A new automatic brain temperature controller was developed based on the conditions required for clinical use from the viewpoint of various aspects of feasibility, ...
Electronics and Communications in Japan, Vol. 97, No. 4, 2014 Translated from Denki Gakkai Ronbunshi, Vol. 132-C, No. 4, April 2012, pp. 615–622
Development of Automatic Brain Temperature Controller Based on Conditions of Clinical Use
TOMOHIKO UTSUKI and HIDETOSHI WAKAMATSU Tokyo Medical and Dental University, Japan
is suppressed by brain cooling to about 32 ◦ C to 35 ◦ C for several days [1]. At present, in brain injury following resuscitation after cardiopulmonary failure or in neonatal asphyxia, improvements in the survival rate and suppression of later impediments have been confirmed [2–5]. Various clinical trails are underway for other forms of brain injury, including cerebral hemorrhage and head trauma. However, in particular when the body temperature falls below 32 ◦ C, side effects such as arrhythmias, heart failure, and infection readily occur, and thus precise control of brain temperature and body temperature is essential [1]. Methods for brain cooling include surface cooling, blood cooling, and intravascular cooling. At present, surface cooling, in which the patient is wrapped in a cooling blanket is the most widespread. This method is noninvasive and the equipment is easy to handle. However, the temperature of the water in the blanket must be regulated in order to control brain temperature. However, determining the appropriate temperature is difficult due to significant differences among individuals and changes over time in patients. Thus precise brain temperature control is difficult in manual methods based on experience [1]. Furthermore, substantial efforts by health care workers and significant medical costs are involved, and as a consequence, the use of devices that automatically control brain temperature is eagerly awaited. The present investigators and their colleagues have developed an algorithm and prototype device for the automatic control of brain temperature using a surface watercooling method [6, 7], and good control precision was found in clinical testing [8]. However, the prototype device does not adequately meet the conditions for clinical use in terms of shape, size, and power consumption. Other surface water-cooling devices are not designed with the idea of automatic control of brain temperature and have insufficient precision for such control, and devices for clinical use developed for high-precision brain temperature control do not yet exist either inside or outside Japan. Therefore, in the
SUMMARY A new automatic brain temperature controller was developed based on the conditions required for clinical use from the viewpoint of various aspects of feasibility, in particular an electric power consumption of less than 1500 W in an intensive care unit. An adaptive algorithm was employed to deal with individual time-varying characteristic changes of patients. The controller in water-surface cooling hypothermia requires significant power for frequent regulation of the water temperature of cooling blankets. Thus, in this study, the power consumption of the controller was checked by several kinds of tests involving simulated control of brain temperature with a mannequin that had thermal characteristics similar to those of adult patients. The required accuracy for therapeutic brain hypothermia, specifically a control deviation within ±0.1 ◦ C, was experimentally confirmed in terms of the root mean square control error, even though the present controller consumes less energy than a conventional controller. It can maintain a reserve power margin of more than 300 W even during full operation. The clinically required water temperature was also confirmed within the limits of the power supply. Thus, its practical application is highly desirable, since it will reduce the physical burden on medical staff and will offer greater usability and better medical cost performance. C⃝ 2014 Wiley Periodicals, Inc. Electron Comm Jpn, 97(4): 19–29, 2014; Published online in Wiley Online Library (wileyonlinelibrary.com). DOI 10.1002/ecj.11483 Key words: brain temperature; automatic controller; adaptive control; power saving; brain hypothermia; clinical use. 1.
Introduction
Therapeutic brain hypothermia is a focused therapy for serious brain injury in which secondary brain cell death
