Preliminary Analysis of Skin Temperature Asymmetries in Elite Young Tennis Players

Abstract

This preliminary study aimed to assess skin temperature (Tsk) asymmetries before and after on-court training in elite young tennis players and to find the correlation between these asymmetries and demographic data, racquet characteristics, and pain and fatigue variation. Thermal images of nine tennis players were taken before and after two and a half hours of standardised training. Thermal asymmetries were correlated to age, years of experience, racquet weight, string tension, body mass index (BMI), and variation in fatigue and pain. In contralateral comparison, Tsk was higher on the dominant side in areas such as the anterior (1.1 ± 0.5 °C, p < 0.001, ES = 0.5) and posterior forearm (1.1 ± 1 °C, p < 0.01, ES = 0.5). Concerning pre- and post-comparisons, post-training Tsk values were lower in all regions except in the posterior forearm, posterior shoulder, and posterior leg. Finally, significant correlations were found between thermal asymmetry and weight of the racquet, body mass index, fatigue and pain variation. Monitoring tennis players’ Tsk with infrared thermography (IRT) gives coaches vital information to evaluate contralateral asymmetries and technical pattern activations during training sessions.

1. Introduction

Infrared thermography (IRT) is a non-invasive, rapid, safe, and objective method to measure skin temperature (Tsk) through body radiation [1]. This technology has been applied in sports science in different areas, such as injury prevention, thermo-physiology, sports wearing or technical pattern studies [2,3]. Focusing on asymmetries, researchers have found that contralateral thermal asymmetries over 0.5 °C can be related to injuries, considering physiological dysfunctions [4,5]. Moreover, thermal asymmetries are used to monitor muscular involvement in sports technical actions in symmetrical and asymmetrical sports [1].

Tennis is an asymmetrical sport in which repetitive movements, such as service or groundstrokes, are executed repeatedly during the game [6]. In the upper limbs, significant differences in anthropometric parameters (muscle circumferences and bone diameters) between the dominant and non-dominant arms have been found [7]. As a result, musculoskeletal disorders may appear, causing injuries such as tennis elbow, muscle strain or shoulder injuries, among others [6]. To the best of the author’s knowledge, few studies have applied infrared thermography to address thermal asymmetries in tennis players.

Thermal asymmetries in the upper limbs associated with the sports technique were found in handball and archery [1,8]. In archery, Tsk was higher on the side where the bow was placed due to greater muscle activation to maintain the shooting position, and only 15% of the athletes presented asymmetry higher than 0.5 °C [8]. In handball, significantly higher temperatures were found in the dominant arm but without exceeding the established threshold of 0.5 °C [1,4]. In the case of symmetrical sports such as scullers or running, no significant differences were found in the Tsk symmetrical area comparisons [1,2].

This preliminary study aimed to assess Tsk asymmetries before and after on-court training in elite young tennis players and to analyse the correlation between these asymmetries and demographic data, racquet characteristics, and pain and fatigue variation. We hypothesised that tennis players’ Tsk would present thermal asymmetries in post-training measurements and higher Tsk values in the dominant arm due to the technical characteristics of the sport.

2. Materials and Methods

Nine elite tennis players (4 male and 5 female) voluntarily participated in this study: 21 ± 5 years, 67.4 ± 6.5 kg body mass, 1.77 ± 0.07 m height, 21.4 ± 1.3 kg/m² body mass index (BMI), 14 ± 4 years of experience, 299 ± 9 g racket weight, and 23 ± 2 kg string tension. Participants’ levels were national and international according to the competitions they participated in (Junior and Senior International Tennis Federation Competitions and Women’s Tennis Association Tour). These players train 5 to 6 days per week, with a daily distribution of 3 ± 1 h of on-court training and 1.5 h of physical preparation. All participants signed the informed consent form before starting the study. In the case of minors, the parent or legal guardian signed the consent. The study procedures complied with the Declaration of Helsinki and were approved by the local university’s ethics committee.

