30-07-2012, 12:49 PM
Methods of Tissue Temperature Measurement
Methods of Tissue Temperature Measurement.pdf (Size: 2.62 MB / Downloads: 68)
In this chapter various techniques for the non-invasive monitoring of tissue temperature
are reviewed, including the proposed NIRS method, highlighting their relative advantages
and disadvantages. The emphasis is on the new-born human brain, since, as discussed
in Section 1.1, one of the major applications of the work developed in this project is
the intended monitoring of cerebral temperature during hypothermic treatment in birthasphyxiated
infants.
Section 4.1 discusses the definition and measurement of core body temperature by
conventional and other non-invasive methods. It can also be important to measure temperature
in a particular part of the body or even a specific volume of tissue. As discussed
in Section 1.2.3, one example of this is the case of selective head cooling, in which cerebral
temperature may significantly differ from that in the rest of the body. Therefore, Section
4.2 describes methods in which non-invasive temperature measurement can be applied to
localised tissue areas, again with particular focus on the infant brain. Finally, Section
4.3 discusses the feasibility of using NIRS to monitor tissue temperature non-invasively,
describing how temperature affects the optical properties of tissue and reviewing studies
in which the temperature-dependence of NIRS tissue measurements is investigated.
73
4.1 Measurement of Core Body Temperature
Deep or ‘core’ body temperature refers to that of the tissues deep within the skull or the
thoracic and abdominal cavities, whereas surface or ‘shell’ temperature refers to the outer
body tissues, essentially the skin. Core body temperature is regulated mainly by the hypothalumus
in the brain, which responds to input from thermoreceptors located centrally
in the body core and peripherally in the skin. The pulmonary artery is considered to be
the ‘gold standard’ measurement site for core body temperature (Fulbrook, 1997). However,
the use of this site is rarely an option in normal clinical practice, and other methods
of temperature measurement are generally employed. There has been much debate about
the most reliable site from which to monitor body temperature in children (Haddock et al,
1996; Robinson et al, 1998; Craig et al, 2000). The majority of the studies conclude that
axillary temperature measurement (from the armpit), whilst safe and easy to perform, is
not sensitive enough to detect fever in infants, and hence rectal temperature measurement
should be used. Rectal temperature has most commonly been monitored by conventional
mercury-glass thermometers. The use of mercury thermometers in neonatal wards has,
however, also been under recent debate, due to safety and cost issues (Sganga et al, 2000),
and the length of time required to achieve an accurate measurement (Craig et al, 2000).
Sganga et al (2000) condone the use of digital thermometers in healthy neonates as a
cheaper alternative to mercury-glass thermometers without the associated risks. Whilst
these methods can provide clinicians with a more or less non-invasive measurement of
body temperature, they do not lend themselves to measurement of temperature within a
particular area of the body, e.g. the brain. The following sub-sections describe two other
methods of measuring core temperature non-invasively, which have specifically been applied
to monitoring cerebral tissue temperature: tympanic thermometry and the zero heat
flow method.
4.1.1 Tympanic Membrane Thermometry
In recent years the tympanic membrane has become a popular temperature measuring site,
in particular with the use of infrared thermometry (Nobel, 1992). The tympanic membrane
is the boundary between the outer and middle ears, commonly known as the
eardrum. Due to its close proximity to the carotid artery, temperature at the TM should
mirror core body temperature well. Ferrara-Love (1991) reports no significant difference
between temperatures measured in post-operative patients using a TM thermometer and
74
via a pulmonary artery catheter. TM temperature is also thought to accurately represent
core cerebral temperature (Childs et al, 1999), which could differ from core trunk
temperature, e.g. during brain hypothermia.
An extensive number of studies have compared brain and TM temperatures by a variety
of methods under differing conditions. Mariak et al (1994) compared invasive brain
temperature measurements with those made at other externally accessible sites during
open-brain surgery in humans. They concluded that TM temperature gave the best approximation
of the average cerebral temperature. During mild hypothermia (35 ◦C) in
piglets, Haaland et al (1996) reported that TM temperature was a better indicator of
brain temperature than colonic temperature. However, they noted that during moderate
hypothermia (29 ◦C) and rapid changes in temperature the tympanic-cerebro correlation
was poorer, an observation also made by Stone et al (1995) during studies of heart surgery
patients. There are further studies that conclude that TM temperature does not give a
good indication of changes in brain temperature (Shiraki et al, 1988). Sato et al (1996)
suggest that the discrepancy between core and TM temperatures is due to poor thermal
insulation of the measuring probe from the surrounding ear canal and poor contact between
the probe and the membrane itself. Contact thermistors or thermocouples are not,
however, suitable in the majority of clinical cases and have associated risks such as injury
to the delicate TM and the spread of infection.
