or the patient along motorized translation stages which both increases the risk of moving artifacts [27, 28]. Furthermore, the TIVITA^® Tissue is able to collect 100 spectral bands simultaneously, including both the visible and near-infrared spectral band that bears the possibility to assess tissue water and hemoglobin content.

Finally, most of the systems investigated previously are experimental assemblies and not commercially available as approved medical devices. The TIVITA^® Tissue camera system is a commercial product with CE (Conformité Européenne) certification and is therefore relevant for different clinical application areas.

The aim of this study was to assess how this new HSI camera can be used in the field of vascular medicine and whether it has the ability to detect the presence or absence of PAD and if the assessed oxygenation parameters correlate to established modalities like ankle-brachial-index (ABI), complaint free walking distance and to vascular quality of life.

It was hypothesized that the oxygenation parameters assessed by the TIVITA^® Tissue correlate to the established modalities and allows a more reliable detection of PAD as an additional tool.

2 Methods

Written informed consent of all volunteers was obtained before recruitment into the study in accordance with the vote of the local ethics committee (A 2017-0153). Exclusion criteria were age younger than 18 years, a history of minor or major amputation at the lower extremities, inability to consent or colonization with multi resistant pathogens.

For hyperspectral imaging the CE certified TIVITA^® Tissue camera was used (Fig. 1A).

Fig.1

The analysis was performed under standardized light conditions and room temperature. First the courser for local tissue analysis was positioned standardized with a defined diameter within the medial plantar artery angiosome in the redgreen image (A). TIVITA^® Tissue. Representative image of the setting for analysis by means of the TIVITA^® Tissue camera system (B). Afterwards quantifications of StO₂ (C), NIR perfusion index (D), THI (E) and TWI (F) were performed.

It provides precise reproducible parameters with high spatial resolution allowing highly reliable determinations of the perfusion state of the measured tissue and analysis of spatial perfusion distribution.

The technical concept of the pushbroom imaging camera is designed to capture one spatial and the spectral dimension at the same time. With this system, a hyperspectral image can be captured without the need of moving the entire camera or the patient. The measurement creates the so-called three-dimensional hyperspectral data cube with the spatial x, y-axis and the spectral λ-axis. It is important to understand that one single image of the CMOS sensor contains the spatial dimension y of the calculated images and the spectral dimension λ. The second spatial dimension x is recorded during the scan cycle. In other systems, the spectral information is recorded over time, for example with a filter wheel. If motion artifacts occur, the assignment of the spectral and spatial information is lost and the spectra can be distorted. In the TIVITA^® Tissue system, on the other hand, motion artifacts are clearly visible in the spatial dimensions but the spectral information is valid at every point.

The internal imaging spectrograph consists of a compact and lightweight construction that allows the camera to scan quickly and precisely. A transmission configuration with an optimized high efficiency holographic grating (>50% efficiency) is used for high light efficiency. The internal spectrometer unit (including the sensor) is calibrated for wavelength. This step also corrects the smile effect, an optical distortion caused by the finite extent of the lens optics, because the wavelength maps’ information is individually created for each line.

A CMOS sensor optimized for the near infrared spectral range is used, with a maximum resolution of 1280×960 pixels, while the region of interest (ROI) for the parameter calculation is set to 960×1008px (spat.×spec.) because relevant spectral and spatial information is projected to this image area size. From 1008 sensor pixels, 501 wavelength points from 500 to 1000 nm are calculated by the software. Therefore, the raw resolution of each scanned image is 960 px×501 wavelengths. The image resolution is reduced by software with a binning algorithm to a resulting value of 480 px×100 wavelength. For each measurement, 640 images are gathered, thus the resulting cube resolution is 640 px × 480 px × 100 wavelengths.

For the measurement shown in the results a varifocal lens with a fixed focal length of 8 mm and an aperture of F1.4 was used. A long-pass optical filter (Schott, GG 495) with additional anti reflection coating with a cutoff below 495 nm was mounted in front of the objective lens to suppress the second order of light intensities below 495 nm. The illumination of the entire investigation area of 21×28 cm is provided by an illumination unit containing six halogen lamps with a total power consumption of 120 W. The six single spots are arranged during production so that the resulting area is illuminated nearly uniformly without the need of additional diffusion elements.

The ambient illumination coming from a standard neon gas lamp from the ceiling was reduced as much as possible by switching off the light of one half of the room in which the camera was placed. Standardized light conditions were achieved by closure of the jalousie.

To convert image data from radiance (spectral flux that reaches the CMOS sensor) to reflectance (relation between the remitted light from an object and the light with 100% remission), a white reference cube was created by recording white reference object during the production progress of the device. This also balances regional differences of lighting. The previously recorded dark pattern is taken into account during the white balancing. The calculation steps give a hyperspectral imaging data cube with a single reflectance spectrum in each spatial point of the image.

HSI analysis includes the parameters Oxygenation - StO₂, NIR perfusion index, THI and TWI. The parameters StO₂ and NIR perfusion index describe the relative oxygen saturation of the blood in the microcirculatory system of the examined tissue area from different tissue depths. The THI describes the existing hemoglobin distribution in the microcirculatory system of the examined tissue area. The TWI has the same meaning in relation to water.

