Intra- and inter-session reliability and repeatability of an infrared thermography device designed for materials to measure skin temperature of the triceps surae muscle tissue of athletes

Study design

The present study was carried out from January 2020 to May 2021 according to The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) criteria (Appendix 1). The Helsinki Declarations, as well as specific human experimentation ethical rules, were taken into account. The ethics committee for clinical interventions of the Hospital Clínico San Carlos, Madrid (Spain) approved this study with internal code number 20/021-E on January 20, 2020. Before the study began, all participants signed the informed consent form. In addition, the present study was supported by Contract 83 between Complutense University and PCE Ibérica S.L. (Reference number: 6-2020), providing a specific grant for this research project.

Sample size

The sample size determination was calculated by bi-variate correlations statistical procedures through G*Power 3.1.9.2 program (G*Power^©, from Dusseldorf University, in Germany), considering a 0.4 correlation coefficient to achieve a moderate correlation (Lobo et al., 2016) between infrared thermography measurements, applying a 1-tailed hypothesis, a 0.05 α error and a 0.80 power. Lastly, the sample size was 34 triceps surae muscles to achieve the required thermography measurements for a 0.801 actual power.

Participants

Thirty-four triceps surae muscles were bilaterally analyzed from 17 healthy athletes considering a consecutive sampling recruitment procedure. Inclusion criteria comprised specifically healthy athletes aged 18–65 years providing the consent information document previously, carrying out sports activities and training for at least 2 h as well as 1 day per week, with moderate (level-II) or vigorous (level-III) intensities for physical activity with metabolic equivalent indexes greater than 600 METs/min/week, measured by the International Questionnaire for Physical Activity (IPAQ) (Roman-Viñas et al., 2010).

Exclusion criteria included muscle soreness, congenital dysfunctions, neuromuscular conditions, rheumatic alterations, body mass index (BMI) greater than 31 kg/m², previous neurological conditions, prior surgeries and skin pathologies. Some alterations in the lower limbs region (i.e., chronic ankle instability presence, prior sprains or previous fractures) were also excluded according to the thermographic influence of compartmental, stress, inflammation, and perfusion-based conditions (Sanchis-Sánchez et al., 2014; Ramirez-GarciaLuna et al., 2022). Lastly, difficulties or inability to carry out the procedure to complete the study course, explained below, were considered as exclusion criteria.

Procedure

Intra- and inter-session reliability and repeatability of the skin temperature of the triceps surae muscles were bilaterally assessed by the Manual Infrared Camera PCE-TC 30 to determine measurement agreement and concordance at the same day separated by 1 h (considered as intra-day measurements) and alternate days separated by 48 h (considered as inter-day measurements), respectively. Indeed, participants were asked to continue with their daily life and physical exercise routine (avoiding unusual efforts or activity changes) between measurements and not taking prescribed medications at the prior week nor vasomotor substances (i.e., caffeine) on the same measurement day, as well as heavy metals were not allowed. A period of 5 min of acclimatization of the subjects to the room was applied (Fig. 1). All measurements were assessed with patients standing up in a relaxed position in the same room within a 24.1 °C ± 1 °C temperature and a 45% ± 10% humidity, without direct ventilation flow toward examiners or participants (Rodríguez-Sanz et al., 2017).

Descriptive data

Descriptive data including sex (categorized as male or female), age (measured in years), height (measured in cm), weight (measured in kg), and BMI (expressed as kg/cm² following the Quetelet’s index) (Garrow, 1986), main sports category (divided into fitness considered as bodybuilding exercise or soccer), side (categorized as right or left), and dominance (expressed as yes or no), and smoker (expressed as yes or no) were detailed (Calvo-Lobo et al., 2019). As a tool with adequate psychometric properties, metabolic equivalent index per minute per week (METs/min/week) was evaluated by the IPAQ to determine physical activity and its categorization as moderate (600–1,500 METs/min/week) and vigorous (≥1,500 METs/min/week) physical activity (Gauthier, Lariviere & Young, 2009).

Infrared thermography

We used a Manual Infrared Camera PCE-TC 30 (PCE Instruments UK Ltd, Southampton, United Kingdom), which displayed a sensor resolution of 80 × 80, a measurement range from 0 °C to 250 °C, a display of 230 × 240 pixels, a thermal sensitivity of 80 mK and an 8-mm lens (De Villoria et al., 2011; Pérez-Urrestarazu et al., 2014). The infrared thermography imaging process was performed with the participant standing up in a relaxed position 1 m from the camera. Bilaterally, the triceps surae complex, including lateral (Figs. 1A and 1B) and medial (Figs. 1C and 1D) gastrocnemius, as well as soleus (Figs. 1E and 1F) muscles, was measured by a region of interest (ROI) through five infrared thermography images for each muscle. Removing the highest and lowest values, the mean of the three measurements was used for data analysis.

