The approach to understanding skin temperature naturally begins with thinking of the process of heat transfer from the body interior to the outside world in terms of conduction. In this view, the heat produced in the body makes its way to the surface through successive layers, with the rate of transfer through each layer related to the temperature gradient across the layer and a specific conductivity. Overall, the body is seen, thermally, as a protected core surrounded by an insulating shell (Fig. 1). Surface temperature is the outcome of the equilibrium between this process of delivery of heat to the surface and the process of transfer of heat to the environment.
Early physical models were almost this simple, except for attempting to take into account two obvious ways in which the outer layers of the body are not a mere insulating blanket; heat is produced within each layer, and the layers are perfused by a moving fluid, blood. The model of Pennes, published in 1948, {699} incorporated tissue heat production and treated the convective influence of blood as an effective conductivity, i.e., with heat transferred across a layer according to a blood flow rate and the trans-layer temperature difference. This model was applied to limbs and predicted parabolic temperature contours which corresponded reasonably well to data obtained by pulling thin wire thermocouples through volunteer's forearms. The Pennes model and its use up to the present day are discussed by Baish in the previous edition of this Handbook series {Baish 2085}.
Starting with this conceptual view of heat transfer, one can move on to thinking about the influence on surface temperature of anomalies, such as localized regions where heat production is greater or less or where blood flow is significantly altered. Obviously, these could influence surface temperature, altering the balance on the overlying skin surface between the rate of arrival of heat from below and the rate of dissipation to the environment.
Anomalies were detected clinically from time immemorial by taking advantage of the exquisite sensitivity of the fingertips to temperature (Hardy and Oppel showed that the threshold of sensitivity to temperature changes is in the order of thousandths of a degree C {1838}). An experiment performed by Cooper et al. {660} showed that the increased metabolism and blood flow associated with exercise of a muscle in the forearm was associated with, in effect, a thermal image on the overlying skin, assessed by measurements with thermocouples that revealed fractional-degree temperature increases .
I will leave it to the other authors in this collection to describe the clinical findings in which the sensitivity of modern infrared imaging has been employed. The purpose of this section is to point out anatomical and physiological features of the vessels that convey blood to, through, and from the skin. These comprise the cutaneous vasculature. Both the physiological and anatomical peculiarities have the potential for altering the relationship between skin temperatures and thermal features of underlying layers.
How hot can the hot spots in the body be? Apart from external influences and the situations in which countercurrent trapping of heat occurs, the obligatory relationship between metabolism and convective cooling by the blood keeps tissue temperatures close to the temperature of their arterial blood supply.
No tissue can survive without blood flow adequate to supply the needs of oxidative metabolism. The amount of oxygen that can be delivered to a active cells is limited by the storage capacity of blood. Each liter of arterial blood contains roughly 200 ml of O2 under normal conditions of respiration. In normal metabolism, approximately 4.8 kcal of energy is converted from the stored chemical form per liter of oxygen consumed. Were this all converted to heat, utilization of the entire oxygen content delivered by a liter of arterial blood would result in liberation of (0.2)*4.8 kcal. This amount of heat would raise the temperature of a liter of blood approximately 1C. This means that if the liter of blood that delivered the oxygen left the tissue 1C warmer, the rate of heat carried away by the blood would balance the rate of heat production. This is the fundamental picture of thermal balance in the deep tissues of the body. The amount of blood flow necessary to supply oxygen is high relative to the amount of heat produced, so metabolically active tissues tend to a thermal equilibrium in which heat is carried away by the venous blood as rapidly as it is produced.
Of the crude approximations in this estimate, the worst is the assumption of 100% extraction of oxygen. No tissue reaches this extreme. The maximum possible is closer to 85%. Consequently, the elevation of venous blood temperature relative to arterial temperature need only be a fraction of a degree in order to carry away the heat produced in the tissue perfused.
