The technique of infrared imaging in medicine*

The room used for thermal imaging must meet certain basic requirements. These are: adequate size for working— where up to 2 m may be needed between the patient and the subject—and adequate space to locate the image processing equipment and space for one or more patient cubicles. A rough indicator for the least distance in one direction can be derived from the optical features of the lens, i.e. the distance between the camera and the patient to take an image of the upper or lower part of the human body or of an object of 1.2 m height. This means that a minimum size of 2 m × 3 m is required, but that a larger room, 3 m × 4 m or more, is preferable.

This is a primary requirement for most clinical applications of thermal imaging. A range of temperatures from 18 °C to 25 °C should be attainable and held for at least 1 h at an accuracy better than 1 °C. At lower temperatures the subject is likely to shiver and an over 25 °C room temperature will cause sweating, at least in most European countries. Variations may be expected in colder or warmer climates, in the latter case, room temperatures may need to be 1–2 °C higher [7].

The type of examination used will determine the ideal ambient temperature. Many clinical examinations are performed with partial disrobing of the patient. When larger areas of the body are unclothed and exposed to the air for longer periods, lower ambient temperatures will cause discomfort [8, 9] and may result in reflex vasoconstriction. Inflammatory lesions are more clearly visualized in a cool environment, typically 20 °C. Examination of the extremities, where the sympathetic nervous system influences the result, a warmer ambient temperature of 22–24 °C is generally recommended.

Additional techniques for cooling particular regions of the body have been developed [10, 11] (e.g. water immersion of the hands in a stress test).

Indication of the air temperature is important, a large digital display which is visible anywhere in the room should be used. Air temperature is affected not only by the heat generated by electronic equipment, but also by the human body. For this reason the air-conditioning unit should be capable of compensating for the maximum number of patients and staff likely to be in the room at any one time. These effects will be greater in a small room of 2 m × 3 m or less. Air-conditioning equipment should be located so that direct draughts are not directed at the patient and that overall air speed is kept as low as possible. A suspended perforated ceiling with ducts diffusing the air distribution evenly over the room is ideal [12], although more expensive to construct, than simple wall mounted systems.

Image processing equipment needs space located away from the patient area, to avoid heat disturbance. A sink with a water supply is often required for water stress tests. Furniture may also include a multi-position chair with attached leg rests and a bowl or table on castors at a suitable height for a sitting patient.

Finally a cubicle or cubicles within the temperature controlled area are essential. These should provide privacy for disrobing and a suitable area for resting through the acclimatization period.

A new generation of infrared cameras have become available for medical imaging. The older systems, normally single-element detectors using an optical mechanical scanning process, were mostly cooled by the addition of liquid nitrogen [13–15]. This had the effect of limiting the angle at which the camera could be used, which restricted operation. Electronic cooling systems were then introduced, which overcame this problem. The latest generation of focal plane array cameras can be used without cooling, providing almost maintenance-free technology [16].

Almost all systems now use image processing techniques and provide basic quantitation of the image [17–19]. In some cases this may be operated from a chip within the camera, or may be carried out through an on-line or off-line computer. For older equipment such as the AGA 680 series, several hardware adaptations have been reported to achieve quantitation of the thermograms [20–22].

Earlier reports stipulated the requirement for a separate thermal reference source for calibration checks on the camera [23]. Many systems now include an internal reference temperature, with manufacturers claiming that external checks are not required. Unless frequent servicing is obtained, it is still advisable to use an external source, if only to check for drift in the temperature sensitivity of the camera. An external reference, which may be purchased or constructed, can be switched on with the equipment and left running throughout the day. This allows the operator to make checks on the camera and, in particular, provides a check on the hardware and software employed for processing. These constant temperature source checks may be the only satisfactory way of proving the reliability of temperature measurements made from the thermogram [24].

A camera stand which provides vertical height adjustment is very important for medical thermography. Photographic tripod stands are inconvenient for frequent adjustments and often result in tilting the camera at an undefined angle to the patient. This is difficult to reproduce and, unless the patient is positioned so that the surface scanned is aligned at 90° to the camera lens, distortion of the image is unavoidable. Studio camera stands are ideal, they provide vertical height adjustment with counterbalance weight compensation. They are stable with a weighted base on wheels which enables the operator to rapidly set up the camera in any reproducible position. Most stands can hold the camera to within 10 cm from the floor (the patient can also stand on a low stool to avoid parallax even in a standing position). The maximum height required will depend on the use of a low couch, or whether all positioning is achieved with the patient standing or sitting in a chair. A pillar height of 2 m or 2.5 m can be used. It should be noted that the type of lens used on the camera will affect the working distance and the field of view, a wide angle lens reduces distance between the camera and the subject in many cases, but may also increase peripheral distortion of the image. The ceiling mounted stands used in radiology and nuclear medicine can also be used, but are likely to require positioning motors if the camera is beyond the operators reach.

The start-up times with modern cameras are claimed to be very short, minutes or seconds. However, the speed with which the image becomes visible is not an indication of image stability. Checks on calibration will usually show that a much longer period from 10 min to several hours with an uncooled system are needed to achieve optimum conditions for quantitative imaging [4, 25].

Software packages for thermal imaging are provided by some manufacturers, few of which are specifically designed for medical applications [17, 18]. One specialized software system designed for medical use meets international standards when used according to recommended techniques [26]. Archiving of both images and relevant clinical data is an important requirement for medical thermography.