Infrared images and data were analyzed by a blinded and experienced evaluator using the Guide™ Report Express (PCE Instruments UK Ltd, Southampton, United Kingdom) (Rodríguez-Sanz et al., 2017). This software provided the mean thermal value (°C) of the selected ROI of 1 cm² coinciding with the center of a landmark for each muscle (Fig. 2). These landmarks were used to determine superficial skin temperature and placed superior to the Achilles tendon for the soleus muscle and in the thickest part of the medial and lateral gastrocnemius muscles according to prior similar studies (Benito-de-Pedro et al., 2019; Rojas-Valverde et al., 2021).

Statistical analyses

The 24.0 Statistical Package Program for Social Science (named SPSS, from IBM-Corp, in Armonk, NY, USA) was used for data analyses. The α error was set at 0.05, and thus a P-value lower than 0.05 was considered statistical significance. The Kolmogorov-Smirnov statistical test and visual inspection of histograms were considered to detail normality distribution. Data adjusted to normal distribution were detailed through means ± standard deviations (SD) in conjunction with the upper and lower limits of 95% confidence interval (CI). Data adjusted to non-normal distribution were detailed through medians ± interquartile ranges (IR). Infrared thermography measurements for intra- and inter-session evaluations were compared through paired-sample Student t-tests considering parametric tests and Wilcoxon tests regarding non-parametric tests. ICC analyzed the reliability and repeatability between each pair of measurements for bidirectional absolute agreement and Pearson (r) or Spearman (ρ) correlation coefficients as parametric or non-parametric tests, respectively. Furthermore, ICC_(2,1) values were specifically interpreted as poor for <0.40 ICC_(2,1), weak for 0.40–0.59 ICC_(2,1), good for 0.60–0.74 ICC_(2,1), and excellent for 0.75–1.00 ICC_(2,1) (Calvo-Lobo et al., 2019).

Next, correlation coefficients were specifically interpreted as weak for 0.00–0.40 r or ρ, moderate for 0.41–0.69 r or ρ, and strong for 0.70–1.00 r or ρ (Lobo et al., 2016). Standard errors for measurements (SEM) values were detailed through SD × $\sqrt{\left(1-ICC\right)}$. After, minimum detectable changes (MDC) values were detailed through $\sqrt{2}\times 1.96\times SEM$ for 95% CI. Both MDC and SEM were detailed through Bland and Altman recommendations (Calvo-Lobo et al., 2019). Limits for agreement (LoA) for each pair of measurements were detailed through differences means $\pm 1.96\times SD$ for 95% CI in line with Bland and Altman (Bland & Altman, 2010; Calvo-Lobo et al., 2019).

In addition, Bland-Altman plots were shown to detail visual agreements for each pair of measurements showing systematic measurement errors of the differences in means distributions for each pair of measurements located at the Y-axis with regards to the means for each pair of measurements located at the X-axis. These Bland-Altman plots were shown in conjunction with linear regression models. R² coefficients were calculated to detail the adjustment quality. The mean values for each pair of measurements were considered independent variables. Lastly, the differences for each pair of measurements were considered dependent variables (Bland & Altman, 2010).

Descriptive data

The final sample comprised 34 triceps surae muscles bilaterally from 17 healthy athletes, nine (52.9%) males and eight (47.1%) females, with mean ± SD (95% CI) age of 41.76 ± 14.42 (36.73–46.79) years, the weight of 68.57 ± 14.57 (63.40–73.64) kg, the height of 1.69 ± 0.09 (1.66–1.73) m, and BMI of 23.45 ± 3.47 (22.24–24.66) kg/m². Regarding the main sports category, these athletes performed in fitness (n = 14; 82.40%) and soccer (n = 3; 17.6%). All athletes presented the dominant right side (n = 17; 100%); most were non-smokers (n = 12; 70.60%). Considering the IPAQ, the mean ± SD (95% CI) of metabolic equivalents index per minute per week was 3,276.08 ± 1,876.49 (2,621.34–3,930.82) METs/min/week, including eight (47.10%) athletes who performed vigorous physical activity and nine (52.90%) athletes who performed moderate physical activity. Table 1 shows the normality statistics and significance according to the Kolmogorov-Smirnov test.

Intra-session reliability and repeatability

According to Table 2, intra-session measurements of the infrared thermography device designed for materials (Manual Infrared Camera PCE-TC 30) in the triceps surae muscle tissue of athletes showed excellent reliability (ICC_(1,2) = 0.969–0.977), measurement errors (SEM = 0.186–0.212 °C; MDC = 0.515–0.587 °C) and did not present any statistically significant systematic error of measurements (P > 0.05).

Regarding intra-session measurements, bivariate correlations were excellent for the soleus (r = 0.953; P < 0.001), medial gastrocnemius (ρ = 0.885; P < 0.001), and lateral gastrocnemius (r = 0.939; P < 0.001).