In a person with normal cardiac output, overall extraction is only 25%, i.e., the mixed venous blood pumped out of the right heart to the lungs retains about 150 ml per liter of oxygen. This is the consequence of the very low extraction fractions in two of the major organ vasculatures (renal, splanchnic) which have a great oversupply of oxygen in relation to their metabolic needs (for a discussion of the organ vasculatures, see Rowell {1839, 1835}).
In short, the convective perfusion of organs is high relative to their heat production, keeping temperatures of the deep tissues near the temperature of arterial blood. This is the physical basis of the near-uniformity of temperatures throughout the deep tissues of the body. From this point of view, arterial blood temperature appears as the best representation of “core” temperature. Insignificant heat transfer occurs between the blood inside the major supply arteries and their surroundings, except in the distal limbs, so arterial blood is delivered to deep structures at a temperature nearly identical to that in the aorta {Afonso 581}. That temperature, in turn is nearly identical to the temperature of mixed venous blood in the pulmonary artery -- despite the heat lost evaporatively from the lungs, pulmonary artery and aortic temperature differ only by hundredths of a degree {Afonso 581}.
Those interested in clinical use of thermal imaging look for surface temperature differences that signal the presence of abnormal tissues below. I am unaware of literature on the metabolism and blood flow of such tissues, but the laws of thermodynamics dictate that the limiting conditions described above apply. To the extent that these tissues differ from surrounding tissues in metabolic rate per unit mass and blood flow rate per unit mass, their equilibrium temperatures will surely differ as well. However, the expected differences are only a fraction of a degree C. The main points of the rest of this article relate to how the physiology and anatomy of the cutaneous vasculature may influence the transmission of these differences to the body surface. But before resuming that tract, consider briefly the situations associated with large temperature discrepancies in the temperature topography of the body.
The big temperature discrepancies relative to central arterial blood temperature are in the periphery, particularly the limbs. Muscle and bone make up a large fraction of their total tissue mass. Under resting conditions, the inactive skeletal muscle has a low perfusion rate in proportion to its low rate of metabolic activity, as does bone. The skin of the limbs can have highly varied blood flow rates. But, in cool resting conditions, rates are similar to the perfusion rate of the underlying muscle. The venous blood returning from skin has been cooled nearly to the temperature of the body surface and may contribute to cooling of deeper tissues traversed by small veins.
A different situation of large discrepancies in the periphery is the thermal picture in the dynamic conditions of exercise. In exercise metabolism increases enormously in large masses of tissue that are generally more superficial. The blood supplied to the active muscle increases (through increase in cardiac output) in proportion to the metabolism increase. Consequently, the temperature increment of venous blood leaving active muscle relative to the arterial blood is limited as described above. However, this warming of a high proportion of a large cardiac output progressively warms the whole blood pool and, in turn, the other organs, until increased dissipation of heat to the environment (via the cutaneous vasculature) matches the increased rate of production.
The large temperature discrepancies persist in a long time constant dynamic after the exercise ceases. The rapid return of blood flow rate to resting levels reduces the perfusion rate to organ mass ratio in the formerly active skeletal muscle. Direct recordings of deep muscle temperature have shown temperature recovering gradually over many tens of minutes {70 Claremont}.
Analyses of heat transfer in perfused tissue speak of microvessels, i.e., those of less than 0.1 mm diameter, as not thermally significant; meaning that blood temperature inside equilibrates with surrounding tissues. For vessels with millimeter or greater diameters, the thermal equilbration length (TEL) is estimated to be greater than the anatomical length. Theoretical treatments along these lines include that by Baish in the previous volume of this series {2085}, the article by Creezee and Langendijk {10}, and the series by Weinbaum, Jiji, and Lemons {364, 1672}. Lemons, et al. {73} made direct measurements in rabbit limbs confirming the predictions that thermal equilibration was complete in vessels of diameters an order of magnitude greater than capillaries.