The calculation algorithms for theses are based on the different absorption of hemoglobin HHb, oxygenated hemoglobin O₂Hb and water. The parameter oxygenation StO₂ is calculated by using the second derivative from the spectrum. With this step, constant and linear influences are eliminated and absorption bands are amplified. Since the absorption curves of melanin and scattering are nearly linear in the interesting wavelength range, this step reduces the influence of scattering and melanin on the calculated values. The parameters NIR perfusion index, THI and TWI are based on the recorded absorbance spectrum, whereby individual wavelength ranges are evaluated for each parameter.

From a theoretical point of view and the suggestion of light path simulations, the remission spectroscopic measurement in the visible range (500–650 nm) gathers information about the superficial parts of the tissue whereas the measurement in the NIR-range (650–1000 nm) gains gathers information about its deeper layers. Based on the well-known optical properties of human skin a typical light penetration depth can be estimated to be between approx. 0.8 mm (500 nm) and 2.6 mm (1000 nm) [28, 29]. Because the StO₂ calculation is based on the VIS spectral range, it displays the oxygenation of the skin at about 0.8 mm depth and the parameter NIR perfusion index allows evaluation of subcutaneous perfusion at about 2.6 mm depth. It must be noted that a depth-selective measurement is not possible and that all layers penetrated by the light contribute to the signal.

The parameter calculation was developed by an empirical approach considering the known optical properties of tissues and blood and was calibrated by occlusion tests with healthy volunteers against a commercial tissue oximeter. Occlusion tests are the standard procedure for tissue oximeters (FDA class MUD) as well, in which two imaging multi-spectral cameras are approved for tissue oxygenation evaluation (K113507, K061848). In the validation study for the TIVITA^® Tissue device, a MOXY Monitor (Fortiori Design LLC, Hutchinson, MN, USA) tissue oximeter was used as a reference device for quantitative comparison of the parameter StO₂ and for qualitative analysis of the StO₂ and THI. The MOXY parameters are denoted as SmO₂ (oxygenation %) and THb (hemoglobin content in g/dl).

Numerical assessment of these parameters was performed using the camera-specific software package TIVITA^® Suite that allows definition of different regions for parameter analysis within the photographed tissue. The parameter calculation and validation were recently explained in more detail [30].

The parameters that are used for statistical representation are calculated in specified circular areas by the software package (Fig. 1C–F). The position of the circled area and its diameter can be chosen freely. Upon positioning of the marker, the average of each marker parameters is displayed with the respective spectrum (Fig. 2). To standardize the quantification of the parameters the circulated area was positioned always in the center of the angiosome and the diameter was unchanged between the studied individuals.

Fig.2

Representative example absorbance spectra from the angiosome of the medial plantar artery of a young healthy participant (green curve) and a patient with PAD (blue curve). The young healthy participant showed StO₂ of 61, NIR perfusion index of 61, THI of 47 and TWI of 55. The patients with PAD had a StO₂ of 43, NIR perfusion index was 47, THI 24 and TWI 48.

2.1 Patient cohort

Altogether, 74 persons were included into this prospective single-center study. To assess physiological perfusion parameters 25 young persons from 18 to 35 years without a history of cardiovascular disease, thrombotic events, coagulopathy or anti-thrombotic medication, including aspirin, were studied. To evaluate the effect of PAD on HSI parameters 24 patients with proven PAD [31] were studied on the scheduled day of admission to hospital, prior to endovascular or surgical treatment. The presence of PAD was established using the criteria of an ABI ≤0.9 [32]. In this group patients with one or more of the typical risk factors for PAD, i.e. diabetes, renal failure, arterial hypertension, smoking and dyslipidemia were included. In addition, a group of 25 hospitalized patients according in age to the patients with PAD but without any signs of PAD were studied to assess the effect of age on the perfusion parameters.

2.2 Conduction of data assessment

The study was performed between January and December 2018. In all tested persons, TIVITA^® Tissue reliably took images of the studied angiosome. To record the hyperspectral data, six seconds are required in which the scanning process is performed. During this time, the patient must remain still and the camera must not be moved in order to avoid motion artifacts. Afterwards the software needs another ten seconds to display the parameters.

Data assessment followed a standardized protocol and was performed in the same room to assure standardized light conditions and an ambient room temperature (20–23°C). One of the authors (N.A.S.) performed all of the HSI studies and data analysis and was not blinded to their results. Participants were required to abstain from smoking for at least 1 hour prior to data assessment. After written informed consent was obtained all study participants were first asked for their medical history, medication, prior vascular interventions or operations and had to fulfill the vascular quality of life questionnaire (VascuQoL 25). The VascuQoL 25 is established for assessment of quality of life in PAD regardless of disease severity (ie, in intermittent claudication and in critical limb ischemia). The questionnaire comprises 25 items, divided into five subscales (domains): pain (4 items), symptoms (4 items), activities (8 items), social (2 items), and emotional (7 items). Every item has a 7-point response scale. When summarizing item responses, an overall score is generated with a maximum of 169 points [33, 34].