Infrared tympanic thermometry (ITT), which has become increasingly popular in hospitals
since the late 1980’s, has been found to provide fast, accurate temperature readings
at relatively low cost (Shinozaki et al, 1988). The non-contact thermometers detect infrared
radiation emitted from the TM and calculate an equivalent black-body temperature.
Since the TM is optically insulated from the outside environment, due to the shape of the
ear canal, it can be approximated as a black-body. In a comparison study between contact
and non-contact (i.e. infrared) TM temperature measurements, Terndrup et al (1997) indicated
that the infrared thermometer gave an accurate estimate of TM, and hence core,
temperature in healthy children. In contrast, many neonatal and paediatric studies have
shown infrared tympanic thermometers (ITTs) to be inaccurate, or insensitive to temperature
changes, when compared to other core-estimating techniques (Muma et al, 1991;
Yetman et al, 1993; Lanham et al, 1999). Childs et al (1999) suggest that conflicting
results from studies evaluating the accuracy of ITTs could mainly be due to probe positioning.
If the sensor is not correctly positioned it may detect radiation emitted from
the ear canal and not the TM. Since the ear canal is not in equilibrium with the arterial
75
supply and is more readily affected by environmental temperature this may give rise to an
erroneous reading of core body temperature.
A further point made by Shibasaki et al (1998) is that current ITTs are not capable
of monitoring temperature continuously, which is especially important in the case of
hypothermic treatment. They have developed a method of continuously measuring TM
temperature using an optical fibre and infrared detector, which showed no significant differences
from TM temperature measured by a contact thermistor during exercise and face
fanning. The adult volunteers gave an indication of when they believed the fibre to be
close to the TM, based on the amplitude of the sound, which travelled down the fibre,
of a mechanical chopper used in the set-up. As the method still involves insertion of the
probe (the fibre in this case) into the ear canal, however, it is unlikely to be suitable for
continuous monitoring of brain temperature in new-born infants.
4.1.2 The Zero Heat Flow Method
Another method of measuring core temperature is based on the ‘zero heat flow’ principle.
The assumption is made that if heat loss from the surface of the body is reduced to zero,
the gradient between core and surface temperature will also tend towards zero. When an
equilibrium has been reached there should be a region of tissue below the surface which will
be at a uniform temperature. The depth to which the tissue reaches a uniform temperature
depends on the size of the area over which the surface is insulated. A number of studies
have reported the use of the zero heat flow method in monitoring tissue temperature in
new-born infants. These studies have taken advantage of the high insulation properties
of the infant mattresses used in neonatal wards and intensive care incubators in order to
achieve the zero heat flow condition. Simbruner et al (1994) investigated the relationship
between brain temperature and cerebral metabolism or circulation in neonates, using the
zero heat flow method to determine temperature within the head. The zero heat flow
condition was achieved by turning the head to one side and positioning the monitoring
thermistor between the infants temple and the incubator mattress. They believed the
method to be accurate and precise enough to determine small temperature differences
between infants with normal or abnormal cerebral tissue or haemodynamics, although
they also suggested a validation of these results with a more direct or invasive method.
The Deep Body Thermometer (DBT), first developed by Fox and Solman (1971),
achieves the zero heat flow condition by heating the skin surface such that outward heat
flow from the core is matched by inward heat flow from the surface. The thermometer
76
Figure 4.1: Schematic diagram of the deep body thermometer probe, based on the zero
heat flow method, developed by Fox and Solman (1971).
probe, shown in Figure 4.1 (Solman and Dalton, 1973), is approximately 6 cm in diameter
and consists of two temperature sensors separated by an insulating pad, which also houses
a heating element. The temperatures of the sensors, one of which is in contact with
the skin surface, are compared by an operational amplifier, the output of which controls
the heating element, switching it on or off depending on the direction of the temperature
gradient between the two sensors (Solman and Dalton, 1973). When the sensors eventually
reach the same temperature, no additional heat can flow out through the pad and the ‘deep
body’ temperature is inferred by the temperature of the skin sensor.
