Most contact temperature measurement technologies use the direct method of temperature measurement. In a typical scenario, a sensor (which is typically small compared with the body or body part) is placed in contact with the body part and comes in thermal equilibrium with it. Examples of such measurements can be a mercury thermometer or a PA catheter (Bridges and Thomas, 2009).

Non-contact (remote) temperature measurement techniques do not measure the temperature directly. Instead, they measure energy flow from the object and derive temperature from these measurements (radiometry). Radiometry is based on the fact that all objects with temperatures above 0K emit electromagnetic radiation in a broad range of wavelengths. However, most of the radiation from surrounding objects (e.g., in the 0–100°C temperature range) is coming for wavelengths 3 μm and longer, commonly referred to as thermal infrared, or IR. For example, the peak of radiation from an object at 300K (27°C) is 9 μm. Objects in this temperature range do not emit noticeable radiation in the visible range of the spectrum. Only objects heated 500°C and above emit noticeable radiation in the visible range of the spectrum (i.e., red-hot metal).

An important concept in radiometry is a blackbody, which is an idealized physical body that absorbs all incident electromagnetic radiation, regardless of frequency or angle of incidence. The blackbody also emits 100% of the absorbed energy, thereby having an emissivity of 1.0 or 100%. The emissivity of an object is the ratio of its energy flow to the energy flow of the blackbody at the same temperature. Thus, a 2-D imaging array of thermal IR sensors does not measure temperature directly, but the energy flow, which is emitted by the target. IR sensors typically measure energy flow over a broad range of spectrums. Then, the temperature is derived from these measurements. In particular, the amount of emitted radiation depends upon both the temperature and the emissivity of the material.

For practical purposes, the middle (MWIR, 3–5 μm) and long (LWIR, 8–14 μm) wavelength infrared (IR) spectral ranges are used.

Finally, thermal optical measurements have an established terminology. According to Glückert and Schmidt (2001), while both pyrometry and thermography refer to the remote measurement of temperature based on heat radiation, a pyrometer is a “radiation thermometer,” which provides point like information, while thermography refers to a “temperature picture,” where the information is optically recorded as an entire scene.

Historically, mechanical temperature sensors (e.g., a mercury thermometer) were the only practical way to measure the temperature. Recently, mercury thermometers were mostly removed from clinical practice due to toxicity concerns. Liquid-crystal contact thermography is another type of direct temperature measurement based on mechanical properties (Benbow et al., 1994).

Currently, electrical temperature sensors have mostly replaced mechanical thermometers in clinical use. The two most common types of electrical temperature sensors are resistance temperature detectors and thermocouples.

Both thermocouples and thermoresistive devices are used in clinical applications. For example, Uleberg et al. (2015) used thermocouples and thermistors as an epitympanic sensor in the auditory canal for continuous monitoring of core temperature in severely injured patients. However, historically, thermocouples were mostly used in many applications. Nowadays, thermistors have become the detector of choice for temperatures below 125°C.

There are two types of sensors used for indirect thermal measurements: non-quantum and quantum (photon) detectors. Non-quantum detectors respond to radiant energy in a way that causes a change of state in the bulk material (i.e., heating the sensing element produces the bolometer effect). They are mostly based on pyroelectric and ferroelectric materials or microbolometer technology. Most IR cameras have a microbolometer-type detector, mainly because of cost considerations. Microbolometers can be created from metal or semiconductor materials. Metal bolometers are produced from thin foils or metal films and usually work without cooling.

Operations of quantum (photon) detectors are based on quantum effects (e.g., excitation of valence electrons to the conduction band), where the state of electrons in a crystal structure changes in reaction to incident thermal photons.

Existing thermal measurement technologies are analyzed across two interplaying dimensions: contact vs. remote and single or few-point vs. multi-point measurements.

The distinction between few-point vs. multi-point measurements is quite nominal. For clarity, we will consider multi-point measurement devices as those, which have a regular (rectangular) grid of sensors (e.g., mat or IR camera). All other modalities with several sensors (e.g., an insole or sock) will be considered few-point measurements.