If vessels that supply skin with arterial blood have long TELs relative to their anatomical length, then they pick up relatively little thermal information, so to speak, from deeper tissues they pass by on the way to the surface. If vessels that travel in the plane of the skin, near the surface, have long TELs, then thermal information supplied to them from below is blurred by being spread out in that plane.
The last part of this article presents anatomical information on the actual size range of vessels in the skin and of arterial vessels that supply the skin, to be considered from the perspective of thermal significance. But the TEL of a vessel also depends upon how rapidly blood flows through it. Accordingly, the range and variability of flow in the vessels of the skin needs to be considered.
In his thermal model of tissue heat transfer, Pennes estimated the flow necessary to supply the nutritive needs of tissues of the forearm as 0.02 ml/min per cubic centimeter of tissue. Various measurements have suggested that this is close to the level present in the skin of a human at rest in neutral thermal conditions {Detry 80}. If so, the two kilogram mass of skin would receive a total blood flow of 40 ml/min, a minor fraction of total cardiac output (normally near 5000 ml/min in a person of this size), but this is below the 250ml/min estimate listed by Williams and Leggett {1420}.
One way to look at the full range of skin blood flow is to record cardiac output in supine resting subjects as they are made hyperthermic. Rowell, et al., used the heat-exchanger undergarment of the Apollo suit to heat their subjects {1236} and measured cardiac output as core temperature rose to 38C and above. Under these conditions, skin blood flow increases due to the combined effects of direct heating of the cutaneous vessels and of the reflex response to the increase in internal temperatures. Blood flow also changes in other organs, but downward except for cardiac muscle. Therefore the measured increase in cardiac output, over 7000 ml/min underestimated the total increase in skin blood flow. This amounts to nearly a 200 fold increase in skin blood flow if we take the resting level as 40 ml/min. For a discussion of cardiovascular changes in hyperthermia see Rowell {1839}.
Skin blood flow has been notoriously difficult to record quantitatively in terms such as ml/min per gram of tissue. Unlike a kidney, say, whose blood flow arrives via a single artery, the boundaries of skin as an organ are not clear, and its blood is delivered via countless arterial twigs that branch from main supply arteries all over the body.
The only non-invasive quantitative method for recording skin
blood flow has been venous occlusion plethysmography (vop). The laser Doppler technique has been
valuable for understanding mechanisms of control of skin blood flow, but never
has been able to yield quantitative figures except through reference to numbers
obtained from vop. The flow axes in
Figures A and B are vop data. The
early adopters of vop chose to express their data in terms of ml/min per 100
ml. Unfortunately, the 100ml reference
quantity refers to the whole limb, not to the quantity of skin. To see why, consider briefly the vop
technique, illustrated in Fig. 2.
The rationale of vop is that a partially inflated blood pressure cuff around the upper arm (or thigh for leg vop) will prevent venous blood from leaving the limb without interfering with arterial inflow. Consequently, the portions of the limb distal to the cuff will swell as blood accumulates. This works quite well with the blood pressure cuff inflated to just below the diastolic arterial pressure. The original vop technology relied on measuring the rate of swelling by displacement of volume from a water-filled chamber surrounding the limb. Later, gauges were used to record changes in limb circumference as the limb expanded. Either way, the geometry forces one to express the changes in terms of the starting volume; thus we see the scales labeled as ml/min per 100 ml in the illustrations of vop data.
Fig. 3 schematically represents vop data obtained in a
classical study that quantified the influence on forearm blood flow of local
temperature {496}. In 1943, Barcroft
and Edholm used a water-filled chamber to record the rate of limb
expansion. When water temperature was
above 35C, limb flow rate increased steadily over tens of minutes. The higher the water temperature, the
greater was the eventual flow. A
temperature of 42C is sufficient to obtain a maximal response; higher levels
produce little or no further flow increase and infringe on the range of thermal
damage. The apparent use of 45C in the
Barcroft and Edholm work refers to temperature in an outer chamber of their
water displacement plethysmograph; undoubtedly greater than the skin surface
temperature.