Stable conditions in blood pressure and hard rate were reached easily in all participangs during this 25-minute lasting procedure. Then HSI was performed standardized in the angiosome of the medial plantar artery [35] by means of the TIVITA^® Tissue. In participants without PAD the angiosome of both feet were assessed while only the angiosome of the affected leg was studied in patients with PAD. The skin temperature in the angiosome was measured using an infrared thermometer (Metene Digital Infrared Non-Contact Forehead Thermometer, Metene, Meilong Town, Longhua District). For this procedure the participants lay in prone position on a standard hospital stretcher (Fig. 1B). On one hand, the angiosome was chosen with the idea to compare our data to the findings of Chiang et al. who used a different HSI camera system in patients with PAD [17]. On the other hand, the plantar angiosomes are covered by glabrous skin, which is rich in arteriovenous anastomoses, have generally higher oxygenation levels, and are generally more reactive compared with skin over other body parts [10, 19]. Afterwards the ABI was assessed on both sides (VascAssist^®, iSYMED, Butzbach, Germany) [36] before the complaint free walking distance was measured standardized.

2.3 Statistics

Statistical analysis was performed using R software (version 3.5.2) [37] variables were evaluated for normal distribution employing the Shapiro-Wilk-Test. In case of normal distribution, the t-test, otherwise a Wilcoxon rank-sum for dichotomous variables were used to compare 2 groups. Percentages were evaluated using either the chi-square test or the Fisher exact test. For detection of differences between all 3 groups a Kruskal-Wallis rank sum test was performed. Hyperspectral parameters were correlated with ABIs, complaint free walking distances and VascuQoL 25 scores using a Pearson (for normally distributed variables) or Spearman (in case of failed normal distribution) bivariate linear correlation analysis. Data were summarized as mean and 95% confidence interval (skin temperature, StO₂, NIR perfusion index, TWI and THI) or as median and interquartile range (age, pack years, ABI, complaint free walking distance and VascuQoL 25). Statistical significance was set at p < 0.05.

3 Results

3.1 Demographics of participants

The study recruited 74 individuals and assessed 126 angiosomes, with each 25 individuals in the group of young participants and of patients older than 50 years without PAD respectively. In the PAD group 24 patients were recruited. Two of these patients were treated on both legs in 2018 and therefore the angiosomes of both feet were studied. Statistical analysis of the characteristics of the groups (Table 1) revealed no significant differences regarding the sex-ratio (p = 0.085).

Table 1
Basic demographics of the study population. The groups differed in age when comparing young healthy participants to patients with and without PAD (p < 0.001 respectively). Patients with and without PAD showed no difference regarding the age (p = 0.18). The overall sex-ratio was also not statistically significant (p = 0.085). The groups further differed regarding the pack years (p = 0.02952). Diabetes and dyslipidemia were more frequent in patients with PAD compared to patients without PAD (p = 0.012 and p = 0.00076, respectively)

Young healthy participants n = 25 Patients without PAD n = 25 Patients with PAD n = 24

Age, years (median (IQR)) 25 (23–30) 60 (54–65) 64 (56.25–73.75)

Male (%) 76 52 79

Pack years (median (IQR)) 0 (0) 0 (0–20) 35 (2.5–45)

Diabetes on medication (%) 0 16 29

Dyslipidemia (%) 0 12 37.5

	Young healthy participants n = 25	Patients without PAD n = 25	Patients with PAD n = 24
Age, years (median (IQR))	25 (23–30)	60 (54–65)	64 (56.25–73.75)
Male (%)	76	52	79
Pack years (median (IQR))	0 (0)	0 (0–20)	35 (2.5–45)
Diabetes on medication (%)	0	16	29
Dyslipidemia (%)	0	12	37.5

Patients with and without PAD did not markedly differ regarding the age (p = 0.18) but were both significantly older than the young control group (p < 0.001, respectively). Moreover, diabetes and dyslipidemia were significantly more frequent in patients with PAD compared to the patients without PAD (p = 0.012 and p = 0.00076, respectively). The number of pack years did not significantly differ between patients with and without PAD (p = 0.21).

3.2 Tissue perfusion, hemoglobin index and tissue water index in participants without PAD

In young healthy participants from 18 to 35 years mean StO₂ and NIR perfusion index were 54.7 (2.55) and 60.34 (1.31) respectively (Fig. 3). Mean THI of the angiosome of the medial plantar artery was 14.8 (2.37). Mean TWI was 48.96 (0.99).

Fig.3

TIVITA^® Tissue findings in the angiosome of the medial plantar artery. HSI analysis revealed significantly reduced StO₂ (A) comparing patients with PAD to patients without PAD (p = 0.02). NIR perfusion index (B) was also significantly reduced in patients with PAD when compared to patients without PAD (p < 0.001). Additionally, THI (C) was significantly increased in patients with PAD compared to young healthy participants (p < 0.001) and to patients without PAD (p = 0.02). TWI (D) showed no statistically significant difference between the 3 groups.