Core body temperature refers to the temperature of the body’s internal organs, such as the heart, liver, brain, and blood. The average normal body temperature is generally accepted as 37°C. However, studies have shown that the “normal” body temperature can have a wide range, from 36.1 to 37.2°C and that it can also be influenced by age, activity, and time of day (Wang et al., 2016).

Core body temperature is tightly regulated by both peripheral and central receptors that are integrated within the hypothalamus (Sessler, 2009). The normal core temperature for an individual can be considered as a “set point” in the thermoregulation system. All the temperature mechanisms continually attempt to bring the body temperature back to this set point. For example, when the body core temperature rises above the critical level of normal body temperature, there is an increase in the rate of heat loss by sweating. Accurate measurements of core body temperature involve invasive medical procedures in hospitals. The PA catheter, which measures blood temperature, is considered the gold standard for measuring core temperatures (Bridges and Thomas, 2009). For non-critical patients, a person’s peripheral body temperature is commonly measured in non-invasive sites such as the mouth, ear, armpit, and rectum because these areas are easily accessible and are believed to provide the best estimation of the core body temperature. However, the accuracy and precision of each of these methods are different. A core temperature measurement can be considered accurate if the mean difference from the PA temperature is ±0.3°C and precise with a standard deviation ranging from 0.3 to 0.5°C (Lawson et al., 2007). A comprehensive summary of accuracy and precision of various non-invasive temperature measurement techniques (the oral, ear-based, temporal artery, the axillary temperature) in intensive care unit (ICU) settings is presented in Bridges and Thomas (2009). The author found that while all methods [with the exception of axillary, which underestimates the PA temperature, were accurate, the precision varied across methods (oral, SD = 0.24–0.6°C; ear-based, SD = 0.4–0.57°C; temporal artery, SD = 0.5–1.1°C; and axillary, SD = 0.16–0.6°C)].

In contrast to the core temperature, the skin temperature falls and rises with the temperature of the surroundings. The skin is an efficient, controlled “heat radiator” system. The flow of blood to the skin is the most effective mechanism for heat transfer from the body core to the environment. The flow of energy to and from the skin determines our sense of hot and cold. For that reason, despite the core body temperature being constant over time, the skin temperature is subjected to changes and is usually lower than the core (Martinez-Jimenez et al., 2021).

Body heat dissipates by radiation, evaporation, convection, and conduction. In normal conditions, radiation is the most significant source accounting for approximately 60% of heat loss (Koop and Tadi, 2021). At rest, evaporation accounts for 22% of heat loss. However, sweating (evaporation) is the primary means of cooling the body during exercise. Different heat loss mechanisms interplay with each other. For example, evaporation and conduction of the air are accelerated by convection. The energy transfer (and skin temperature) is determined also by the temperature and the moisture of the surrounding environment.

The normal skin temperature is about 33°C. However, the skin temperature is not maintained at a constant value across the entire body. Lower temperatures are characteristically observed in proximity to superficial veins, relative to superficial arteries, and over protruding body parts including the toes, fingers, ears, and nose. Similarly, skin surface temperature has been observed to be higher over organs with high metabolic rates rather than those at rest, as well as over muscles rather than tendons or bones (Kanitakis, 2002). The body can vasoconstrict regions of skin (other than chest, neck, and face) exposed to cold temperatures to prevent heat loss. The hand is probably the most active body part in responding to the body’s thermoregulation requirements. In warm conditions, the hand is fully vasodilated, and the fingertips are the warmest areas of the hand. This pattern is reversed when cooling. Vasoconstriction of the hand blood vessels causes the skin temperature of the hand to vary on the order of 8°C. When the hand is cold, it ceases to transfer much body heat to the environment. Because of this, an IR image taken a few minutes after a subject moved from a warm environment (30°C) to a slightly cool (22.6°C) environment, shows that, although the rest of the upper body temperature has not changed much yet, the blood vessels of the hand were already well constricted (Lim, 2020). Likewise, a 3°C difference in finger skin temperature has also been observed in slightly cool environments caused by the difference in muscular exertion between typing vs. holding a computer mouse (Huizenga et al., 2004).

