Measurement of tissue oxygen saturation levels using portable near-infrared spectroscopy in clinically healthy dogs


  • Funding: Internal private funding provided by Dr. Carl Osborne, Professor of Medicine, University of Minnesota Veterinary Medical Center.
    Previously presented: Presented in abstract form at the International Veterinary Emergency and Critical Care Symposium, Phoenix, AZ, September 2008.
    Conflict of interest: Dr. Greg Beilman is a member of the Scientific Advisory board of Hutchinson Technology Inc.

Address correspondence and reprint requests to Dr. Kelly E. Hall, University of Minnesota College of Veterinary Medicine, 1352 Boyd Avenue, Saint Paul, MN 55108, USA.


Objective – To establish a reference interval for tissue oxygen saturation (StO2) levels measured by a portable near-infrared spectroscope and determine site(s) for reproducibly measuring StO2 levels in dogs.

Design – Prospective experimental study.

Setting – Veterinary teaching hospital.

Animals – Seventy-eight healthy dogs.

Measurements and Main Results – A portable device that quantitatively measures StO2 levels directly in muscle tissue using near-infrared spectroscopy (NIRS) was topically applied to shaved sites over 4 muscle bodies. Readings from the sartorius muscle were obtained 100% of the time. The digital extensors and biceps femoris muscles provided similar readings, but less consistently obtained StO2 values (70% and 67%, respectively). Mean StO2 levels measured over these 3 sites were not statistically different from one another. When readings from these 3 sites were combined, a mean ±1 SD of 92.9±7.4% was obtained. The epaxial muscles produced a significantly lower mean ±1 SD (68.5±22.4%), and readings were obtained only 60% of the time.

Conclusions – In dogs, a mean ±1 SD of 92.9±7.4% can be used to investigate clinical applications of NIRS. The sartorius muscle most consistently allows for detection of StO2 levels (100%). The epaxial muscles are not consistent or reliable for obtaining StO2 readings and are not recommended for clinical application of near-infrared spectroscope. Sex does not significantly affect StO2 readings at any site. Body condition score only affects readings obtained from the sartorius muscle.


A cornerstone in the treatment of critically ill patients is to ensure appropriate oxygen delivery to meet tissue oxygen demands. Pulmonary arterial catheterization is a highly invasive technique that allows for determination of an individual patient's oxygen delivery and oxygen extraction capabilities. Unfortunately, this technique requires placement of a specialized catheter into the pulmonary artery: a time intensive, invasive, and expensive procedure. The InSpectra Tissue Spectrometer is a patented, noninvasive monitoring technology that quantitatively measures tissue oxygen saturation (StO2) directly in muscle tissue using near-infrared (NIR) light. The device provides immediate, continuous, and reproducible StO2 measurements at various tissue depths using a noninvasive monitoring contact probe that rests on or adheres to the skin over the area to be measured.1 In addition to its noninvasive nature, its portability and user-friendly features make it an ideal device for use with veterinary patients.

Extensive laboratory data from animal models and early human clinical trials have been obtained with the portable device, with promising results for its use as a monitoring tool, reflection of global perfusion, and a prognostic indicator.2–8 For example, peripheral muscle StO2 levels and mixed venous oxygen saturation (SvO2) were shown to predict resuscitatable from nonresuscitatable animals in a porcine model of hemorrhagic shock.7 Additionally, evidence suggests that StO2 levels may be an end-point resuscitation data point for hemorrhagic shock.6,8 The only published clinical studies in animals evaluating the use of NIRS evaluated a similar, nonportable device for use in lactating cows and horses with laminitis.9,10

The purpose of this study was to (1) establish a reference interval for StO2 levels measured by a portable NIRS device in healthy dogs and (2) determine appropriate sites for reproducibly measuring StO2 in dogs. We hypothesized that the epaxial muscles would be the least consistent and accurate site for obtaining StO2 levels due to increased frequency of fat deposition over the lower back in dogs. The establishment of StO2 reference intervals for dogs will enable further study in the use of the device for clinical patients.