Human skin temperature is the product of heat dissipated from the vessels and organs within the body, and the effect of environmental factors on heat loss or gain. There are a number of further influences which are controllable, such as cosmetics [27], alcohol intake [28] and smoking [29–31]. These should form part of the request made to the patient when calling him or her for examination. In general terms the patient attending the examination should be advised to avoid all topical applications such as ointments [32, 33] and cosmetics on the day of the examination to all the relevant areas of the body. Large meals and above-average intake of tea or coffee should also be avoided, although studies supporting this recommendation are hard to find and the results are not conclusive [34–36]. Patients should be asked to avoid tight fitting clothing and to keep physical exertion to a minimum. This particularly applies to methods of physiotherapy such as electrotherapy [37–39], ultrasound [40], heat treatment [41–43], cryotherapy [43–46], massage [47–49] and hydrotherapy [50–52] because the thermal effects from such treatments can last for 4–6 h under certain conditions. Heat production by muscular exercise is a well-documented phenomenon [53–57].

Drug treatments can also affect the skin temperature. This phenomenon was used to evaluate the therapeutic effects of medicaments [3]. Drugs affecting the cardiovascular system [58–61] must be reported to the thermographer in order that the correct interpretation of thermal images will be given.

On arrival at the department, the patient should be informed of the examination procedure, instructed to remove appropriate clothing and jewellery, and asked to sit or rest in the preparation cubicle for a fixed time. The time required to achieve adequate stability in blood pressure and skin temperature is generally considered to be 15 min, with 10 min as a minimum [62–64]. After 30 min of cooling, oscillations of the skin temperature can be detected in different regions of the body with different amplitudes resulting in a temperature asymmetry between the left and right sides [63]. Some evidence has been found for a circadian rhythm of both core and skin temperatures [20, 34, 65, 66]. During this preparation time the patient must avoid folding or crossing their arms or legs, or placing their bare feet on a cold surface. If the lower extremities are to be examined, a stool or leg rest should be provided to avoid direct contact with the floor [67]. During the examination paper or linen towels may be required to avoid overcooling of the feet.

As in radiology, it is preferable to standardize on a series of standard views for each body region. The EAT Locomotor Diseases Group recommendations include a triangular marker system to indicate anterior, posterior, lateral and angled views [2, 68]. Modern image processing software provides comment boxes which can be used to encode the angle of view which will be stored with the image [26]. It should be noted that the position of the patient for scanning and in preparation must be constant. Standing, sitting or lying down affects the surface area of the body exposed to the ambient temperature, therefore an image recorded with the patient in a sitting position may not be comparable with one recorded on a separate occasion in a standing position.

Image size is dependent on the distance between the camera and the patient and the focal length of the infrared camera lens. The lens is generally fixed on most medical systems, so it is good practice to maintain a constant distance from the patient for each view in order to acquire a reproducible field of view for the image. For example a single hand may be recorded in a 20 cm × 20 cm field of view, while a picture of both lower limbs from knees to ankles may require a 50 cm × 50 cm field. If deviations in camera angle from the normal position on the stand are required, these should be recorded for future use. If different thermograms with different fields are compared for the same subject, the variable resolution can lead to false temperature readings [69]. Most hospitals now use a standard format for patient identification and demographic detail. As much detail as possible should be recorded with the thermogram to avoid misidentification. The time and date of the examination should also be recorded, which most computer software will provide as standard [70].

Individual software programmes now largely dictate the layout for a clinical report. This will normally consist of the images, the demographic data and any measurements made from image processing. Every image or block of images must carry an indication of temperature range, with colour code/temperature scale. The colour scale itself should be standardized. Industrial software frequently provides a grey-scale picture and one or more colour scales. The default colour scale will often show white as hot, then yellow and then red, following the hot metal scale. The so-called rainbow or spectral order of colours is more widely recognized, especially by colleagues who are not used to the other colour scales used by engineers. A false colour finely graded scale is also possible, this can range from dark blue at the cold end through green to red and will convey a similar degree of image contrast to a monochrome black and white picture.

Background temperatures which can obscure the clinical image should be avoided and cleaning the image by processing, e.g. squeezing the temperature range or overwriting the lower temperatures with a background of white, grey or black, will improve the visual presentation. Care should be taken when the pictures are recorded to minimize these problems. The use of hardboard or cold towels arranged just prior to image recording will often improve the image clarity. Regions of interest, spot measurements sites, etc, should be indicated and, if these mask the clinical picture, a separate image without these processing indicators should be provided. It is important to note that while a high number of colour shades can be displayed on a computer monitor screen, most printers to date are less able to reproduce all the fine detail. Settings for the image on the screen should be tested on the printer to ensure that subtle differences in the image which are important are not lost in a lower contrast printed image.

Computer file archives are now commonplace and are a valuable reference for repeated investigations on the same patient. A multiple window facility in the software will allow the operator to recall a series of earlier pictures for comparison and to ensure that the same positions and temperature settings have been used. A matrix of 4 × 4 images is adequate with the possibility to zoom in on any combination of frames for the report [24]. The numerical data relating to each image must be clearly identified with the original image to which they relate. If a standardized challenge test such as a cold stress test for Raynaud's phenomenon has been used, the relevant images should be printed, preferably to show the pre- and post-stress thermograms together with the temperature data extracted from them. Where normal values for indices [71–73] or temperature values [74, 75] are known, these should be included in the report.