Much later, this phenomenon was re-studied to assess the reproducibility of the response at 42 C, the temperature at which the maximum response is obtained, and to learn more of the variability within subjects {Savage reproducibility 1549}. Subjects’ forearms were sprayed for an hour with 42C water. Limb blood flow, recorded by the vop technology illustrated in Fig. 2, rose from initial values in the 2-6 ml/min per 100 ml to the levels indicated in Fig. 4, different for each subject but more or less reproducible on different occasions in a given subject. Flow began to increase immediately when the water spray began and was usually near peak levels within twenty minutes. At the time of this study, this was known to be the direct response of the cutaneous vasculature to local heating, i.e., the entire increase in flow to the limb occurred via the cutaneous vasculature (see below).
The point of Fig. 2 is to illustrate maximal levels of the cutaneous vascular response. Similar peak levels of flow have been recorded in hyperthermic individuals with much lower surface temperatures on the arm from which the vop data was obtained. This response is mediated by an active vasodilator system, i.e., if the sympathetic nerves are blocked, blood flow falls to resting levels. Both this response and the direct response to heating are extraordinarily developed in human skin and are absent or of a different character in other species.
Besides these mechanisms related to thermal variables, skin blood flow is also under the influence of a separate functional capability of the sympathetic nervous system. Reflex vasoconstriction is modulated in response to cardiovascular reflexes. This adjustment of smooth muscle tension in the resistance vessels of skin is mediated through release from the sympathetic fibers of the neurotransmitter norepinephrine, and is the dominant means of control of blood flow in the other organ systems. Also, of course, skin blood flow is altered when substances that alter vascular tension circulate in the blood, e.g., epinephrine. For a comprehensive treatment of the mechanisms of control of blood flow in human skin, see Rowell {1839}.
The intent here is only to discuss the range of blood flow within human skin, not to dwell on the mechanisms. At first glance, vop data such as that in Figs. 3 and 4, puts the upper limit at roughly 20 ml/min per 100 ml. But this number needs closer examination.
First, these increases in limb blood flow are known to be due entirely to change in the vessels of the skin, with no contribution from the much larger tissue mass below, muscle.
Though we may think that hot pads applied to the skin over a sore muscle improve the circulation, the blood flow changes that have been recorded in heated skeletal muscle are comparatively small. Fig. 5 is from a paper cited as showing that heat increases blood flow in skeletal muscle. But the flow increases recorded were associated with temperatures above 42C, well above the levels that produce maximal dilation in skin. In another figure in the same paper, with maintenance of muscle temperature near 42C by diathermy, maximal flow was 8 ml/min per 100 ml {1132 Sekins}. This is trivial compared to the increases of which the cutaneous vasculature is capable.
Another line of evidence showing that muscle flow does not increase in reflex response to heating of the overlying skin nor in response to its own temperature are the findings obtained when skin blood flow was arrested. For example, in hyperthermic subjects with skin blood flow blocked in one forearm by epinephrine, vop revealed little or no flow increase in the blocked arm, despite a fully developed vasodilator response in the opposite arm {Detry 80 Edholm Fox; 2086}. Incidentally, at the high core temperatures these subjects reached, the muscle of the blocked arm would have been heated through elevation of the incoming arterial blood temperature. The lack of flow increase is another indication of the weak direct influence of temperature on the resistance vessels of the vasculature of skeletal muscle.
So, we may look at Fig. 4 and regard the flow increase, relative to baseline levels, as due to flow change within the cutaneous vasculature. Unfortunately, this has been taken as the basis of estimates of maximal skin blood flow as near. 20 ml/min per 100 ml of tissue. These overlook the fact that the flow increase should be related not to 100 ml of arm but to the volume of skin enclosed by the plethysmograph.
