Short-Term Skin Temperature Responses to Endurance Exercise: A Systematic Review of Methods and Future Challenges in the Use of Infrared Thermography

Background: Body temperature is often assessed in the core and the skin. Infrared thermography has been used to measure skin temperature (Tsk) in sport research and clinical practice. This study aimed to explore the information reported to date on the use of infrared thermography to detect short-term Tsk responses to endurance exercise and to identify the methodological considerations and knowledge gaps, and propose future directions. Method: A web search (PubMed, Science Direct, Google Scholar, and Web of Science) was conducted following systematic review guidelines, and 45 out of 2921 studies met the inclusion criteria (endurance sports, since 2000, English, full text available). Results: A total of 45 publications were extracted, in which most of the sample were runners (n = 457, 57.9%). Several differences between IRT imaging protocols and ROI selection could lead to potential heterogeneity of interpretations. These particularities in the methodology of the studies extracted are widely discussed in this systematic review. Conclusions: More analyses should be made considering different sports, exercise stimuli and intensities, especially using follow-up designs. Study-derived data could clarify the underlying thermo physiological processes and assess whether Tsk could be used a reliable proxy to describe live thermal regulation in endurance athletes and reduce their risk of exertional heat illness/stroke. Also more in-depth analyses may elucidate the Tsk interactions with other tissues during exercise-related responses, such as inflammation, damage, or pain.


Introduction
Current sport dynamics have evolved rapidly from a physical, physiological, technical, and tactical point of view [1]. With this growth, new technologies have emerged to control and monitor the different sports capacities related to optimal performance. Among these technologies, those that stand out due to their portability, easy access to information, innocuousness, relatively simple interpretation of data, and mainly due to their non- 42 Racinais et al. 2021 [69] checked using 20 × 20 cm reference plates  Figure 1. Flow diagram of the study identification, selection, and inclusion. Figure 1. Flow diagram of the study identification, selection, and inclusion.

Data Sources
An electronic literature search was performed in four different databases: PubMed (MEDLINE) (n = 221), Science Direct (EMBASE) (n = 948), Web of Science (WoS) (n = 319), and Google Scholar (n = 1433). The search was performed until 2:00 p.m. on 17 November 2021. The authors did not differentiate or discriminate by journal or manuscript authors. The search strategy considered Boolean phrases as search descriptors as follows: "thermography OR thermology OR thermometry OR thermal imaging OR skin temperature OR body temperature AND exercise OR endurance OR sports". All references were extracted and imported into an open-source research tool (5.0.64, Zotero, Fairfax, VA, USA) to systematise studies.

Data Selection
The following inclusion criteria were considered: studies containing keywords in the title or abstract, experimental designs in humans, studies published from 2000 to 2021 (November) in peer-review academic journals, studies exploring the effects on Tsk of practicing endurance sports, but not limited to swimming, cycling, running, skiing, triathlon, or kayaking. Studies written in the English language were considered. Other languages were also considered if a translation could be performed (e.g., Spanish, Italian, Portuguese). A single author collected the original database of studies and compiled them into a data spreadsheet (Microsoft Excel, Microsoft, Redmond, WA, USA). After duplicate removal, two independent authors analysed studies, considering inclusion and exclusion criteria.

Data Collection and Extraction
Two different authors performed the analysis and selection of studies following the PRISMA protocol (see Figure 1). Specific exclusion criteria were followed to remove low-quality studies or those irrelevant to the primary aim of this systematic review, such as duplicates, articles with critical language limitations, studies in animals (e.g., dogs, horses), studies where the full text was not available, studies involving factors other than endurance sports, such as strength or conditioning research, different evaluation methods or technologies (e.g., thermistors, thermocouples, contact patches), book chapters, abstracts, conference papers and other technical reports, and articles which showed a severe lack of key information (e.g., participants information, discipline execute, confusing exercise protocol).
The protocol followed for selecting the studies was: a. an identification of potential studies, b. duplicate removal, c. title, abstract, and year examination, d. method quality and relevance review, and e. exploration of full texts. Finally, f. those studies with a lack of information (e.g., lack of sample data, lack of technical characteristics of equipment) were excluded.
The finally selected studies were analysed, and the following data were extracted using a descriptive table (see Table 1): authors and year of publication, study design and task performed, results and main study outcomes. Besides, Tables 2 and 3 present methodological relevant data of each study considering the Thermographic Imaging in Sports and Exercise Medicine checklist (TISEM) [28] as follows: participants and environment characteristics (participants data, previous instructions, extrinsic factors, environmental condition, environmental setup, equipment, image background, thermal adaptation) (see Table 2); and camera settings and setup (camera preparation, image recording, camera position, emissivity, assessment time, body position, method of drying the skin and image evaluation) (see Table 3).