Materials and Methods


Eighty-seven healthy client-, staff-, and student-owned dogs were evaluated for enrollment in the study. All dogs evaluated for the study were volunteers, not clinical patients. The study protocol was reviewed and approved by the Institutional Animal Care and Use Committee (IACUC), and informed consent was obtained from the owner of each dog evaluated. The goal of the study was to evaluate a minimum of 60 animals, based on research for validating reference intervals.11

Inclusion criteria included adult dogs weighing >9 kg with a normal physical exam, normal complete blood counta and serum biochemical analysis,b systolic blood pressure ≥85 mm Hg when measured using an ultrasonic Doppler flow detector,c and either pulse oximetry reading ≥96%d or arterial partial pressure of oxygen (PaO2) ≥85 mm Hg.e Of the 87 dogs evaluated, 9 were excluded for various reasons including resistance to restraint for blood collection and diagnostic testing (n=1), diagnosis of endocrine disease (n=2), positive Borrelia burgdorferi serum statusf (n=2), healing skin wounds (n=1), body condition score (BCS) 1/9 (n=1), reported diarrhea (n=1) and body weight <9 kg (n=1). The data reported and analyzed represent results from the 78 dogs that met all inclusion criteria.

Experimental Protocol

All dogs were examined by 1 of 2 veterinarians (KEH, LLP). Parameters recorded included body weight, attitude, rectal temperature, heart rate, femoral pulse quality, respiratory rate and effort, mucous membrane color, capillary refill time and BCS (based on a 1–9 scale, with 1=emaciated, 5=ideal, and 9=grossly obese).12 Venous whole blood samples for CBC and serum biochemical analysis were obtained via the jugular vein. All venous samples were immediately submitted, routinely processed, and analyzed by the in-house clinical laboratory. Arterial whole blood samples were collected anaerobically from the dorsal pedal or femoral artery directly into a lithium heparin blood gas syringe.g Arterial samples to obtain blood gas data were analyzed within 5 minutes of collection by 1 of 2 trained medical technologists to avoid operator bias.

Blood pressures were obtained using an ultrasonic Doppler flow detector and standard clinical methods (probe placed over an artery on a peripheral limb, and cuff size chosen based on limb circumference). Blood pressure measurements and blood samples were obtained by 1 of 3 certified veterinary technicians.

Muscle StO2 values were obtained using the Inspectra Tissue Spectrometer.h The spectrometer works by analyzing returned light from muscle tissue illuminated at a standardized depth of approximately 14 mm below the sensor by 4 calibrated wavelengths of NIR light. The device does not require manual calibration and can display readings within 20 seconds of being turned on. The 'system produces an absolute measurement of oxygen saturation in the microcirculation, where oxygen is exchanged with tissue.'1 When placed over a muscle body (Figure 1), the lower right corner of the screen (Figure 2) displays current StO2 levels every 2 seconds. The main screen allows for monitoring trends in StO2 levels over minutes, hours or days, depending on user programmed selection.

Figure 1.

 Near-infrared spectroscope (NIRS) probe on biceps femoris.

Figure 2.

 Portable near-infrared spectroscope (NIRS) with continuous data readings (InSpectra Tissue Spectrometer).

The lower left corner of the unit (Figure 2) displays the tissue hemoglobin index (THI). The THI measurement is reported by the manufacturer to be an indicator of signal strength based on the total amount of measured hemoglobin present in the monitored tissue. In humans, THI levels >5.0 indicate sufficient hemoglobin to obtain an adequate StO2 signal in most circumstances. ‘Values <5.0 indicate a weaker hemoglobin signal in which case the accuracy of the StO2 value may be reduced.’1