The tissue enclosed by the volume chamber (plethysmograph) or the cross sectional area enclosed by circumference gauges was that of the whole limb, bone, muscle, connective tissue, and skin. Overall skin thickness may be a millimeter or so in some parts of the body, and as much as two elsewhere {Moretti 2041}. For the typical forearm circumference of subjects from the study illustrated in Fig. 4, the fraction of the arm volume comprised by skin would be 5 to 10%, depending on whether the skin layer is taken as 1 or 2 mm.
If the volume changes registered by vop reflect flow increase limited to the skin, then the blood flow change, in terms of ml/min per 100 ml of skin is properly described by multiplying the vop numbers by the inverse of the fraction that skin makes up of the total limb. If that factor is 20, then the 20 ml/min per 100 ml of limb reflects an increase in SkBF of 400 ml/min per 100 ml of skin, putting skin close to kidney in terms of potential flow per unit mass of tissue. Interestingly enough, if total skin volume amounts to approximately 2000 ml, this peak figure corresponds to a cardiac output increase of 8000 ml/min, in reasonable correspondence to values that have been recorded in hyperthermic supine resting subjects {1839}.
Ordinarily, clinicians interested in thermal imaging would be working with individuals who are not under thermal stress. People adjust their behavior and clothing to remain in the so-called “neutral” zone of thermoregulation. Within this zone, one feels neutral, thermally. Skin temperature, averaged over the whole surface, ranges not much beyond a 33 to 35C. But a considerable range of skin blood flow is possible, in fact necessary, within the limits of the neutral zone.
Skin temperature at 33 as opposed to 35C corresponds to a substantial change relative to a core temperature near 37C. Skin blood flow is altered in reflex response to the necessity to deal with the present core-surface temperature difference. Not only is the environmental temperature a factor in skin blood flow in these conditions, but also recent history, e.g., elevated core temperature due to recent exercise.
Savage and Brengelmann showed examples of the changes in skin blood flow possible within the neutral zone. Limb blood flow, recorded by vop, changed roughly 2 ml/min per 100 ml when skin temperature was driven between 33 and 35C{Savage 1796). Taking the proportion of skin as 5%, the vop data showing a blood flow increase of 2 ml/min per 100 ml of limb would correspond to skin blood flow increase of 40 ml/min per 100 ml of skin, a huge increase over baseline nutritive levels. A slight increase in core temperature would have a much bigger effect.
Quite possibly, then, a person studied with thermal imaging under what are thought to be neutral conditions could be in a state of skin blood flow high enough to interfere with the resolution of the method.
Given the estimates above of maximal flows during active vasodilation or direct heating, and taking minimal nutritive levels as near 2 ml/min per 100 ml, we have skin blood flow potentially ranging over a 200:1 scale. A full traverse of this range would entail warming an initially cool a person to the core temperature above 38C, or heating a patch of initially cool skin up to 42C.
But, even the relatively minor skin and core temperature changes associated with thermoregulation in the neutral zone elicit blood flow changes between bare nutritive levels and factors of tens above.
Thermal equilibration lengths and the size range of vessels that have insignificant heat exchange with their surroundings must vary widely along with skin blood flow. My intuition leads me to expect that the best chance of observing presence of an abnormal structure below the skin through a temperature contour on the skin would be under conditions of minimal skin blood flow. Such intuitive notions can be far off the mark, but perhaps manipulation of skin blood flow should be considered as a possible way of altering transmission of thermal information to the surface.
The fascinating architecture of the microvessels of the skin was described early on by Spalteholz. For a dramatic graphical rendering, see the comprehensive article on cutaneous anatomy by Moretti {2041}. Particularly impressive are the multiply interconnected meshworks of vessels in planes parallel to the skin, within about a millimeter of the surface. .According to theoretical estimates and the direct measurements of Lemons et al {73}, vessels in this size range would be categorized as thermally insignificant, in that effective heat transfer prevents any temperature difference between the blood inside and the surrounding tissue. Of course, in another sense, they are highly significant, minimizing temperature gradients in the plane parallel to the skin as well as in the outer millimeter or so of the plane perpendicular to the skin.