Results
Tables 1-3. shows the compilation of the analysis of the 45 selected studies for this systematic review; they describe the study's design, analysis, and main outcomes (Table 1), methodological aspects based on TISEM related to participant and environment characteristics (Table 2), and camera settings and setup (Table 3).
Finally, the main results of the studies showed that, during exercise, the Tsk decreases (n = 12, 26.7%), and after the effort, during recovery, increases (n = 19, 42.2%); there were no lateral asymmetries due to exercise (n = 5, 11.1%); there was an inverse relationship between core and Tsk during exercise (n = 3, 6.7%); Tsk could be influenced by body composition, muscle activity, and sweating rate (n = 5, 11.1%); there is no consensus in the differences by sex (n = 3, 6.7%); and compression gear had no influence on Tsk (n = 1, 2.2%) (see Table 1).
A total of 22 (48.9%) out of 45 studies reported following previous instructions, such as no diuretics, drug, tobacco, alcohol, caffeine, heavy meals, or tea intake before the thermograms were taken. Moreover, no sunbathing, UV ray exposure, exercise, creams, or lotions were allowed. An additional six studies (0.7%) reported as exclusion criteria present skin burns, kidney problems, symptoms of pain, osteomioarticular injury, using compression garments, or attending physical therapy or using recovery methods prior to the evaluations.
Moreover, some room environmental conditions were reported as temperature and humidity in 84.4% of the cases (n = 38), ranging from 18.0 to 30.0 • C and 27.9 to 80% of humidity. Besides, a total of 20 (44.4%) studies reported avoiding direct airflow, electronic equipment (e.g., heating/cooling air conditioning systems) near the evaluation area, and artificial or natural light. Finally, 16 (35.6%) studies reported using an antireflective panel as a background of the thermograms to avoid temperature bias.

Camera Settings and Setup
Regarding the camera preparation, such as calibration (e.g., a reference plate connected to a thermistor), turned-up time, and stabilisation, only 20 (44.4 %) studies reported the procedure followed. There was a lack of information regarding the specifics of these procedures, which does not allow replication. The

Discussion
This study aimed to explore the information reported to date on using IRT to detect short-term Tsk responses to endurance exercise to propose future study directions in the application of thermography and identify those gaps in knowledge to be filled. A total of 41 publications, including a sample of 660 participants, were extracted.

Methodological Considerations to Assess Tsk
The most studied sport was running, followed by cycling. Studies analysed Tsk using cross-sectional designs through acute pre-post exercise protocols lasting from 15 to 90 min of running, cycling, swimming, and kayaking. The thermograms were collected using 19 different cameras under different environmental conditions and settings. Most of the authors preferred to set the IRT camera 1 m from participants' skin, using a 10 min thermal adaptation with a temperature of 18.0-27.0 • C and 50.0-64.0% relative humidity. Moreover, the ROIs selected for analysis varied widely, and only one study used a statistical technique (e.g., principal component analysis) to identify the most relevant ROIs.
Although there are some guidelines regarding the optimal settings that should be followed to obtain better thermogram results [9][10][11], this systematic review has shown that some studies did not report crucial data, such as room temperature, humidity, or thermal adaptation time. Considering multiple thermographic brands and settings selected to obtain the thermograms, studies should be performed on how different settings, such as room temperature and relative humidity, the camera distance from the skin, thermal adaptation time, and participants' positions, could potentially influence temperature data. How do the stimuli influence temperature changes depending on the exercise settings (e.g., exercising in hot, humid, windy environments) [31,58]?
Additionally, considering the studies extracted, the selection of several ROIs sometimes did not consider the segmentation of an area (e.g., quadriceps into rectus femoris, vastus lateralis, and medialis); it should be analysed whether the sub-areas of a body segment could present differences in temperature due to the different ROIs selected [35].
There is a lack of objectivity when selecting the ROIs to be analysed. This is one of the main concerns; usually they are based on technical considerations, such as the role of the area in the sport mechanics and performance. Some statistical data mining techniques that have been used in other sport and exercise science studies to discriminate the most relevant variables to be analysed could be used to select the ROIs [35]. One of the preferred statistical methods is principal component analysis. Still, other options (e.g., Bat algorithm, MLP networks, Bayesian classificatory, SMO algorithm, random forest, logistic regression, Levenberg-Marquart algorithm, K-means grouping algorithm, hierarchic grouping) are also valid and applicable to this kind of data.
Moreover, despite the vast amount of evidence regarding the Tsk responses in acute conditions to relatively short stimuli, such as time to exhaustion tests, incremental tests, and other similar tests lasting 15-60 min at different intensities, there is a lack of studies recording the physiological skin responses to heavier and longer stimuli, such as the marathon, long-distance triathlons, multi-stage cycling, open-water swimming, and other similar efforts, that could allow a more in-depth analysis of the behaviour of Tsk and its relationship with other well-known physiological responses, such as muscle damage, pain, and temperature regulation [31,58].
The preferred acute stimuli used to explore Tsk change using thermography were a 60 min aerobic test in running (n = 7) and an incremental test until exhaustion in cycling (n = 5). To better understand the Tsk responses after exercising, more longitudinal studies are needed to elucidate the physiological cascade influencing Tsk and how it is related to adjacent tissue responses [73]. This may allow scientists, researchers, athletes, medical staff, coaches, and other stakeholders to detect, monitor, and control the Tsk changes in endurance practitioners at both the group and the individual level. This is key for extending the knowledge of the physiological effects of endurance exercise beyond a single moment in time.
Some methodological guidelines are available as TISEM [28] to report critical methodological procedures when using IRT in sports and exercise sciences. Indeed, there is a lack of information in the studies selected considering the participants and environment characteristics, and camera settings and setup. Future studies must follow this kind of methodological guidelines to make the studies replicable.