Interpreting an absolute skin temperature measurement (especially a single-point measurement) may pose a challenge because of skin thermoregulation. Therefore, a temperature gradient (the difference in temperatures between two points) is a more objective measure, which will depend less on the ambient conditions. Figuratively speaking such an approach is “self-calibrating” where one point (e.g., intact skin) in the same region is selected as a reference. In this case, the positive gradient may be indicative of increased local blood flow (e.g., inflammation), while the negative gradient may be indicative of the reduced blood supply (e.g., obstructed blood flow) (Martínez-Jiménez et al., 2018a; Martinez-Jimenez et al., 2021).

One known problem with thermography is the distortion in the image caused by evaporative water loss in the wound bed. This problem can be solved either by allowing the wound to dry completely (which may delay the timing of the assessment) or by applying a non-permeable covering to the wound bed, which eliminates the problem of evaporation (Cole et al., 1991).

The emissivity and apparent temperature of an object varies with the angle of viewing. Watmough et al. (1970) calculated the emissivity, ε as a function of a refractive index of the material, n and viewing angle, φ (see Eq. 1):

In Figure 2, we plotted dependence of the emissivity on the viewing angle for several refraction indexes. It can be seen that the emissivity is flat in the range of 0 < φ < 60°. Thus, in practical applications, the angle of viewing with respect to tissue surface should not exceed 60° or ideally 45°, since for larger angles the emissivity of the surface will be significantly reduced. For skin viewed obliquely, the lower emissivity may lead to a reduction in the apparent temperature of more than 4°C, so a significant “hotspot” in this region might be undetected in a thermogram.

Several practical considerations should be considered for optimal thermal imaging (Moreira et al., 2017). Namely:

Diabetic foot ulceration (DFU) is a major complication of diabetes. The annual prevalence of DFU is estimated to be 4–10%, and the lifetime risk of developing these ulcers in people with diabetes is estimated to be anywhere from 15 to 25%. Foot ulceration increases the risk of lower limb amputation, one of the most debilitating complications of diabetes. If untreated, DFU may become infected and require total or partial amputation of the affected limb. More than 50% of non-traumatic lower-limb amputations are attributable to diabetic ulcers (Raghav et al., 2018).

High throughput screening allows earlier interventions in a patient population that already presents clinically with late-stage complications, significant morbidity, and mortality risk. Thermography has significant potential as an adjuvant technique in diabetic foot assessment. For example, elevated temperature is a reliable marker of inflammation and can thus predict the risk of ulceration, infection, and amputation (Figure 3; Bharara et al., 2012). Similarly, the decreased temperature may be a sign of insufficient blood supply and indicate ischemia (Lin and Saines, 2017).

When skin is subject to prolonged pressure or shearing forces, a pressure injury can occur. Although the terms decubitus ulcer, pressure sore, and pressure ulcer have often been used interchangeably, the National Pressure Injury Advisory Panel considers pressure injury (PI) the best term to use because open ulceration does not always occur (Table 1; Black et al., 2007). PIs are considered a preventable disease because the risk factors that lead to pressure injury can be identified in advance using tools like the Braden Score. Despite this, PI continues to occur in 6–18% of hospitalized patients (Li et al., 2020), with significantly higher prevalence in the chronically ill, patients in the intensive care unit, and those with spinal cord injury (Labeau et al., 2021). Given the considerable personal and financial costs the use of a practical method to detect early pressure injury is greatly needed.

The high incidence rate of PI calls for reliable and practical methods for early detection to guide therapeutic measures to mitigate progression. Inflammatory and apoptotic changes in the epidermal and dermal layers may precede surface changes by 3–10 days (Moore et al., 2017). In addition, clinical signs and symptoms which initially suggest a stage 1 PI may conceal a deep tissue injury. The tissue damage of deep tissue injury (DTI) in an area of intact skin evolves from the deep layers to the surface over 2–3 days and within 7–10 days further deterioration to a necrotic state occurs (Mayrovitz et al., 2018). The diagnosis of DTI is based on visual inspection alone, which is affected by subjective bias and user experience. In addition, the diagnosis can be exceedingly difficult in patients with dark skin (Fitzpatrick skin tone V–VI). More sensitive and accurate methods for detecting DTI are needed.