In patients without PAD mean StO₂ and NIR perfusion index were 52.22 (2.38) and 58.04 (1.54) respectively. Mean THI of the angiosome of the medial plantar artery was 23.9 (3.38). Mean TWI was 53.24 (2.0).

3.3 Tissue perfusion, hemoglobin index and tissue water index in patients with PAD

Respective comparison of the patients without PAD to the patients with PAD, HSI analysis revealed significantly reduced values for StO₂ and NIR perfusion index in patients with PAD (Fig. 3). StO₂ ranged from 40–72 in young healthy participants, from 39–70 in patients without PAD and from 28–81 in patients with PAD. NIR perfusion index ranged from 53–71 in the young group, from 46–65 in patients without PAD and from 32–72 in patients with PAD. Compared to patients without PAD mean StO₂ and NIR perfusion index were reduced to 47.3 (5.11, p = 0.02) and 49.6 (3.66, p < 0.001) in patients with PAD. Mean THI was significantly increased compared to both young healthy participants and to patients without PAD with 30.36 (4.29, p < 0.001 and p = 0.02, respectively). Mean TWI was found at 51.58 (2.4) and was therefore comparable to the other studied groups.

3.4 Skin temperature, ABI, VascuQoL 25 score and walking distance free of complaints

Skin temperature in the assessed angiosome did not significantly differ between the three studied groups (p = 0.98). Median ABI was similar in healthy young participants and older patients without PAD with 1.13 (1.07–1.155) and 1.2 (1.16–1.3) (p = 0.1638) respectively. Compared to these two groups median ABI was found significantly reduced with 0.81 (0.66–1.00) in patients with PAD (p < 0.001). However, in nine from the 24 patients with PAD the ABI could technically not be assessed due to a markedly impaired perfusion of the legs.

Analysis of the vascular quality of life questionnaire revealed a median score of 169 points and 168 (162–169) points in healthy young participants and older patients without PAD, respectively. Compared to these groups patients with PAD had significantly lower scores with a median score of 91 (68–110) points (p < 0.001, respectively).

While walking distance was not restricted in healthy young participants and in patients without PAD, median complaint free walking distance was found at 70 m (50–127.5 m) in patients with PAD (Table 2).

Table 2
Established modality for detection of PAD. Analysis of skin temperature, ABI, complaint free walking distance and VascuQoL 25 score. Skin temperature was comparable between the groups (p = 0.98). ABI, complaint free walking distance and the VascuQoL 25 score were significantly reduced in patients with PAD when compared to young healthy participants and patients without PAD (p < 0.001 for both). In nine from 24 patients with PAD the ABI could technically not be assessed)

Young healthy participants n = 25 Patients without PAD n = 25 Patients with PAD n = 24

Skin temperature (mean (CI-95%)) 35.68 (0.11) 35.99 (0.15) 35.98 (0.18)

ABI (median + IQR) 1.13 (1.07–1.155) 1.2 (1.16–1.3) 0.625 (0–0.91)

Walking distance (m) (median + IQR) not restricted not restricted 70 (50–127.5)

VascuQoL 25 (median + IQR) 169 168 (162–169) 91 (68–110)

	Young healthy participants n = 25	Patients without PAD n = 25	Patients with PAD n = 24
Skin temperature (mean (CI-95%))	35.68 (0.11)	35.99 (0.15)	35.98 (0.18)
ABI (median + IQR)	1.13 (1.07–1.155)	1.2 (1.16–1.3)	0.625 (0–0.91)
Walking distance (m) (median + IQR)	not restricted	not restricted	70 (50–127.5)
VascuQoL 25 (median + IQR)	169	168 (162–169)	91 (68–110)

3.5 Correlation between tissue oxygenation and ABI, complaint free walking distance and VascuQoL 25 score

A bivariate linear correlation analysis of limbs with PAD showed a ratio between limb ABI and the hyperspectral-derived values for StO₂ (R = 0.43, p = 0.09) and for NIR perfusion index (R = 0.11, p = 0.69) (Fig. 4A and B).

In parallel, NIR perfusion index but not StO₂ showed significant ratio to the complaint free walking distance (R = 0.43, p = 0.03 and R = 0.36, p = 0.07, respectively) (Fig. 4C and D).

Fig.4

Correlations of NIR and StO₂ to established modalities of PAD screening in patients with PAD. Scatter plots showing the correlation between NIR perfusion index and StO₂with ABI (A and B), complaint free walking distance (C and D) and vascular quality of life score (VascuQoL 25, E and F).

In patients with PAD NIR perfusion index but not StO₂ showed statistically significant ratio to the VascuQoL 25 score (R = 0.42, p = 0.03 and R = 0.37, p = 0.07, respectively) (Fig. 4E and F).

In contrast, neither young healthy participants nor patients without PAD showed statistically significant correlations for StO₂ or NIR perfusion index with ABI, complaint free walking distance or VascuQoL 25 score.