Challenges in the Use of IRT in Endurance Sports
Researchers have been concerned with understanding how exposure to prolonged and strenuous exercise could impact short-and long-term performance and health, physiologically and physically [12]. Moreover, there has been a great increase in endurance running participation [74], mostly among increasingly younger participants commonly requiring medical assistance while participating in these endurance events [75]. Following this information, thermography has been shown to detect acute changes in Tsk provoked by exercise. During physical movement, the skin blood vessels contract slightly due to the higher muscle activity as a response to higher sympathetic activity [76], which is why this systematic review identified articles suggesting a decrease in Tsk during effort. In contrast, after exercise, an increase in Tsk is expected due to the heat produced by muscle contraction, when the cutaneous vessels may dilate and dissipate heat. In this regard, more studies are needed to understand how the control circuit of heat conduction and dissipation from the muscles to the skin works and how these physiological processes could later impact Tsk [77]. Moreover, more evidence is needed regarding the interaction between core and peripheric tissue temperature with Tsk, since there are several factors influencing Tsk, such as blood flow rate, thermal conductance, thermal capacitance, skin surface area, body segment lengths, and metabolic rates [78].
Skin is the outer layer that controls thermoregulatory functions and heat transfer for sweating, vasodilation and vasoconstrictions, and shivering. Although, a better understanding of the functionality of perforator and perforasome (potential thermal connection to other nodes by convection and conduction) as the interconnected vascular system between the skin and deeper tissues during pain, inflammatory cascade, and tissue damage as regular outcomes after prolonged exercises is needed [67]. Consequently, multi-segment and multi-node (e.g., core, muscle, fat, skin) thermal modelling studies could be performed for endurance exercising [78], considering Tsk change during endurance sports could oscillate [76] while intramuscular temperature remains constant [79]. Studies that clarify these dynamics and physiological interaction between tissues are necessary, as well as the understanding of how Tsk may or may not suggest changes in structures at a deeper level.
In line with the abovementioned, in endurance sports, it seems that the rise in Tsk could be due to the heat generated by the prolonged exercise and its subsequent physiological processes as the increase in endothelial nitric oxide, glycogen resynthesis, or increase in systemic hormones [67]. Despite there being a need for a more in-depth analysis of the thermo physiological responses to endurance exercise, skin blood flow and Tsk play a critical role in thermoregulatory processes. Due to the prolonged duration of these, heterogeneous behaviour in Tsk that could be found in the different ROIs could be explained by the different environmental conditions, exercises durations, intensities, and mechanical load [76]. The permanent switch from skin vasoconstrictions to vasodilation and vice versa, due to the need to lose or maintain body temperature heat, could provoke a decrease and subsequent increase in Tsk. This could be the reason why there could be changes in Tsk during a constant-load exercising by time. Considering these continuous and heterogenous changes in Tsk, there is a need to explore new technology that allows monitoring of the athlete constantly. As has been evaluated so far, recording the temperature before and after an endurance event does not seem to be sufficient to record these potential changes throughout an event of long duration and relatively variable efforts (e.g., trail running, kayak, open-water swimming, cross-country skiing).
Tsk requires a standardised protocol commonly considering environmental conditions, calibration, and consistent set-up for the thermographic camera [80]. All these protocols were selected for laboratory-based research, and no field protocol have been proposed due to some set-up limitations. In this sense, considering that endurance sports are usually practiced outdoors, IRT have a great limitation to outpatient and field-based assessments. In this sense, the TISEM [28] methodology guidelines may be followed to standardise the reporting of key imaging protocols, equipment characterisation, participant preparation, and temperature assessment techniques. A revised protocol should be proposed in future studies to evaluate Tsk using IRT in field-based settings.

Limitations
While this study aimed to explore the knowledge reported to date on the use of IRT to detect short-term Tsk responses to endurance exercise, the results and outcomes must be seen in the light of some limitations. One of the main limitations is the wide range and difference between study designs, assessment settings, and temperature. This issue makes the studies and outcomes difficult to compare and interpret, and, finally, difficult to link with other evidence and combine to obtain more precise results. Considering the organic characteristics of the sample and that endurance sports participants are relatively uncommon, having a sufficient sample size or follow-up designs is challenging. More studies need to be performed regarding the sample characteristics considering a more diverse population in order to have more generalisable results to then propose a more global solution.

Conclusions
Although there is the relatively large number of published articles regarding the use of IRT to assess acute Tsk responses to endurance exercise and sports practice, several