2.1. Procedures and Experimental Design

An experimental study with pre- and post-measurements using IRT was conducted (Figure 1). Specifically, measurements were concerted on three different training days over four weeks (14.6 ± 2.6 °C temperature; 45 ± 8% relative humidity). All the players during training wore the same type of clothing, a T-shirt and shorts with a polyester composition. Four images were taken before and after a standardised training session of 2.5 h (Figure 1). Variations in fatigue and soreness were calculated by completing a 15 cm visual analogue scale before and after training (0 was no fatigue/soreness, and 15 was maximum fatigue/soreness) [9]. Personal performance variables such as age, years of experience, racquet weight, string tension and body mass index (BMI) were obtained by filling in a questionnaire.

2.2. Skin Temperature Assessment

The design of the study considers training in court (outside) and thermal image acquisition in an indoor space. The reasons for this design were, firstly, that we assessed high-level players and we did not want to interfere in their daily training routine. Secondly, the thermographic images were taken in an indoor space to control the variables influencing the data acquisition (e.g., environmental conditions, sources of infrared radiation, etc.) to have a reproducible environment.

Tsk measurements were taken in a conditioned room near the training courts. The room’s ambient temperature (20.1 ± 0.5 °C) and relative humidity (47 ± 2%) were always controlled (Figure 2). The images were taken with athletes in anatomical positions at 2.5 m for a correct framing of the areas under study. Thermal images were taken with a FLIR E60bx infrared camera (FLIR Systems, Wilsonville, OR, USA), with a resolution of 320 × 240 pixels, a measurement uncertainty of 2 °C and a sensitivity of <0.05 °C.

Participants were informed 24 h before the measurements of the protocol requirements for IRT interventions. They were asked to avoid: (1) coffee, tea, tobacco, alcohol, or medicines consumption 12 h before the test; (2) heavy meals 2 h before the test; (3) body creams and sprays; and (4) therapeutic treatments within at least 24 h before the test [8,10]. Furthermore, they were required to come 15 min before the training session. For the skin acclimatisation to room temperature, participants went 10 min without exercising with their bodies uncovered (boys without shirts and shorts pulled up; girls with sports tops and shorts pulled up) [11] (Figure 2).

Regarding the image processing, regions of interest (ROI) were determined in each of the four images taken before and after the training session (Figure 3). The ROIs were obtained with FLIR ThermaCAM Researcher Pro software (version 2.10, FLIR Systems, Wilsonville, OR, USA). A 0.98 emissivity was used, and the mean temperature was obtained for each ROI. These results calculated the asymmetries between the dominant and non-dominant side (dominant minus non-dominant) and the variation of thermal asymmetry between pre-training and post-training (post minus pre). The dominant or non-dominant side was established for the upper and lower limbs, according to which was the preferred and most-used arm during training.

Regarding the image processing, regions of interest (ROI) were determined in each of the four images taken before and after the training session (Figure 3). ROIs were obtained by the same evaluator, increasing their reproducibility and considering anatomical proportions and anatomical points visible in the images. Considering the thermal resolution of the camera, all the ROIs had a considerable number of pixels: greater than 25, which is the minimum indicated by ISO regulations (TR 13154:2009 ISO/TR 8-600).

2.3. Statistical Analyses

IBM SPSS Statistics 23 (IBM, Armonk, NY, USA) was used for statistical analysis. First, the Shapiro–Wilk test was applied to confirm the normality distribution of temperature data (p > 0.05). After this, differences in dominance before and after training were analysed using the Student’s t test for related samples. The same analysis was performed to evaluate the differences between the mean temperature of the different ROIs before and after training. Furthermore, data normality distribution of fatigue and pain variation was confirmed with the Shapiro–Wilk test in all values except for in the knees, the shoulder, and the arms (p < 0.05). Student’s t test for related samples was used for those with normal distribution. Meanwhile, the Wilcoxon test was applied for those with a non-normal distribution. For significant pair differences, Cohen’s effect size (ES) was calculated and classified as small (ES = 0.2–0.5), moderate (ES = 0.5–0.8), or large (ES > 0.8) [12]. Finally, a bivariate analysis was carried out using Pearson’s correlation coefficient to deepen the understanding of asymmetry Tsk variation of each ROI, considering age, racquet weight, string tension, years of experience, BMI and variation in fatigue and general pain (p < 0.05). Significant correlations were classified as weak (0.2 < ∣r∣ < 0.5), moderate (0.5 ≤ ∣r∣ < 0.8), or strong (∣r∣ ≥ 0.8) [13].