Risk assessment is the first step in the prevention of PIs. Unfortunately, the predictive value of current risk assessment scales is variable with modest performance (Adibelli and Korkmaz, 2019). A systematic review from Cochrane found no empirical evidence that using risk assessment scales can reduce the incidence rate of PI (Moore and Patton, 2019).

Infrared thermography offers the advantages of speed, simplicity, reproducibility and objective measurements. Numerous studies have reported thermal abnormalities occur with PI. Temperature measurements for skin assessments are recommended by international experts as part of the comprehensive skin and tissue assessment; the evidence-based statements published in the 2019 Clinical Practice Guideline include (Kottner et al., 2019; Internationalguideline, 2021):

Infrared thermography has been shown to be a useful physiological marker for assessment of PI development (Sae-Sia et al., 2005). Compared to normal skin, the site of injury may exhibit a rise or a fall in temperature (Figure 5).

Two studies deserve special mention. The study by Simman and Angel (2021) was a blinded prospective cohort study of 70 ICU patients. Participants had their sacral area and bilateral heels scanned with the thermographic device. Follow-up imaging took place throughout the patient’s stay (upon admission and 3, 7, 14, and 25 days after). Clinicians were blinded from the thermal images and all participants received standard of care for the prevention and treatment of wounds. The objectives were an assessment of thermography for the detection of non-visual physiologic changes of DTI that can precede the visual changes, minimizing the subjectivity and difficulty of the visual assessment, and evaluating the financial impact of implementing thermographic technology. Of the 70 participants enrolled in this study, four patients with intact skin progressed to visually identifiable DTI’s. On all of these individuals, the thermographic device identified a pre-visual temperature anomaly before becoming a visually identifiable DTI. Three had an increased temperature (>1°C) above normal skin and one had a decreased temperature (−5.3°C) below average regional skin values. These temperature changes preceded the visual appearance of the DTI by 5–18 days (Figure 6).

The second notable study by Cai et al. (2021) evaluated 415 ICU patients enrolled by a convenience sampling method. The risk of PI was assessed via Braden Scale and thermal images of the sacral area were obtained by an infrared thermal imager once a day for 10 days. The predictive effects of infrared thermography and the Braden scale on PI were compared by the receiver operating characteristic curve from which the optimal cut-off value of skin temperature for predicting pressure injury was determined. The authors found that the relative temperature of the sacral area was negatively correlated with the risk of PI. The efficiency of infrared thermography for diagnosing PI was better than that of the Braden Scale. Based on the relative temperature optimal cut-off value (−0.1°C), Kaplan–Meier curve and Cox proportional hazard regression model analysis showed the incidence of pressure injury with relative temperature below −0.1°C was higher than the group with relative temperature above −0.1°C.

In the context of skin inflammatory disorders, infrared thermography has proven useful in detecting inflammation and staging the severity of the disease. For example, a positive temperature gradient of 2–4°C was found in the axillae and groins of people suffering from hidradenitis suppurativa, an inflammatory disease of the hair follicles (Zouboulis et al., 2019). Two recent studies demonstrated that in the same disease, when quantified, the positive gradient temperature has a high degree of correlation with several of its clinical staging systems (Ramirez-GarciaLuna J. et al., 2021; Ramirez-GarciaLuna J. L. et al., 2021). Due to the subjective nature of these scoring systems, the authors of these reports recommend using thermography as an objective tool to monitor disease severity and assess response to treatment during clinical trials (Figure 7).