To minimize interference during measurement from the presence of hair, an approximately 5 cm × 5 cm (2 in × 2 in) site was shaved on each dog over the 4 areas analyzed: epaxial (lower back), digital extensors (forearm), biceps femoris (lateral thigh), and sartorius (medial thigh). In situations where the epaxial muscle was not palpable, readings were attempted mid paralumbar region (L3–L5). Other than clipping of hair, the skin was not altered in any way. The probe of the tissue spectrometer was then placed on the skin over each of the 4 sites and gently held in place by one of the investigators for 5–30 seconds and the highest obtained reading was recorded (Figures 1 and 2). If a reading was obtained from the site (ie, the machine displayed a value in the right lower corner of the unit), a value was usually displayed on the screen within 10 seconds of being applied. Time from application of the probe to display of a reading was not recorded. When no StO2 level was displayed on the screen within 30 seconds of application of the probe to the animal, it was recorded that no reading was able to be obtained from that particular site (reading failure). During measurement, most animals were in the standing position; however, depending on animal demeanor and comfort, some readings were obtained with the dog in lateral recumbency. Animal positioning was not standardized for this study. Readings were obtained from either the left side or the right side for each muscle body (not both in any given animal). Readings were taken from the right side approximately half of the time (epaxial 67%, digital extensor 56%, biceps femoris 55%, sartorius 45%).

Statistical Analysis

The mean and SD were determined for StO2 levels at each site and by subpopulation (sex and BCS). Interval ranges were determined for each site. Interval ranges are from 3rd smallest to 3rd largest for single site, and 9th smallest to 9th largest with readings from statistically similar sites combined (digital extensor, biceps femoris, sartorius). This provides a 91–95% prediction interval (see Table 1) calculated as (n−2x/n) × 100 where n is the number of readings obtained and x the number of values removed from smallest/largest values obtained. For example, for single site measurements, x=2 and therefore at the epaxial site, prediction interval is (47−4/47) × 100=91%. We did not use large sample normal theory because the distribution of values for individual dogs is skewed to the left.

Table 1.   Measurement success and results of tissue spectroscopy by site in clinically healthy dogs
Site measuredNumber of readings
(% of 78 attempted)
StO2 mean ± 1 SD (%)Prediction
StO2 interval
range (%)
  • *

    Values significantly different than all other sites (P<0.001).

  • Prediction interval is [(n−2x)/n] × 100 where n is the number of readings and x the number of values removed from smallest/largest values obtained.
    StO2 interval range reports 3rd (9th) lowest to 3rd (9th) highest value for each site (3 statistically similar sites combined).

  • StO2: tissue oxygen saturation.

Epaxial (lower back)*47 (60%)68.5 ± 22.491%25–94
Digital extensors (forearm)55 (70%)94.1 ± 5.993%82–99
Biceps femoris (lateral thigh)52 (67%)93.8 ± 7.592%76–99
Sartorius (medial thigh)78 (100%)91.4 ± 8.195%71–99
Statistically similar sites combined (digital extensors, biceps femoris, sartorius)185 (79% of 234 readings)92.9 ± 7.491%76–99

Dogs were classified by BCS into 3 groups (BCS 5–5.5, 6–6.5, >7). Mean StO2 measurements at the sartorius site were compared with BCS and sex using a one-way anova and the Student t-test. In order to compare measurements at each site, we first fit a two-way anova, then compared the 4 sites using the Tukey HSD test.i


Dogs evaluated weighed 9–53.6 kg (mean 26.3 kg, median 25.8 kg) and were 1.5–8.5 years old (mean 4.5 years, median 4 years). There were 39 males (37 neutered, 2 intact) and 39 females (38 spayed, 1 intact). BCS ranged from 5 to 8.5 (mean 5.97, median 6). The mean, SD, and interval range of StO2 levels measured at each of the 4 sites for the entire population of included animals are reported in Table 1. Readings from the epaxial site are significantly different than those obtained from the other 3 sites (digital extensors [P<0.001], biceps femoris [P<0.001], and sartorius [P<0.001]). When readings from these 3 sites were combined (185 readings), a mean and SD of 92.9±7.4% and a 76–99% interval range with a 91% prediction interval was obtained.