The concern here is with the larger vessels that feed the cutaneous microvasculature, vessels well above 0.1 mm in diameter. Obviously skin, as an organ, does not deceive its arterial supply through major arteries identifiable as allocated to its functions and in the size range of, say, the renal arteries. But neither is it true that the skin’s arterial supply is delivered exclusively by tiny arterial twigs that are many generations derived from offshoots of a deep artery in a branching pattern in the plane perpendicular to the skin with each generation progressively smaller in diameter along the ascent to the skin. Instead, the architecture is more like that illustrated in Fig. 6. Robust arterial vessels called perforators make the connection between deep supply vessels and the arterial meshwork within the skin.
For the modern view of the arterial supply of the skin, we are indebted to plastic surgeons. They were motivated to find large perforators they could depend upon as the arterial supply for a patch of skin removed from its original underpinnings and grafted over a wound. For a comprehensive treatment of the literature, fascinating illustrations of the rich distribution of perforators and for renderings of the various types of perforators, see “The Arterial Anatomy of Skin Flaps”, by Cormack and Lamberty {2081}.
Besides this insight into the distribution of perforators, seen in the plane more or less perpendicular to the skin, we also have the amazing images obtained by Salmon of arterial vessels in the plane of the skin, made available in English in 1988 {2083}. A sample is shown in Fig. 7.
The images published by Salmon were obtained from cadavers by infusing a radio-opaque substance, removing the skin, and making X-ray photographs. They reveal a complex meshwork in the plane of the skin of vessels in the fractional millimeter diameter range. In comparing these images with the maps of perforators described above one can visualize the connections at occurring at the major intersections in the meshwork revealed in Salmon’s radiographs. Cormack and Lamberty indicate that connections from below also intercept the larger vessels shown in Fig. 7 along their length {2081}.
Plastic surgeons have recently employed thermal imaging to locate perforators {2084}. They deliberately manipulated skin blood flow to do so. They applied ice packs to reduce flow. The uniform skin temperature present just after removal of the ice pack was soon replaced by a field mottled by freckle-sized spots, interpreted as the effect of recovering blood supply via perforators.
The classical Spalteholz description of a dense meshwork of microvessels, below 0.1 mm diameter, inclines one to see the skin, thermally, as a homogenous medium, albeit well-perfused. If this view is combined with a notion that the cutaneous microvessels are supplied from below by tiny arterial vessels that are themselves thermally insignificant, i.e., in which blood temperature is equilibrated with the surroundings, then one can imagine skin as a simple conductive layer that would preserve thermal contours of underlying thermal anomalies.
But the cutaneous vascular architecture revealed in images such as those of Comack and Lamberty and Salmon {2081, 2083} does not permit this view. They show the skin to be supplied by arterial supply vessels with a size range up to substantial fractions of a millimeter, above that in which equilibration would be expected.
At the very highest rates of flow possible within the skin, the perforator vessels must be delivering arterial blood within the skin at temperatures nearly identical to that in the major supply arteries below, unaffected by the temperatures of tissues that the perforators pass by. This thermal “significance” must persist in the large vessels in the meshwork in the plane of the skin that receive flow from the perforators (Fig. 7), surely complicating any transmission of thermal anomalies from below.
Whether this would be true at lower flow rates is not clear, but recall the point made above, that skin blood flow rates even in the neutral range of thermoregulation are far above the nutritive levels typically assumed in thermal models. Perhaps the approach of the Japanese surgeons who used ice packs to shut off skin blood flow {2084} has potential in other areas of clinical thermal imaging.
This all boils down to a wish that those interested in applying thermal imaging clinically will consider the peculiar physiology of flow variability in the cutaneous vasculature and possibly make use of anatomical information such as that in the publications of Salmon and of Comack and Lamberty {2081, 2083}.