Thermography can be used for treatment planning and monitoring. In particular, a case report suggested that thermography of patients with hidradenitis suppurativa has value in planning the surgical excision of the diseased area and that once the inflamed tissue is removed, the areas of positive temperature gradient normalize over time (Polidori et al., 2017), or that the lesions observed can be mapped to other medical imaging modalities, such as magnetic resonance (Derruau et al., 2018). Likewise, thermography has been used to monitor treatment response in patients with psoriasis as normalization of temperature gradients is correlated with disease improvement and affection to joints (Castillo-Martínez et al., 2013; Capo et al., 2015). Finally, thermography has been extensively used to map microvascular involvement in systemic inflammatory diseases such as lupus erythematosus and scleroderma (Saygin et al., 2019). For example, thermographic imaging of the hands of subjects with systemic sclerosis has an overall accuracy of 74% for correctly identifying people suffering from this disease (Murray et al., 2009) and is highly predictive of the development of digital ulcers (Hughes et al., 2016).

Beyond skin diseases, thermography has also been found to be effective for the identification of systemic conditions. Thermographic imaging of the abdomen of adults with acute abdominal pain was found to have a diagnostic accuracy comparable to that of ultrasound imaging to diagnose appendicitis (Ramirez-GarciaLuna et al., 2020). The authors of this study demonstrated that temperature gradients between the right and left lower abdominal quadrants as little as 0.35°C are highly predictive of appendicitis. Thus, if these temperatures are assessed with lower sensitivity tools, such as digital thermometers, diagnosis may be missed (Steele, 1986). The abdominal temperature difference may be related to pain, which is associated with blood flow redistribution due to autonomic nerve activation. Indeed, evidence of thermal changes in the lacrimal carunculae have been found to appear in response to noxious stimuli (Kolosovas-Machuca et al., 2016). This research opens the door for using thermography as a non-invasive imaging modality for measuring pain and its associated physiological responses; thereby, lowering the intrinsic subjectivity of these experiences (Erel and Özkan, 2017). Finally, in the context of the ongoing COVID-19 pandemic, infrared thermographic assessment has been proposed as an alternative to screen people to identify facial inflammatory changes suggestive of infection (Martinez-Jimenez et al., 2021). In contrast, regular fever screening with digital thermometers has been found to be poorly predictive of infection in minimally symptomatic adults (McConeghy et al., 2020). However, further research needs to be carried out to fully demonstrate the effectiveness of thermal infrared screening for this application.

It is estimated that in North America alone, up to 30% of surgical patients develop a surgical site infection (SSI; SSI et al., 2021). SSIs are defined by the United States Centers for Disease Control (CDC) as the presence of purulent exudates, a systemic fever of more than 38°C, site pain, tenderness, localized swelling, redness or heat within 30 days after surgery (Horan et al., 2008). These definitions are very broad and non-specific, objective measurements to classify and diagnose patients with SSI are needed. Thermography has shown promise to be such an objective tool (Figure 8). For example, a study found differences in the thermal pattern in response to injury in the surgical incision in days 1–4 after surgery in patients who underwent enterostoma closures (Siah et al., 2019). The authors of this paper found that non-infected post-surgical wounds showed an increase of approximately 1.5°C in the periwound area within the first 48 h after surgery. In contrast, infected wounds did not show this pattern and took almost 4 days to reach similar temperatures. Furthermore, the abdomen of patients with post-surgical infections showed “coldspots” during the first 4 days after surgery, which were highly suggestive of SSIs. Another study where follow-up of obese women who underwent cesarean sections was done at 7, 15, and 30 days postoperative found a similar pattern (Childs et al., 2019). A temperature gradient of −1°C between the periwound area and the abdomen at day 7 after surgery was associated with a 3-fold increase in the likelihood of having a SSI. The authors also found that after this time point, temperatures between patients with or without SSI were very similar; thus, concluding that the value of thermal imaging for diagnosing SSIs was in the first days after surgery. In contrast, thermographic monitoring of patients following total knee arthroplasties showed that a 1.6°C increase between presurgical temperatures and the peak temperature in the first 3 days after surgery was highly suggestive of septic arthritis (Romanò et al., 2012). Likewise, persistent increases in the periwound temperature after procedures involving sternotomies were found to be suggestive of infection in two studies (Robicsek et al., 1984; Fujita et al., 2013). Finally, a recent report showed that temperatures above 34°C in the pin sites of patients who underwent external fracture reductions had a sensitivity of 73%; specificity, 67%; positive predictive value, 10%; and negative predictive value, 98% to detect infection. Additionally, the authors of this report demonstrated that the intra-rater agreement accuracy for thermography was excellent (Rahbek et al., 2021). Taken together, these findings strongly suggest that thermography may be a viable tool to objectively detect early SSIs; however, they also highlight the variability of measurements for different surgical procedures and anatomic locations and calls for more studies to be performed. Nonetheless, because infrared thermography can easily be paired with mobile devices (Wang et al., 2017) and smartphone apps have been demonstrated to be promising technologies for SSI monitoring (Wang et al., 2020), it can be expected that in the near future these combined technologies can be leveraged for the early identification of SSIs.