THI readings were recorded for each StO2 level obtained. In this study, data points with a THI≤5 were as follows: 29 at the epaxial site, 0 at the digital extensor site, 3 at the biceps femoris site, and 4 at the sartorius site. Table 2 reports results obtained when data points with a THI≤5 are removed. The mean ±1 SD of measured StO2 levels by subpopulation (sex, BCS) are reported in Tables 3 and 4. There was no significant difference between readings in males versus females at any of the sites. When body condition score was analyzed in reference to the sartorius (medial thigh) site, there was a significant difference between normal BCS (5–5.5) and moderately to severely overweight (BCS>7) dogs (P=0.002), and a significant difference between normal BCS (5–5.5) and mildly overweight (BCS 6–6.5) dogs (P=0.04) with overweight dogs having lower readings. There was no significant difference between mildly (BCS 6–6.5) and moderately (BCS>7) overweight dogs (P=0.08) at the sartorius site. There was no significant difference in mean StO2 values by BCS at any of the other 3 sites when compared with one another (Table 5).

Table 2.   Measurement success and tissue spectroscopy results by site in clinically healthy dogs with THI ≤5 removed from data
Site measuredNumber of readings
(% of 78 attempted)
StO2 mean ± 1 SD (%)Prediction
StO2 interval
  1. Prediction interval is [(n−2x)/n] × 100 where n is the number of readings and x the number of values removed from smallest/largest values obtained. StO2 interval range reports 3rd (9th) lowest to 3rd (9th) highest value for each site (3 statistically similar sites combined).

  2. StO2: tissue oxygen saturation.

Epaxial (lower back)18 (23%)87.6 ± 8.778%77–94
Digital extensors (forearm)55 (70%)94.1 ± 5.993%82–99
Biceps femoris (lateral thigh)49 (63%)95.4 ± 4.092%90–99
Sartorius (medial thigh)74 (95%)92.9 ± 5.195%82–99
Statistically similar sites combined (digital extensors, biceps femoris, sartorius)178 (76% of 234 readings)93.9 ± 5.291%83–99
Table 3.   Measurement success and tissue spectroscopy results by site and sex subpopulation in clinically healthy dogs
Site measuredSubpopulation
Number of readings
(% of attempted)
StO2 mean ± 1 SD (%)P-value
  1. StO2: tissue oxygen saturation

Epaxial (lower back)Male (39)26 (67%)70.8 ± 21.30.77
Female (39)21 (54%)65.8 ± 24.0 
Digital extensors (forearm)Male (39)28 (72%)95.0 ± 6.10.87
Female (39)27 (69%)93.1 ± 5.8 
Biceps femoris (lateral thigh)Male (39)26 (67%)94.5 ± 6.50.74
Female (39)26 (67%)93.1 ± 8.4 
Sartorius (medial thigh)Male (39)39 (100%)92.5 ± 5.50.88
Female (39)39 (100%)90.3 ± 10.1 
Table 4.   Measurement success and tissue spectroscopy results by site and body condition score (BCS) subpopulation in clinically healthy dogs
Site measuredBody
of dogs)
Number of
(% of 78
StO2 mean ± 1 SD (%)
  • *

    P-value 0.04.

  • §

    P-value 0.002.

  • StO2: tissue oxygen saturation.

Epaxial (lower back)5–5.5 (25)13 (52%)73.2 ± 15.6
6.0–6.5 (41)28 (68%)66.6 ± 24.5
>7.0 (12)6 (50%)67.3 ± 26.7
Digital extensors (forearm)5–5.5 (25)18 (72%)94.7 ± 4.8
6.0–6.5 (41)31 (75%)93.7 ± 6.7
>7.0 (12)6 (50%)94.5 ± 6.0
Biceps femoris (lateral thigh)5–5.5 (25)16 (64%)95.8 ± 2.8
6.0–6.5 (41)29 (70%)93.2 ± 7.7
>7.0 (12)7 (58%)91.9 ± 12.9
Sartorius (medial thigh)5–5.5 (25)25 (100%)94.9 ± 3.1
6.0–6.5 (41)41 (100%)90.8 ± 7.6*
>7.0 (12)12 (100%)86.3 ± 13.2
Table 5.   Significance of differing StO2 levels at each site by body condition score (BCS)
  1. Significant values in bold (P<0.05).