3. Results

3.1. Dominant and Non-Dominant Comparison

As shown in Figure 4, concerning the upper body, significant differences were only found in the forearms (anterior and posterior) and the posterior arm. In the forearms, differences appeared during post-training, while in the posterior arm, differences appeared during pre-training. In both cases, the temperature of the dominant side was higher than in the non-dominant side (anterior forearm 1.1 ± 0.5 °C, p < 0.001, ES = 0.5; posterior forearm 1.1 ± 1 °C, p < 0.01, ES = 0.5; posterior arm 0.3 ± 0.4 °C, p = 0.05, ES = 0.2). Concerning the lower body, no significant differences were found between the dominant and non-dominant sides, neither before nor after training (Figure 4).

3.2. Pre-Training and Post-Training Comparison

Regarding the analysis in Figure 4, only three regions did not have significant differences between pre- and post-training on either side. These regions were the posterior forearm, posterior shoulder, and posterior leg. Significant differences were observed between the pre-dominant and non-dominant sides in the rest of the regions. More specifically, it can be seen how, on both the dominant and non-dominant side, the temperature was lower after training (as an example, the differences in the dominant side were: anterior forearm −1.2 ± 1.5 °C, p = 0.04, ES = 0.7; anterior arm −1.6 ± 1.2 °C, p < 0.01, ES = 1.0; posterior arm −1.3 ± 1.1 °C, p < 0.01, ES = 0.8; anterior shoulder −0.9 ± 1.1 °C, p = 0.04, ES = 0.7; abdominal −1.8 ± 1.6 °C, p = 0.01, ES = 1.1; lumbar −1.4 ± 1.1 °C, p = 0.04, ES = 1.1; anterior thigh −1.5 ± 1.5 °C, p = 0.02, ES = 0.8; posterior thigh −1.5 ± 0.8 °C, p = 0.01, ES = 1.0; anterior knee −1.3 ± 1.2 °C, p = 0.01, ES = 1.3; posterior knee −1.2 ± 1.1 °C, p = 0.01, ES = 0.9; anterior leg −1.0 ± 1.2 °C, p = 0.04, ES = 0.9).

3.3. Correlation Analysis

Players completed a visual analogue scale indicating their pre- and post-training fatigue and pain status (

Table 1. Pre- and post-training fatigue and pain data. Mean (SD). 0 (no pain/fatigue); 15 (very painful/fatigued).

	Pre-Training (SD)	Post-Fatigue (SD)	Pre vs. Post (p Value)
Fatigue perception
General	4.6 (2.1)	6.7 (3.1)	0.01
Trunk	3.3 (2.3)	4.3 (2.9)	0.05
Shoulder	2.7 (2.0)	4.6 (2.8)	0.01
Arms	2.9 (2.1)	5.4 (3.8)	0.04
Gluteus	2.9 (2.6)	4.0 (3.4)	0.03
Thigh	3.3 (2.5)	5.1 (3.1)	0.02
Knee	1.2 (2.1)	2.0 (2.6)	0.01
Leg and Feet	3.2 (2.3)	6.0 (3.9)	0.01
Pain perception
General	2.5 (1.4)	2.1 (1.3)	0.39
Trunk	2.2 (2.0)	2.1 (1.9)	0.66
Shoulder	1.4 (1.8)	2.6 (3.3)	0.06
Arms	1.5 (2.2)	1.3 (1.7)	0.44
Gluteus	1.9 (1.4)	2.2 (1.2)	0.19
Thigh	1.8 (1.7)	1.4 (1.0)	0.27
Knee	1.3 (1.8)	1.5 (2.2)	0.25
Leg and Feet	1.1 (1.0)	1.5 (1.4)	0.10

). Significant post-training fatigue increase was found in all regions (p < 0.05). In contrast, no variation in pain perception was observed.