Following its development some thirty years ago, many studies were done to assess the
accuracy of the DBT. In a study by Singer and Lipton (1975), a close correlation was
found between the temperature measured by the DBT, applied to the upper sternum, and
the temperature of the tympanic membrane. The upper sternum has been recommended
as a suitable measuring site (Fox and Solman, 1971) since it is relatively flat, has little
subcutaneous fat and is close to the large systemic blood vessels which will be at core
temperature. Togawa (1979) comments that the forehead is also a convenient measuring
site for the thermometer probe and cites findings from a study by Tsuji et al (1976) in
which temperature measured at the forehead by the DBT probe was seen to correlate
closely to blood temperature in the jugular vein. A modified probe, developed by Togawa
et al (1976), incorporates an aluminium casing (approximately 2 cm in diameter) around
a smaller version of the original probe, which minimises radial heat flow from the probe
centre and thus improves the accuracy of the DBT. The DBT response time, i.e. the time
taken for the initial equilibrium temperature to be reached, is reported to be between 15
and 20 minutes (Togawa, 1985), or slightly less if the probe is allowed to heat up to 37 ◦C
prior to application (Solman and Dalton, 1973). The temperatures measured by this probe
77
have been shown to reflect the temperature at a depth of 18 mm or greater below the skin
surface, i.e. the core temperature (Matsukawa et al, 1996).
A relevant point to consider is whether the zero heat flow method (achieved via the
DBT or other method) could be used to monitor tissue temperature during mild hypothermia.
In a study of patients undergoing cardiopulmonary bypass and moderate hypothermia
(25–28 ◦C), the modified DBT was used to monitor deep brain temperature at the
forehead (Muravchick, 1983). The author concluded that the DBT was not suitable for
monitoring the rapid changes in temperature seen during cardiopulmonary bypass. This
hypothesis was supported in a review by Togawa (1985), who suggests that zero heat flow
measurement at either the forehead or the torso is a reliable method of measuring deep
body temperature except during rapid cooling or warming. Conversely, in a recent study
(Harioka et al, 2000), temperature monitored by the DBT at the (adult) forehead was
seen to correlate well with blood temperature measured in the pulmonary artery during
intraoperative mild hypothermia (∼34 ◦C), although the authors noted that the probe
was allowed to equilibrate for 20 minutes prior to induction of anaesthesia. Dollberg et al
(1993) point out that the zero heat flow method relies on heat conduction to the surface,
but that heat is also transferred by the circulation and the two forms of heat flow are not
independent. Thus if heat flow via the circulation were to change during hypothermia, for
example due to vasoconstriction, this would affect heat flow to the surface and hence the
time response of the measuring system. A potentially serious hazard is the possibility that
the imposed heat insulation at the skin surface may in fact alter cerebral tissue temperature
during the treatment (Togawa, 1979). Gunn and Gunn (1996) investigated the effect
of a radiant heater on the temperature gradient in the head of new-borns in a neonatal
intensive care unit, using a similar zero heat flow method to Simbruner et al (1994). They
had expressed concern that since brain temperature depends partly upon radiative heat
loss from the scalp, which in turn depends on skin and ambient temperatures, the use
of overhead heaters could in fact reduce the natural core to surface temperature gradient
in the head, thus increasing brain temperature. It has been shown, in adult rats, that
an increase in body temperature of just a few degrees can significantly affect the extent
of neuronal damage following an ischaemic insult (Baena et al, 1997). In their study
Gunn and Gunn (1996) found no significant differences between core head and core body
temperatures (as measured by the zero heat method) and rectal temperature when the
overhead heater was switched on, although scalp temperature was seen to rise sharply.
They concluded that the subsequent reduction in core to surface temperature gradient in
78
the head must affect the temperature of intermediate or superficial brain structures. This
alteration in temperature of the surface tissues would not be measured by the zero heat
flow method, as indicated by the findings of the study (Gunn and Gunn, 1996), since it
represents the deep body or head temperature.