4 Discussion

In this study, it was shown that tissue perfusion analysis by TIVITA^® Tissue reliably revealed the decrease of tissue oxygenation in patients with PAD in the angiosome of the medial plantar artery (i). Furthermore, the decreased oxygenation values positively correlated with ABI (ii), complaint free walking distance (iii) and vascular quality of life (iv). The whole examination, comprising the survey of the VascuQoL 25 score, TIVITA^® Tissue analysis, ABI analysis and assessment of complaint free walking distance, took about 25 minutes per participant. TIVITA^® Tissue analysis alone took about 20 seconds after positioning of the participant which is faster compared to the time required for ABI assessment. HSI analysis was performed under standardized conditions in North Europeans to assure comparability and reproducibility of the gained data. Although light conditions and room temperature or skin pigmentation could affect HSI analysis they were not especially addressed in this proof of concept study.

PAD is an increasingly prevalent disorder affecting millions of patients across the world [19]. Clinical signs of PAD like decreased complaint free walking distance, painful legs under resting conditions or a delay in wound healing are sometimes misdiagnosed as a result of spinal disc herniation, chronic venous insufficiency or other reasons. Therefore, the underlying pathology of arteriosclerotic stenosis or occlusion of arteries is not diagnosed in time with often disastrous consequences for the patients. Currently used techniques to diagnose PAD are sometimes limited by patient’s physical characteristics as obesity, local inflammation and edema might limit palpation of pulses. The ABI can be distorted by calcified arteries or by diabetic media sclerosis [38] and other techniques like Doppler- or duplex ultrasound analysis of arterial perfusion depends on experience and therefore needs a specialized examiner. Neither palpation of pulses nor assessment of ABI give exact information regarding localization of the vascular pathology and are not appropriate for precise follow up of patients after endovascular or surgical revascularization. In order to simplify diagnosis and lead to an earlier detection and treatment of impaired arterial tissue perfusion HSI could be a useful tool for user-independent ambulant screening for PAD.

As mentioned above, the studied angiosome was chosen to compare our findings to the data of Chiang et al. who used the OxyVu™ HSI camera system in patients with PAD [17]. This camera system only detects wave lengths of the visible light [19]. StO_2, the oxygenation parameter assessed in the visible light range from 500–660 nm, also positively correlated to the ABI (R = 0.43) when compared to the study of Chiang et al. (R = 0.42). However, in 9 of the 24 studied patients with PAD ABI could technically not be assessed due to a progressed atherosclerosis. However, in these patients HSI analysis allowed assessment and quantification of the respective oxygenation parameters.

Skin temperature did not significantly differ between the study groups which is also in line with the findings of Chiang et al. [17]. The similar skin temperature between the groups has to be underlined as it excludes a relevant effect on microcirculatory perfusion that is significantly influenced by the local temperature that could therefore affect the HSI perfusion analysis.

In addition, both NIR perfusion index and StO₂ showed a positive correlation to the VascuQoL 25 score with a significant correlation for NIR perfusion index and a distinct correlation for StO₂, even though not reaching statistically significance. This is also the case for the complaint free walking distance. NIR perfusion index revealed a significant positive correlation while StO₂ also distinctively correlated to complaint free walking distance. This aspect is important as a decrease in ambulatory ability is one of the most frequent reasons for patients with PAD consulting a vascular specialist initially. In this case diagnostics, additionally to the established modalities, by means of a HSI camera could confirm the presence of a PAD or exclude it so that other, e.g. muscular or neurologic causes have to be clarified. Thereby, both StO₂ and NIR perfusion index are indicative for PAD and should be interpreted together.

However, no statistically significant correlations were found between ABI, complaint free walking distance and VascuQoL 25 score and StO₂ or NIR perfusion index in young healthy participants and patients without PAD. This could be explained by the almost maximal VascuQoL 25 score and the unimpaired walking distance in both groups.

Another clinical examination of patients with vascular diseases by means of the OxyVu™ HSI camera system to study tissue oxygenation in the upper extremities upon vascular occlusion by a cuff also approved the potential for assessment of tissue perfusion in patients with cardiovascular disease [18]. In this study oxygenated hemoglobin was found reduced in patients with PAD and coronary artery disease. This is in line with the TIVITA^® Tissue analysis showing that StO₂ and NIR perfusion index, both reflecting the oxygenated hemoglobin, are significantly reduced in patients with PAD.

TWI is an additional parameter, based on the light absorption of tissue water, which allows to evaluate whether the tissue water changes over the course of time as a parameter for tissue edema. This could enable the physician to assess edema formation in soft tissue that might be based on renal dysfunction, on impaired venous or lymphatic drainage due to thrombosis or external compression or on an ischemia reperfusion injury after revascularization. TWI analysis revealed no statistical difference between the studied groups. Therefore, it can be assumed that the data assessment was not influenced by tissue edema. The clinical value of this tissue water analysis additionally needs to be studied in detail in patients before and after surgical or endovascular revascularization.