Furthermore,

Table 2. Pearson’s correlation coefficient (r) and p value (p) between each ROI’s variation (Δ) in thermal symmetry and age, racket weight, string tension, years of experience, body mass index (BMI) and variation in fatigue and pain. Significant correlations are highlighted in bold letters.

	Age		Racket Weight		String Tension		Years of Experience		BMI		Δ General Fatigue		Δ General Pain
	r	p	r	p	r	p	r	p	r	p	r	p	r	p
Δ Anterior Forearm	0.39	0.30	−0.01	0.99	−0.28	0.47	0.28	0.47	−0.27	0.49	−0.19	0.62	0.76	0.85
Δ Posterior Forearm	−0.11	0.79	0.55	0.12	−0.39	0.30	−0.21	0.59	−0.25	0.52	0.62	0.77	−0.87	0.82
Δ Anterior Arm	0.48	0.19	0.71	0.03	−0.51	0.16	0.05	0.91	−0.42	0.26	0.39	0.30	0.13	0.7
Δ Posterior Arm	0.11	0.78	0.61	0.08	−0.23	0.56	0.05	0.90	−0.39	0.30	0.81	0.01	−0.52	0.90
Δ Anterior Shoulder	−0.17	0.66	0.66	0.06	−0.48	0.20	−0.34	0.37	−0.68	0.04	0.44	0.23	0.38	0.31
Δ Posterior Shoulder	−0.21	0.59	0.55	0.13	−0.37	0.32	−0.28	0.46	−0.25	0.52	0.81	0.01	−0.11	0.77
Δ Abdominal	−0.33	0.39	0.48	0.19	−0.43	0.25	−0.33	0.39	−0.30	0.43	0.19	0.62	0.07	0.87
Δ Lumbar	−0.36	0.34	0.70	0.04	−0.28	0.46	−0.34	0.37	−0.42	0.26	0.51	0.16	0.30	0.44
Δ Anterior Thigh	−0.01	0.98	−0.15	0.70	−0.53	0.14	0.21	0.59	0.37	0.33	0.11	0.79	0.11	0.79
Δ Posterior Thigh	−0.47	0.20	0.06	0.87	0.53	0.15	−0.50	0.17	−0.27	0.48	0.42	0.26	−0.38	0.31
Δ Anterior Knee	0.37	0.33	0.23	0.55	−0.26	0.50	0.58	0.10	0.59	0.09	0.41	0.28	−0.50	0.17
Δ Posterior Knee	0.29	0.45	−0.95	0.81	−0.20	0.61	0.43	0.25	0.81	0.01	−0.8	0.83	−0.81	0.01
Δ Anterior Leg	−0.33	0.39	−0.34	0.37	0.05	0.90	−0.14	0.71	−0.16	0.69	−0.53	0.14	0.75	0.02
Δ Posterior Leg	0.03	0.94	0.56	0.12	−0.33	0.38	0.14	0.73	0.30	0.43	0.33	0.39	−0.39	0.30

shows the results of the correlations between thermal symmetry variation and age, racket weight, string tension, years of experience, BMI, general fatigue and pain changes. The most significant results are shown in bold (Table 2), and in

Table 3. Mean variation (Δ) of thermal symmetry and standard deviation (SD) for each region analysed.