Flap loss is the most feared complication following flap surgery. Traditionally, flap monitoring is performed by clinical assessment, including the assessment of the tissue coloration, temperature, capillary refill and turgor (Shen et al., 2021). However, it is estimated that between 8 and 14% of cases of flap reconstructions present some degree of failure due to thrombosis of the tissue. If thrombosis is recognized promptly, the tissue may be salvaged; however, due to the subjective nature of clinical inspection alone, this does not always occur (Brands et al., 2010). Infrared thermography may be an attractive technology for measuring flap viability after surgical reconstruction because there is a direct relationship between tissue perfusion and its temperature, (de Weerd et al., 2011; Sagaidachnyi et al., 2017). One study found that hotspot images on infrared imaging after flap surgery were always associated with arterial Doppler signs over them (Sjøberg et al., 2020). Another study found that free flap failure was correlated with negative temperature gradients during surgery (−2.64°K), at 24 h post-operative (−1.22°K), and 10 days after surgery (−0.058°K) (Just et al., 2016). Similarly, Meyer et al. (2020) found that intraoperative non-rewarming of the tissue after surgical microvascular grafting was associated with flap loss. Before anastomosis, the mean temperature of flaps showed a negative temperature gradient of −11.47°F which was normalized to 0°F when the flow was adequately restored. Taken together, these findings strongly suggest that infrared technology can be used during flap surgery to predict patient outcomes (Figure 9). Indeed, infrared imaging has previously been used to predict the healing outcome, the need for surgical treatment, and the time needed to achieve wound healing in patients with burns (Martínez-Jiménez et al., 2018a; Carrière et al., 2020). However, due to the scarcity of scientific reports, caution for extending the conclusions on flap survival is warranted.

Angiographic studies demonstrate that the skin’s temperature correlates directly with overall perfusion (Nagase et al., 2011). Thus, not surprisingly, patients with peripheral artery disease (PAD) exhibit disparities in their skin temperature distribution, particularly in the soles (Ilo et al., 2020). This finding was used to develop a prognostic measurement of vascular compromise and cardiovascular risk akin to the widely known ankle-brachial index (ABI). The thermographic ABI (tABI) consists of obtaining thermograms of the hands and feet of patients and dividing the peak lower extremity temperature by the upper extremity temperature (Wallace et al., 2018). The researchers who developed the method found a Pearson correlation of 0.83 (p < 0.001) between ABI and tABI measurements. Because this measurement is non-contact and non-invasive, it could be an attractive alternative to traditional ABI measurements if validated in larger patient cohorts. Likewise, the 6-min walk test (6MWT), another classic non-invasive study for the assessment of PAD, has been studied in a thermographic protocol with measurements before and after the patients got on the treadmill (Huang et al., 2011). The post-exercise temperature was found to drop in the lower extremities of patients with arterial stenosis, but they were maintained or elevated slightly in the extremities of people with patent arteries (temperature changes at sole in PAD vs. non-PAD patients: −1.25 vs. −0.15°C; p < 0.001). The exercise-induced temperature changes at the sole were also positively correlated with the 6MWT distance walked (Spearman correlation coefficient = 0.31, p = 0.03) and correlated with the ABI (Spearman correlation coefficient = 0.48, p < 0.001). Thus, the investigators of this study concluded that sole measurements after completing the walk had a sensitivity of 81.7% and specificity of 65% to diagnose PAD. Finally, successful revascularization of limbs with critical ischemia has demonstrated a classic thermographic pattern characterized by a significant increase in the temperature of the revascularized limb, compared to the contralateral side within the first 24 h after surgery (Staffa et al., 2017; Ilo et al., 2021; Figure 10). Thus, the plantar temperature has been used to predict amputations after endovascular therapy in this patient population with a sensitivity of 70% and specificity of 54% (Chang et al., 2020).