THI analysis revealed on one hand an increase in hemoglobin in the older, hospitalized participants of the study when compared to the young group. On the other hand, patients with PAD showed significantly increased THI than patients without PAD and young healthy people. So, both the age and the presence of PAD seem to affect the tissue hemoglobin content. A relevant effect of skin pigmentation on the measurement can be excluded as the studied plantar skin is usually untanned. It might be assumed that patients with PAD show increased tissue hemoglobin content as the blood flow in the periphery is typically reduced due to vasodilatation and new formed blood vessels comprise erythrocytes which could explain the higher THI. At the same time high levels of THI in patients with PAD are connected with lower StO₂ and NIR perfusion index. This means that the most amount of the assessed total tissue hemoglobin is deoxygenated. This is contrary to the findings of Chin et al. who described a decrease in deoxygenated hemoglobin in patients with PAD [19]. This might be explained by a different camera system used.

Although the young group of participants had no metabolic, cardiac or respiratory diseases compared to the older group of patients without PAD the values of StO₂ and NIR perfusion index did not significantly differ between these groups. So, neither the age nor these diseases, that are well known risk factors and comorbidities of PAD, seem to affect HSI analysis of tissue oxygenation.

The use of vasoactive substances like caffeine or antihypertensive medication were not considered in this analysis. It might be assumed that caffeine, nicotine or antihypertensive medication induce vasoconstriction that could impair cutaneous and subcutaneous microcirculation but would not affect the oxygenation of hemoglobin. Therefore, neither caffeine, nicotine nor antihypertensive medication are expected to affect HSI analysis substantially. The number of pack years, reflecting the nicotine consumption over the life time, was markedly increased in patients with PAD when compared to young healthy individuals and patients without PAD. This finding was expected as smoking is one of the most important risk factors for PAD. Although skin characteristics like moisture or thickness of the skin can affect HSI they were not considered in this study as they are patient-individual, non-modifiable parameters during perfusion analysis. To minimize the effect of skin characteristics on HSI analysis only North Europeans were recruited to assure comparable skin pigmentation.

However, until now no valid range, cut-off levels or threshold values were defined that are necessary to characterize tissue perfusion and to estimate the degree of PAD. To do so and to analyze the sensitivity and the specificity of HSI, a bigger study population is necessary. In addition, a separated follow-up of endovascular or surgically treated patients is needed to assess whether HSI by means of TIVITA^® Tissue allows post-therapeutically success control. In addition, systematic analysis of the other angiosomes of the lower extremity could finally allow to determine the location of vascular pathology.

Due to the possible side effects mentioned above, it is also possible that the measurement of perfusion and oxygenation values with HSI at a defined point in time is not completely suitable for a reliable diagnosis without considering the time course.

Combining this diagnostic technique with additional methods such as venous occlusion and measurement over an extended period of time as investigated by Khalil et al. may be an interesting approach for further investigations [39]. In this study, the dynamic change of the hemoglobin content due to an applied venous occlusion was evaluated with an experimental NIRS tomography device. The results show that this method may be well suitable for diagnosing PAD and verifying the success of an intervention. As the shown device is not an easy to use and approved medical device, the simplicity and speed of the TIVITA^® Tissue HSI camera could be a significant improvement with a great impact to clinical practice.

The small number of included individuals is another limitation of this study that does not allow to determine the sensitivity and specificity of the TIVITA^® Tissue. Due to the small number of patients the severity of PAD, i.e. claudication or critical ischemia, was not considered. In addition, the location of arterial stenosis or occlusion in patients with PAD was not considered during the recruitment and the HSI analysis. The patients selected for this study had PAD, severe enough for indicating an intervention. This design was chosen to test whether the TIVITA^® Tissue allows to detect PAD in general. In future, patients with mild or minimal asymptomatic PAD need to be studied as well to prove the suitability of this technique for PAD screening. Therefore, this work has to be interpreted as a proof of concept study that aimed to demonstrate whether this new HSI camera system allows for detection of perfusion disorders in patients with PAD.

5 Conclusion

In summary, the above established hypothesis was confirmed regarding the correlation of the HSI parameters to established modalities. The assumption that HSI alone allows to detect the presence or absence of PAD could not be confirmed. As StO₂ better correlates to ABI than NIR perfusion index and NIR perfusion index, but not StO₂, significantly correlates to complaint free walking distance and VascuQoL 25 score, both parameters should be interpreted when assessing PAD.

However, it could be shown that TIVITA^® Tissue can be an important additional tool to support and improve the diagnosis of PAD. This study underlines the feasibility of HSI for easy, fast and reliable detection of impaired tissue perfusion in patients with PAD. It shows the great potential of this new camera system for easy screening and potential follow up assessment of patients with PAD. However, further clinical studies are needed to define physiological parameters for tissue oxygen saturation, hemoglobin- and water content and to assess whether the TIVITA^® Tissue enables monitoring and post-therapeutic follow up of patients with PAD.

Disclosures

The hyperspectral camera described in this study was developed by Diaspective Vision GmbH. Mr. A. Holmer is employee of this company. In the long term, Diaspective Vision has proprietary interest in the development of the camera system resulting in a product for routine clinical use. We certify that the clinical investigators and co-authors have no financial interests and financial arrangements with Diaspective Vision GmbH und have received no funding for the preparation of this manuscript.

Footnotes

Acknowledgments

The authors cordially thank all participants and patients who were involved in this study.

References

Zuzak

, Schaeberle

, Lewis

, Levin

IW.