In addition, Bland-Altman plots presented an adequate agreement for intra-session measurements of the infrared thermography device designed for materials (Manual Infrared Camera PCE-TC 30) in the soleus (Fig. 3), medial gastrocnemius (Fig. 4) and lateral gastrocnemius (Fig. 5) muscles of athletes, due to visual distributions of the difference means for each pair of measurements at Y axis concerning the mean for each pair of measurements at X-axis did not present any systematic measurement error and most thermographic measurements were between the upper and lower LoA. In addition to Bland-Altman plots, linear regression models did not show any statistical significance for soleus (R² = 0.000; β = 0.003; F_1,32 = 0.003; P = 0.954), medial gastrocnemius (R² = 0.002; β = −0.017; F_1,32 = 0.065; P = 0.800), and lateral gastrocnemius (R² = 0.019; β = −0.050; F_1,32 = 0.626; P = 0.435) intra-session measurements.

Inter-session reliability and repeatability

According to Table 3, inter-session measurements of the infrared thermography device designed for materials (Manual Infrared Camera PCE-TC 30) in the triceps surae muscle tissue of athletes presented excellent reliability (ICC_(1,2) = 0.956–0.974), measurement errors (SEM = 0.187–0.232 °C; MDC = 0.518–0.643 °C), and did not present any statistically significant systematic error of measurements (P > 0.05).

Considering inter-session measurements, bivariate correlations were also excellent for the soleus (r = 0.949; P < 0.001), medial gastrocnemius (r = 0.914; P < 0.001), and lateral gastrocnemius (r = 0.938; P < 0.001).

Lastly, Bland-Altman plots showed an adequate agreement for inter-session measurements of the infrared thermography device designed for materials (Manual Infrared Camera PCE-TC 30) in the soleus (Fig. 6), medial gastrocnemius (Fig. 7), and lateral gastrocnemius (Fig. 8) muscles of athletes, since visual distributions of the difference means for each pair of measurements at the Y axis concerning the mean for each pair of measurements at the X axis did not present any systematic measurement error, and most thermographic measurements were between the upper and lower LoA. In conjunction with the Bland-Altman plots, linear regression models did not show any statistical significance for soleus (R² = 0.014; β = 0.039; F_1,32 = 0.463; P = 0.501), medial gastrocnemius (R² = 0.000; β = −0.009; F_1,32 = 0.014; P = 0.907), and lateral gastrocnemius (R² = 0.019; β = −0.050; F_1,32 = 0.621; P = 0.436) inter-session measurements.

Future studies

Further studies should be designed as randomized clinical trials to determine if these asymmetries reference cut-off values could prevent triceps surae muscle injuries (Côrte et al., 2019). According to prior studies (Hiemstra et al., 2007; Chung et al., 2015), the uninvolved normal side after injury may often be not normal, i.e., presenting temperature values different from healthy subjects, and cut-off values should also be detailed in the future muscle recovery studies. In addition, thermographic measurement of the triceps surae with this device should be analyzed by intra- and inter-rater reliability determining SEM and MDC values and correlated with the other high-end infrared devices as possible gold standards such as ThermoHuman® (Fernandez-Cuevas et al., 2017; Requena-Bueno et al., 2020), IRTIS-2000® (Zaproudina et al., 2008) and Thermofocus® (Petrova et al., 2018) tools. Furthermore, correlations with electromyography measurements of the triceps surae muscle activity should be carried out (Rodriguez-Sanz et al., 2019). Lastly, the intramuscular temperature should also be correlated with this device to determine concurrent validity concerning a gold standard (Burnham, McKinley & Vincent, 2006).

Limitations

Various limitations should be considered for the use of this thermographic device. First, the MDC was superior to the lower limbs asymmetry reference range from 0.3 °C to 0.4 °C proposed as a cut-off for following-up before a preventive protocol (Côrte et al., 2019) and therefore, this device should not be used for values lower than 0.5 °C–0.6 °C.

Second, the concurrent validity of this device has not been performed for human tissue temperature, and this validity should be assessed in the future (Burnham, McKinley & Vincent, 2006).

Third, our sample only comprised healthy athletes, and further studies should analyze skin temperature with this tool over injured muscle tissue (Alburquerque Santana et al., 2022).

Fourth, our sample size calculation was accurately detailed to determine a moderate bivariate correlation between measurements, but our sample size was low to achieve the actual power to perform comparisons classifying groups depending on sport category, BMI, and other valued characteristics. In addition, despite statistical analyses were carried out according to our prior sample size calculation model to detail bivariate comparisons for intra- and inter-session measurements, future nested type studies should be designed as nested statistical models such as analyses of variance (ANOVA) to determine more accurate temperature comparisons.

Finally, in spite of the device was calibrated according to the manufacturer, the fact that the PCE-TC 30 thermography device was not designed for skin temperature assessment could affect the repeatability of the measurement to a lesser extent than its accuracy. Nevertheless, a comparison between the temperature assessed by a validated thermal camera and the PCE-TC 30 was not reported. Future studies should evaluate its accuracy due to a possible wrong estimation of the absolute skin temperature. Procedures for thermographic assessment in sports and exercise sciences have been reviewed by Moreira et al. (2017) in a consensus statement of the experts in the field. However, our reliability study aimed to standardize the proposed procedure and contributed to improving the methods behind measures.