4.2 Non-Invasive Brain Temperature Measurement
This section will discuss non-invasive methods of brain temperature measurement that
are currently being researched and developed world-wide in academic and industrial institutions.
These include microwave radiometry, magnetic resonance thermometry and
ultrasound thermometry. Many studies based on these methods have investigated the
application of remote thermometry to hyperthermic oncological treatments (Carter et al,
1998; Dubois et al, 1993). The discussion here, however, will focus if possible on thermometry
in hypothermic therapies, in particular in the neonatal brain. The advantages and
disadvantages of the methods, with regard to the specific application, are summarised.
4.2.1 Microwave Radiometry
The technique of microwave radiometry in the application to non-invasive thermometry is
based on the detection of electromagnetic radiation in the microwave region emitted from
a ‘hot’ body (Foster and Cheever, 1992; Leroy et al, 1998). Any material body at a temperature
above that of absolute zero will emit thermal radiation, the spectrum of which
is directly related to the temperature of the body. For living human tissue, at a temperature
of around 37 ◦C, spectral emission is most intense in the mid-infrared at about 9.7
μm (Togawa, 1985). At a microwave frequency of 3 GHz (10 cm wavelength), a typical
frequency used in microwave radiometry (Leroy et al, 1998), the emitted intensity is approximately
108 times less than that at the peak (Togawa, 1985). However, with sensitive
enough equipment this radiation can be detected and a temperature resolution of 0.1 ◦C
achieved (Edrich et al, 1980). At microwave wavelengths the penetration depth through
human tissue is the order of several centimetres, much deeper than the few millimetres
or less characteristic of the mid-infrared (Godik and Gulyaev, 1991). Spatial resolution
is, however, compromised by the use of longer wavelengths: at a wavelength of 3 cm the
spatial resolution has been reported to be about 2 cm (Edrich et al, 1980).
Microwave radiometers detect the average power emitted from a particular volume
of the medium beneath the detecting antenna. The received signal depends on many
79
factors, such as the temperature distribution in the measured volume and its dielectric
properties (Cheever and Foster, 1992). The effective depth of sensing, i.e. the depth at
which the signal is e−1 that at the surface, depends on the antenna geometry: a larger
antenna increases the effective depth at the cost of resolution (Cheever and Foster, 1992).
The radiometric temperature (known as the brightness temperature) is determined by
a complex two-step process (Leroy et al, 1998). Initially, the received radiation field
pattern of the antenna is estimated by assuming it to be the same as the absorbed field
distribution seen in the medium were the antenna to be used to transmit energy into the
medium (Foster and Cheever, 1992). Subsequently, the temperature is found by applying
an inversion process to the radiometric data, which has been weighted by the antenna field
distribution. There are many inversion techniques used for recovery of the temperature
distribution in biological tissue (Foster and Cheever, 1992). Some require the use of multifrequency
radiometers (Mizushina et al, 1992), whilst others depend on a fixed-frequency,
multi-angle method (Montreuil and Nachman, 1991).
Gustov et al (1985) investigated the non-invasive measurement of cranio-cerebral temperatures
in healthy adults by the use of microwave radiometry. The system they used
was reported to measure temperature to an accuracy of 0.1 ◦C. In three patients, brain
temperature at a depth of 5–6 cm was measured directly during surgery. The radiometric
temperature measured at the corresponding site prior to surgery was found to differ from
brain temperature at depth by only 0.4–0.8 ◦C. Recently, an interest in the application
of microwave radiometry to monitoring neonatal brain temperature during hypothermic
treatment has arisen (Hand et al, 2001).
Foster and Cheever (1992) discuss the relative advantages and disadvantages of microwave
thermometry. In their opinion, the advantages include the comparative simplicity
of the hardware, e.g. in comparison to magnetic resonance systems (see Section 4.2.2),
and the potential for greater tissue-depth temperature determination compared to infrared
thermomgraphic techniques. Another obvious advantage is the inherent safety of
the technique, since no external field is applied to the tissue. Amongst the disadvantages,
Foster and Cheever (1992) include the relatively poor temporal and spatial resolution of
the measurements, the susceptibility to interference from other electrical signals and the
possibility that the radiometric signal may reflect temperature-dependent changes in tissue
dielectric properties as well as tissue temperature changes.