In the field of venous pathology, infrared thermography has been used to assess wound healing, response to treatment, and predict infection in patients with venous ulcers in a similar manner to patients with diabetic ulcers (Martínez-Jiménez et al., 2013, 2018b; Dini et al., 2015; Cwajda-Białasik et al., 2020). Finally, infrared thermography has also been used to predict fistula maturation in a cohort of 100 patients with end-stage renal disease who needed vascular access for hemodialysis (Al Shakarchi et al., 2017). Because the creation of an arterio-venous fistula (AVF) creates significant hemodynamic changes to blood flow in the arm, the authors of this report found that in turn, these changes lead to thermal changes distally. Negative temperature gradients between the patients’ hands were highly predictive of a successful AVF, while no changes or positive gradients almost always were associated with procedure failure. The authors of this report also found that for AVF clinical patency, infrared thermal imaging had a positive predictive value of 88% and a negative predictive value of 86% and that for functional maturation, it was 84 and 95%, respectively. Furthermore, they also concluded that thermography was highly superior to the commonly used intra-operative predictor of the presence of thrill or other independent pre-operative patient factors. Taken together, these reports strongly suggest that infrared thermography is an imaging modality with significant value in vascular surgery.

Like all mammals, humans exhibit a wide degree of adaptive physiological changes to exercise, and because many of them entail thermoregulation, they can be readily assessed using thermal imaging (Mota-Rojas et al., 2021). As opposed to the previous use cases discussed in this review which have been done using static thermal imaging, the assessment of physiological responses to exercise requires dynamic imaging to capture the changes produced (Hillen et al., 2020). The studies that assess thermoregulation during exercise can be divided into three types: (a) constant long endurance exercise (more than 10 min with constant intensity), (b) resistance exercise (less than 10 min, typical resistance exercise with load), and (c) incremental exercise testing (increasing intensity protocol through exhaustion). In the first group, endurance exercise, common findings have been a blood flow restriction to the skin that produces a decrease in its temperature at the beginning of the exercise period followed by a rise of the skin temperature to basal or slightly above basal immediately after termination of the activity, with an in-between thermal plateau. This thermographic pattern reflects the redistribution of the blood flow to muscles during exercise and back to the skin after it to promote heat dissipation (Zaïdi et al., 2007; Balci et al., 2016; Drzazga et al., 2018; Hillen et al., 2020). In contrast, resistance exercise produces an increased skin temperature pattern related to the stressed muscles which is more pronounced in trained athletes, compared to untrained individuals, and in younger people compared to older individuals (Weigert et al., 2018; dos Santos Bunn et al., 2020). These findings can be linked to local and systemic adaptations of myocutaneous blood flow and reflect the muscles’ metabolism while under stress, as they have also been seen in individuals experiencing ergonomic injuries (Alexandre et al., 2017). Finally, for incremental exercise testing, a mixed pattern with no thermal changes at mild exercise levels and a similar pattern to endurance athletes of heat dissipation at the end of the workout session was observed at the highest intensity levels (Priego Quesada et al., 2017; Drzazga et al., 2018). These changes have been proposed as an adaptive response to heat dissipation of the core temperature, which is considered one of the most limiting factors for exercise performance (Lahiri et al., 2012; Hillen et al., 2020). Taken together, the findings strongly suggest that thermal imaging of athletes can be highly predictive of performance. Thus, it is not surprising that research using near-infrared imaging (NIR), an imaging modality closely related to thermal imaging but that can also measure oxygen content and that can be integrated into clothing materials, is a promising technology to assess athletes’ fitness, cardiovascular performance, ease of recovery, and predict injury risk (Bontemps et al., 2021).