ROI	Δ Thermal Symmetry (SD)
Anterior Forearm	0.8 °C (0.5)
Posterior Forearm	0.7 °C (1.1)
Anterior Arm	0.1 °C (0.6)
Posterior Arm	0.0 °C (1.2)
Anterior Shoulder	0.2 °C (0.5)
Posterior Shoulder	0.0 °C (0.5)
Abdominals	0.0 °C (0.4)
Lumbar	−0.2 °C (0.4)
Anterior Thigh	0.0 °C (0.3)
Posterior Thigh	0.0 °C (0.4)
Anterior Knee	−0.2 °C (0.4)
Posterior Knee	0.0 °C (0.3)
Anterior Leg	−0.1 °C (0.2)
Posterior Leg	0.0 °C (0.3)

, the mean variations of the thermal symmetries found in each ROI are expressed. The correlations found were:

• Racket weight and variation in thermal symmetry in the anterior arm (p = 0.03, r = 0.71) and lower back (p = 0.04, r = 0.70), both positively and moderately correlated.
• BMI correlates negatively with symmetry variation in anterior shoulder temperature (p = 0.04, r = −0.68) but positively with posterior knee temperature (p < 0.01, r = 0.81). The former was a moderate correlation, while the latter was strong.
• The variation in overall fatigue correlates positively with the variation in temperature symmetry in the arm (p < 0.01; r = 0.81) and posterior shoulder (p < 0.01; r = 0.81) regions. Both correlations were strong.
• Overall pain variation and thermal symmetry variation in the posterior knee (p < 0.01; r = −0.81) and anterior leg (p = 0.02; r = 0.75) show a correlation. The first was negative and strong, while the second was positive and moderate.

4. Discussion

This study aimed to assess Tsk asymmetries before and after on-court training in elite young tennis players and to find the correlation between these asymmetries and demographic data, racquet characteristics, and pain and fatigue variation. The main findings demonstrated that in the contralateral comparison, there is a thermal asymmetry with higher values in the anterior and posterior forearm of the dominant arm at the end of the training session. Concerning the pre- and post-comparison, all the studied regions showed lower Tsk values in the post-training data, except in the posterior forearm, posterior shoulder, and posterior leg, where no significant differences were found. Finally, thermal asymmetries correlate with racquet weight, body mass index and variation in fatigue and pain. We will interpret these data with the existing literature, emphasising their practical applications and future research directions.

4.1. Dominant and Non-Dominant Comparison

Few studies use IRT in tennis, but some studies analyse Tsk in symmetrical and asymmetrical sports, comparing dominant and non-dominant sides [1,2,14]. Tennis has some characteristics that make it asymmetrical [7]. Our hypothesis that higher temperature values would be observed on the dominant side was partially accepted, as asymmetries were observed for the anterior and posterior forearm (after training) and the posterior arm (before training). However, no significant differences were found between both sides in the remaining regions.

At baseline, asymmetries higher than 0.5 °C may indicate physiological dysfunctions [2,11]. In this study baseline, a difference of 0.3 °C was only found in the posterior arm. These data verify that the participants did not show any injury or pain, as demonstrated in the pain perception results (Table 1). However, post-training differences in forearms were higher than 1 °C in both the anterior (1.1 ± 0.5 °C, p < 0.001, ES = 0.5) and posterior forearm (1.2 ± 1.0 °C, p < 0.01, ES = 0.5), possibly because of the isometric tension needed to support volleys (preventing wrist flexion and extension), to execute prone-supination movements in strokes, and to keep the racket in playing position throughout the hold training session [7,15].

As mentioned above, these results are in line with previous research carried out on archers [8], handball players [1] and tennis players [6], where a higher temperature on the dominant side was associated with the technical specificity of the sports [1,8]. This can be explained from a physiological perspective, considering the fact that more significant vasodilatation occurs in the muscles with greater metabolic needs (dominant side), facilitated by vasoconstriction of the less active muscles, which can increase thermal asymmetry (non-dominant side) [8]. It is important to emphasise that, unlike previous studies where these thermal differences did not exceed 0.5 °C [1,16], in this study, the differences found were twice the reference temperature of 0.5 °C. Thus, it is essential to highlight the need to individualise the thermal asymmetry reference value in asymmetric sports according to the game’s technical characteristics and the ROIs which are affected.