BCS 5–5.5 versus BCS 6–6.50.38970.58340.26440.04
BCS 5–5.5 versus BCS>70.60160.95360.25080.002
BCS 6–6.5 versus BCS>70.94370.76160.67930.08


NIRS technology has been developed to allow for continuous monitoring of StO2 levels in various tissues. Unlike other oxygen measuring devices (pulse oximetry [SpO2], arterial blood gas [PaO2], pulmonary arterial catheter [SvO2]), NIRS measures hemoglobin saturated with oxygen in the microcirculation at the level of the tissues. To the authors' knowledge, the application of this technology to small animal clinical patients has not been investigated. The current study was undertaken to establish reference values for StO2 in clinically healthy dogs for future studies investigating the use of the device in ill veterinary patients.

Criteria for inclusion were selected to ensure study animals were adult dogs of good health (CBC, chemistry, physical exam) with appropriate perfusion ability (systolic blood pressure) and normal oxygenation of arterial blood (SpO2 or PaO2). Body weight (>9 kg) was selected to ensure that the light emitting and light sensing sites on the tissue spectrometer device would be over the same muscle body at all 4 selected sites for each dog. A 6 kg dog was evaluated (owner unaware of exclusion criteria) due to availability of resources. Data points gathered from this dog were excluded from the final analysis; however, the dog's StO2 values fell within the reference interval determined for each muscle body. The investigators have no reason to believe the device will not be applicable to animals weighing <9 kg. Further investigation in a group of smaller dogs will be necessary to verify this hypothesis. As this device has been developed for use in humans, the investigators believe this device would also be applicable to giant breeds that may not have been well represented in the population evaluated.

In humans, the thenar eminence (muscle body at the base of the hand, below the thumb) has been selected for obtaining StO2 readings in clinical situations. This site was selected due to its ease of identification, ease of probe placement, and consistency of the thickness of underlying tissue from patient to patient, regardless of size.5 An initial report using healthy human volunteers reported mean ±1 SD StO2 values of 88±5%.13 Crookes et al5 evaluated 707 human volunteers and reported mean ±1 SD StO2 values of 87±6%. A prospective cohort study found that a cutoff StO2 of 75% optimized sensitivity and specificity in identifying poor perfusion and predicting the development of multiple organ dysfunction syndrome or death in severe torso trauma patients.3 This cutoff of 75% approximates 2 SDs from the mean in both of the initial volunteer studies.

Data reported here include mean ±1 SD, as well as interval ranges (Table 1). The tissue spectrometer reports a percentage of saturated versus unsaturated hemoglobin in the tissue microvasculature, limiting upper values to 100%. As a result, our data are not distributed evenly around a mean (ie, Gaussian), but has a skew to the left. Based on statistician consultation (Weisberg), interval ranges that evenly remove data points from each end of the distribution curve are also reported (Table 1). As a point of interest, when the 3 statistically similar sites are combined, the 9th lowest data point (76%) is very close to 2 SDs from the mean (78%). As reported above, in human clinical trials, 2 SDs from the mean approximates the cutoff for clinically significant readings. Future prospective trials will be necessary to determine if this applies clinically to canine patients as well.

In humans, the THI is an indicator of signal strength and StO2 values are not considered accurate at THI levels≤5.1 This finding has not been verified in canine and porcine models evaluating THI and StO2 readings.14 However, in Table 2, we report data points by site with a THI≤5 removed. The epaxial site has the most readings removed (n=23), reinforcing the conclusion that this site is not recommended for clinical application of the device. Additionally, when the 3 statistically similar sites are combined, the mean ±1 SD reported is not clinically different (92.9±7.4% versus 93.9±5.2%).