. Visible reflectance hyperspectral imaging: Characterization of a noninvasive, in vivo system for determining tissue perfusion. Anal Chem. 2002;74(9):2021–8.

, Fei

. Medical hyperspectral imaging: A review. J Biomed Opt. 2014;19(1): 10901. doi: 10.1117/1.JBO.19.1.010901

Grambow

, Dau

, Holmer

, Lipp

, Frerich

, Klar

, et al.Hyperspectral imaging for monitoring of perfusion failure upon microvascular anastomosis in the rat hind limb. Microvasc Res. 2018;116:64–70. doi: 10.1016/j.mvr.2017.10.005

Wahabzada

, Besser

, Khosravani

, Kuska

, Kersting

, Mahlein

, et al.Monitoring wound healing in a 3D wound model by hyperspectral imaging and efficient clustering. PLoS One. 2017;12(12):e0186425. doi: 10.1371/journal.pone.0186425

Chin

, Chappell

, Giatsidis

, Perry

, Lujan-Hernandez

, Haddad

, et al.Hyperspectral Imaging Provides Early Prediction of Random Axial Flap Necrosis in a Preclinical Model. Plast Reconstr Surg. 2017;139(6):1285e–1290e. doi: 10.1097/PRS.0000000000003352

Calin

, Coman

, Parasca

, Bercaru

, Savastru

, Manea

. Hyperspectral imaging- based wound analysis using mixture-tuned matched filtering classification method. J Biomed Opt. 2015;20(4):046004. doi: 10.1117/1.JBO.20.4.046004

Daeschlein

, Langner

, Wild

, von Podewils

, Sicher

, Kiefer

, et al.Hyperspectral imaging as a novel diagnostic tool in microcirculation of wounds. Clin Hemorheol Microcirc. 2017;67(3-4):467–74. doi: 10.3233/CH-179228

Yudovsky

, Nouvong

, Pilon

. Hyperspectral imaging in diabetic foot wound care. Hyperspectral imaging in diabetic foot wound care. J Diabetes Sci Technol. 2010;4(5):1099–113.

Nouvong

, Hoogwerf

, Mohler

, Davis

, Tajaddini

, Medenilla

. Evaluation of diabetic foot ulcer healing with hyperspectral imaging of oxyhemoglobin and deoxyhemoglobin. Diabetes Care. 2009;32(11):2056–61. doi: 10.2337/dc08-2246

10.

Khaodhiar

, Dinh

, Schomacker

, Panasyuk

, Freeman

, Lew

, et al.The use of medical hyperspectral technology to evaluate microcirculatory changes in diabetic foot ulcers and to predict clinical outcomes. Diabetes Care. 2007;30(4):903–10.

11.

Greenman

, Panasyuk

, Wang

, Lyons

, Dinh

, Longoria

, et al.Early changes in the skin microcirculation and muscle metabolism of the diabetic foot. Lancet. 2005;366(9498):1711–7.

12.

Chin

, Freniere

, Lo

, Saleeby

, Baker

, Strom

, et al.Hyperspectral imaging for early detection of oxygenation and perfusion changes in irradiated skin. J Biomed Opt. 2012;17(2):026010. doi: 10.1117/1.JBO.17.2.026010

13.

Chin

, Freniere

, Bonney

, Lancerotto

, Saleeby

, Lo

, et al.Skin perfusion and oxygenation changes in radiation fibrosis. Plast Reconstr Surg. 2013;131(4):707–16. doi: 10.1097/PRS.0b013e3182818b94

14.

Fabelo

, Ortega

, Ravi

, Kiran

, Sosa

, Bulters

, et al.Spatio-spectral classification of hyperspectral images for brain cancer detection during surgical operations. PLoS One. 2018;13(3):e0193721. doi: 10.1371/journal.pone.0193721

15.

Jansen-Winkeln

, Maktabi

, Takoh

, Rabe

, Barberio

, Köhler

, et al.Hyperspectral imaging of gastrointestinal anastomoses. Chirurg. 2018;89(9):717–25. doi: 10.1007/s00104-018-0633-2

16.

Chin

, Babchenko

, Lujan-Hernandez

, Nobel

, Ignotz

, Lalikos

JF.

. Hyperspectral Imaging for Burn Depth Assessment in an Animal Model. Plast Reconstr Surg Glob Open. 2016;3(12).e591. doi: 10.1097/GOX.0000000000000558

17.

Chiang

, Jain

, Sleigh

, Vasudevan

. Evaluation of hyperspectral imaging technology in patients with peripheral vascular disease. J Vasc Surg. 2017;66(4):1192–201. doi: 10.1016/j.jvs.2017.02.047

18.

Sumpio

, Citoni

, Chin

, Sumpio

BE.

. Use of hyperspectral imaging to assess endothelial dysfunction in peripheral arterial disease. J Vasc Surg. 2016;64(4):1066–73. doi: 10.1016/j.jvs.2016.03.463

19.

Chin

, Wang

, Kibbe

MR.

. Evaluation of hyperspectral technology for assessing the presence and severity of peripheral artery disease. J Vasc Surg. 2011;54(6):1679–88. doi: 10.1016/j.jvs.2011.06.022

20.