4.2. Pre-Training and Post-Training Comparison

Exercise induces heat production [16], increasing metabolic activity and blood flow to exercising areas [17]. Considering this statement, we hypothesised that post-training Tsk values would be higher than pre-training. However, the intervention results did not allow us to confirm these hypotheses, given the lower temperatures in most of the ROIs after training. Other researchers have found that Tsk decreases at the beginning of exercise due to cutaneous vasoconstriction [18]. Later, cutaneous vasodilatation provokes a change in the blood flow direction, dissipating heat from internal organs to the skin, and causing an increase in Tsk [18]. However, sweat rate and evaporation can be activated depending on the type of exercise, or if heat stress increases, leading to a decrease in Tsk [18,19]. This is a heat transfer process of great importance since, through evaporation, excess heat is lost due to the energy consumed in transforming the water that moistens the skin and mucous membranes (sweating) into vapour [20], maintaining thermal homeostasis [21]. It is also important to consider the moderate climatic conditions under which the study was conducted (14.6 ± 2.6 °C temperature; 45 ± 8% relative humidity): neither excessively hot nor excessively cold, but warm enough for sweating and subsequent evaporation to occur.

4.3. Correlation Analysis

The variation in the thermal symmetry of the different ROIs analysed can be caused by factors such as environment, clothing, or training load, among others [22]. More specifically, Tsk asymmetries can be affected by the years of experience in the sport [6,8], BMI [8], racquet weight [23], and string tension [24]. Regarding the results of the present study, direct correlations were observed between racket weight and Tsk asymmetries of the anterior arm and lower back; BMI and the posterior knee; fatigue variation and the posterior arm and posterior shoulder; and between pain perception and the anterior leg. Moreover, negative correlations were observed between BMI and the anterior shoulder; and between pain variation and the posterior knee.

Considering the sample size of the present study, we should be cautious with the correlations observed and consider them as aspects to take into account in future studies. The correlations between BMI and the Tsk asymmetry of the different ROIs can be explained by their asymmetrical body composition, resulting in lower Tsk on the sides with higher fat proportion [8]. Furthermore, previous studies have shown that the racket’s weight affects the technical execution and the number of muscle fibres involved in each movement [23,25]. This would explain the positive correlation between racket weight and thermal asymmetries of the anterior arm and lumbar, as well as the correlation between fatigue variation and thermal asymmetry of the posterior arm and shoulders. In addition, greater motor unit recruitment would result in greater heat production and thus more significant thermal asymmetry [8]. Finally, the correlations found in the lower limb and pain are interesting. Significantly, there is positive correlation between the perception of pain and the thermal variation of the anterior leg. This correlation could be associated with the high demand for changes of direction in tennis [26], as well as the technical execution of the forehand, where the hold body weight is placed on the forefoot (dominant leg) to promote trunk rotation for the stroke movement [26].

4.4. Limitations and Future Studies

Concerning the study’s limitations, as mentioned above, the sample used is small, although this can be explained by the fact that it is difficult to access elite athletes in competitive periods for this type of intervention. On the other hand, a standardised training approach was used for all the athletes, but the control of the loads on the track could not be determined objectively, and there could be variations in the different sessions. Finally, we must consider that on-court interventions better represent the reality of the game but make it more difficult to control external variables such as the weather or the game context itself.

Finally, future research could replicate this intervention in larger populations to verify that the results and correlations are reproduced similarly. It would also be interesting to generate groups according to age and sex to determine if differences in thermal asymmetries behaviours are found or by comparing injured and uninjured groups. It would also be interesting to extend the variables under study by taking into account not only the mean temperature values of the ROIs, but also the minimum and maximum temperatures.

5. Conclusions

Monitoring skin temperature with IRT in young elite tennis players provides coaches with relevant information to assess thermal asymmetries associated with the technical behaviour of athletes during training. Young tennis players present contralateral asymmetries of more than 1 °C immediately after training due to the technical characteristics of the game, and this thermal asymmetry can be correlated with racket weight, BMI, fatigue and pain. Moreover, the most relevant ROIs were the anterior and posterior forearm and the anterior and posterior shoulder. The staff can use this information for decision making concerning injury prevention or sport technique adjustment.