In our study, 3 of the muscle bodies (sartorius, digital extensors, biceps femoris) were selected based on size and accessibility in most clinical veterinary patients. The epaxial muscle body was selected for data collection as a negative control, as it is a site frequently associated with fat deposition. The difficulty in acquisition of a data point (60% of readings) and significantly lower readings (mean 68.5%) with a large SD (22.4%) of data gathered at the epaxial location suggest that this is an inappropriate site for obtaining StO2 levels in dogs. Analysis of results obtained from the other 3 sites revealed no significant difference in readings (sartorius, digital extensors, biceps femoris). Additionally, results from each site had comparable reference intervals, and were associated with a narrow SD (5.9–8.1%). When the 185 readings obtained from the digital extensors, sartorius, and biceps femoris sites were combined and analyzed, a mean ±1 SD of 92.9±7.4% was obtained. This result will be used as a reference range for further studies evaluating StO2 levels in clinically ill dogs. One suggestion for the difference in reference range between humans and dogs is the relative activity and oxygen use of the muscles used in dogs, compared with the less metabolically active thenar eminence in humans.

When subpopulation analysis was performed, there was a statistically significant difference between normal weight dogs (BCS 5–5.5) and mildly to severely overweight dogs (BCS 6–6.5 and BCS>7) at the sartorius site, with overweight dogs having lower readings. Whether these differences have clinical significance will need to be established in future studies. There was no significant difference in BCS subgroups noted in readings when compared with the other 3 sites. However, the epaxial site has quite large reference intervals and SDs, further supporting the recommendation that this site is not ideal for obtaining measurements in clinical patients. Reading failures (no number displayed on device) at the digital extensor and biceps femoris sites were more commonly associated with obese patients (BCS>7) and may indicate that fat interferes with the ability to obtain a reading at these sites, but not with the reading itself if one is obtained (Table 4).

Because of ease of positioning, limited hair, and minimal skin pigment, the medial thigh (sartorius) is recommended as the ideal site for obtaining StO2 readings. While the digital extensors and biceps femoris sites had comparable reference intervals, readings were less consistently obtained from these sites. In our study, objective data on pigmentation of the dogs were not collected and reported; however, subjectively, the investigators noted that dark pigmentation seemed to interfere with the ability to obtain a reading. In the absence of objective data, the authors also noted that the skin in the area of the sartorius muscle (medial thigh) was rarely pigmented.

In human studies, there was a statistical, but not clinically significant, difference between male and female thenar eminence StO2 levels (male 88.38±5.51%; female 85.40±6.69%) of 707 volunteers.5 In our study, there was no statistical difference between males and females at any of the sites. This difference may be associated with the more profound hormonal differences in humans whereas a majority of our study dogs were neutered (96%).

This study evaluated the use of a noninvasive, continuous readout, portable NIRS technology to enable measurement of StO2 levels for clinical use in veterinary medicine. Reference ranges for StO2 levels in clinically healthy dogs were established. Limitations to this study include weight restrictions (dogs<9 kg excluded, few giant breeds included) and lack of inclusion of a geriatric population. Future studies should be performed to analyze the use of this technology in dogs with clinical illness resulting in decreased tissue perfusion. If findings in clinical veterinary patients parallel the promising results found in human literature, this device could become useful in guiding management of trauma, shock, sepsis, and other critical illnesses in dogs.


Special thank you to Dr. Carl Osborne for his financial, emotional, and intellectual support of this project; Dora Schroeder, CVT, VTS (ECC) for her expertise and assistance in gathering data; Dr. Leslie Sharkey for her mentorship and guidance; Sanford Weisberg, School of Statistics, for his help with the statistical analysis.


aCell-Dyn 3500, Abbott, Chicago, IL.

bOlympus AU400, Olympus America Inc, Center Valley, PA.

cParks Medical Model #811B, Parks Medical, Aloha, OR.

dNonin Model #8500AV, Nonin Inc, Plymouth, MN.

ei-STAT 1 analyzer, CG4+cartridge, Abbott Laboratories, East Windsor, NJ.

fIdexx 4Dx, Idexx, Westbrooke, ME.

gPortex Arterial Blood Sample Syringe, Smiths Medical ASD Inc, Keene, NH.

hInSpectra Tissue Spectrometer, Hutchinson Technology Inc, BioMeasurement Division, Hutchinson, MN.

iJMP 7.0, 2007, SAS Institute Corp, Cary, NC.