Koutsiaris

. Deep tissue near infrared second derivative spectrophotometry for the assessment of claudication in peripheral arterial disease. Clin Hemorheol Microcirc. 2017;65(3):275–84.

21.

Kluz

, Małecki

, Adamiec

. Practical importance and modern methods of the evaluation of skin microcirculation during chronic lower limb ischemia in patients with peripheral arterial occlusive disease and/or diabetes. Int Angiol. 2013;32(1):42–51.

22.

Neumann

, Schmaderer

, Finsterer

, Zimmermann

, Steubl

, Helfen

, et al.Noninvasive quantitative assessment of microcirculatory disorders of the upper extremities with 2D fluorescence optical imaging. Clin Hemorheol Microcirc. 2018;70(1):69–81. Doi: 10.3233/CH-170321

23.

Sternitzky

, Keßler

, Mrowietz

, Jung

. Effect of naftidrofuryl on intramuscular oxygen pressure prior to, during, and after load on the treadmill in apparently healthy subjects. Clin Hemorheol Microcirc. 2002;27:185–92.

24.

Mrowietz

, Franke

, Pindur

, Sternitzky

, Jung

, Wolf

. Evaluation of Laser-Doppler-Fluxmetry for the diagnosis of microcirculatory disorders. Clin Hemorheol Microcirc. 2019;71(2):129–35. doi: 10.3233/CH-189402

25.

Schioppo

, Orenti

, Boracchi

, De Lucia

, Murgo

, Ingegnoli

. Evidence of macro- and micro-angiopathy in scleroderma: An integrated approach combining 22-MHz power Doppler ultrasonography and video-capillaroscopy. Microvasc Res. 2019;122:125–30. doi: 10.1016/j.mvr.2018.07.001

26.

Martini

, Bagno

. The wavelet analysis for the assessment of microvascular function with the laser Doppler fluxmetry over the last 20 years. Looking for hidden informations. Clin Hemorheol Microcirc. 2018;70(2):213–29. doi: 10.3233/CH-189903

27.

Denstedt

, Pukstad

, Paluchowski

, Hernandez-Palacios

, Randeberg

LL.

. Hyperspectral imaging as a diagnostic tool for chronic skin ulcers. SPIE Proceedings. 2013;85650N. doi: 10.1117/12.2001087

28.

Schmidt

, Liebold

, Fassler

, Wollina

UW.

. Contact-free spectroscopy of leg ulcers: Principle, technique, and calculation of spectroscopic wound scores. J Invest Dermatol. 2001;116(4):531–5.

29.

Bashkatov

, Genina

, Kochubey

, Tuchin

VV.

. Optical properties of human skin, subcutaneous and mucous tissues in the wavelength range from 400 to 2000nm. J Phys D: Appl Phys. 2005;38(15):2543–55. doi: 10.1088/0022-3727/38/15/004

30.

Holmer

, Marotz

, Wahl

, Dau

, Kämmerer

PW.

. Hyperspectral imaging in perfusion and wound diagnostics – methods and algorithms for the determination of tissue parameters. Biomed Tech (Berl). 2018;63(5):547–56. doi: 10.1515/bmt-2017-0155

31.

Martini

. Current opinions about the definition of critical limb ischemia: A debate still open after three decades. Clin Hemorheol Microcirc. 2019. doi: 10.3233/CH-190576

32.

getABI Study group. getABI: German epidemiological trial on ankle brachial index for elderly patients in family practice to dedect peripheral arterial disease, significant marker for high mortality. Vasa. 2002;31(4):241–8.

33.

Nordanstig

, Wann-Hansson

, Karlsson

, Lundström

, Pettersson

, Morgan

MB J.

. Vascular Quality of Life Questionnaire-6 facilitates health-related quality of life assessment in peripheral arterial disease. J Vasc Surg. 2014;59(3):700–7. doi: 10.1016/j.jvs.2013.08.099

34.

Morgan

, Crayford

, Murrin

, Fraser

SC.

. Developing the Vascular Quality of Life Questionnaire: A new disease-specific quality of life measure for use in lower limb ischemia. J Vasc Surg. 2001;33(4):679–87.

35.

Taylor

, Palmer

JH.

. The vascular territories (angiosomes) of the body: Experimental study and clinical applications. Br J Plast Surg. 1987;40(2):113–41.

36.

Kraushaar

, Dressel

, Maßmann

. Vascular robustness: The missing parameter in cardiovascular risk prediction. Prev Med Rep. 2018;9:107–13. doi: 10.1016/j.pmedr.2018.01.008

37.

R Core Team. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria, ISBN 3-900051-07-0, (2013) URL http://www.R-project.org/

38.

Nawalany

. Endovascular therapy for limb salvage. Surg Clin North Am. 2010;90(6):1215–25. doi: 10.1016/j.suc.2010.08.007

39.

Khalil

, Kim

, Flexman

, Dayal

, Shrikhande

, et al.Dynamic diffuse optical tomography imaging of peripheral arterial disease. Biomed Opt Express. 2012;3(9):2288–98.