C⃝
19
2014 Wiley Periodicals, Inc.
present research we developed an automatic brain temperature controller with the conditions for clinical use taken into consideration, and performed an evaluation with a focus on conserving power, the most important issue at present. 2. 2.1
Necessary Conditions for Clinical Use Necessary conditions
Brain hypothermia is performed under general anesthesia in order to prevent shivering due to low body temperature. As a consequence, the vital signs must be monitored constantly, and systemic control through artificial respiration, blood transfusion, and medication must be performed continuously. Therefore, it is ordinarily performed in an intensive care unit complete with patient monitors and specialized nurses. An intensive care unit is a place for the treatment of patients in life-threatening states. Therefore, the types of equipment used must be as quiet as possible and must not cause pollution. Furthermore, because the space available to each patient is limited, small devices are preferable, and fail-safe functions that give warnings and automatically shut down when operational problems or failures occur are also needed. In equipment that is to be used over the long term, durability with respect to continuous operation and reliability with respect to faults are also required. The power available for use from an ordinary or emergency power supply is usually no more than 1500 VA (single phase, 100 V). Thus the maximum power consumption must be less than 1500 W. From the point of view of the medical staff involved, equipment with good operability and display characteristics and easy maintenance is desirable. The conditions described have been established for all medical equipment used clinically. In the present research we also developed a brain temperature controller with these conditions given due consideration. 2.2
Fig. 1. Right-side view of automatic brain temperature controller using the surface water-cooling method. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.] the temperature rapidly and precisely, the mixing flow rate of the warm water and cold water is adjusted by a flow PI control unit (Buerkert, Ty8071, Ty6022, Ty8623-2) in which the aperture of the proportional solenoid valve is controlled by target flow level signals. The warm water used in mixing is produced in the warm water tank and the cold water is produced in the cold water tank. A heater is installed for the warm water tank, and cooling coils from a compressor are installed for the cold water tank. The heater is turned off when the
Equipment configuration
Figure 1 shows photographs of the automatic brain temperature controller based on the surface water-cooling method developed by us. Figure 2 shows a schematic diagram of the mechanism. The equipment is 60 cm wide, 160 cm deep, and 170 cm high. Water is sent to the blanket from a mixing tank by a pump. Water can be delivered to up to four blankets at the same time, and the used water is then returned to the mixing tank by the internal pressure of the blanket. In order to control brain temperature, the mixing tank temperature is adjusted to the appropriate control temperature by adding warm water at approximately 40 ◦ C or cold water at approximately 5 ◦ C to the tank. In order to adjust
Fig. 2. Schematic description of the automatic brain temperature control system using the surface water-cooling process.
20
temperature of the warm water tank exceeds 40 ◦ C, and the compressor halts when the temperature of the cold water tank falls below 5 ◦ C. When the warm water and the cold water are mixed, the amount of water in the mixing tank rises, and any excess flows to a buffer tank. The buffer tank returns water to the warm water tank and the cold water tank by operating a pump when the water level in either tank drops. A tank to store supplemental water is also provided for cases of water deficiency due to a change in the type or number of blankets, or evaporation. This supplemental water tank also functions as an injection port before the start of control. In order to assess the operation of the device, temperature sensors, flow rate sensors, water level sensors, and the leak detection sensor described later were installed in various places inside it. A touch panel display for operation, display, and various warnings was also provided. The instructions and processing for the mechanical operation of the device are performed by a programmable logic controller (PLC). The computations in the control algorithm are performed by a small personal computer (PC). As a result, the PC and the PLC send and receive measurement data and calculation results via serial communications. The PC also handles the setting of target brain temperature values and storage of control data files.
In earlier development of the prototype device, the capacity to produce the necessary warm water and cold water was determined based on evaluations of the capacity required for automatic control of brain temperature, and a heater and compressor were selected. In order to reduce the amount of warm and cold water used in mixing, the control sampling period was lengthened from 12 s to 30 s. Next, the heater and compressor were selected as described after determining that the temperature of the warm water tank and the cold water tank could be maintained even if the capacity to create warm water and cold water were reduced by approximately one-half. Extending the sampling period degrades the immediacy of operational control, and there are concerns that an overall deterioration in control precision may occur. However, because the time constant for the brain temperature response to the body surface temperature is slow, approximately 3 h in an adult, we determined that there would not be any significant effects even if a sudden disturbance such as a transfusion or other interference from outside the body surface occurred. To further conserve power, two buffer tanks, one to return water to the warm water tank and one to return water to the cold water tank, were set up. Thus when the temperature of the mixing tank is high, overflow is sent to the warm water buffer tank, and when the temperature is low, overflow is sent to the cold water buffer tank. Drops in the temperature of the warm water tank and rises in the temperature of the cold water tank are limited and the power consumption for the production of warm water and cold water can be reduced. Stainless steel, vinyl chloride, and polyolefin were used as the piping materials. Thus, sludge does not readily accumulate and a variety of cleaning agents and disinfectants can be used. In addition, if one of the hand-operated valves set up is opened, water can be forcibly removed from the hot water, cold water, or mixing tank via the buffer tank, so that replacement with clean water can be readily performed. In order to suppress bacterial growth, an ultraviolet germicidal light is installed in the mixing tank water circulation system. Because countermeasures for basic grounding and protective grounding were performed, the electrocution protection class of the device is class 1.
2.3 Functions and features based on clinical conditions In a clinical setting, the possibility of damage to or deterioration of the blanket as a result of handling or inspection, and of damage to tubes and connectors due to overloading cannot be ruled out. Therefore, we considered a mechanism to detect water leaks around the device based on flow level differences between the water sent and the water received by each blanket. We programmed our device so that if a water leak is detected, the water in the blanket is drawn off and an alarm is issued, together with automatic shutdown as a fail-safe function. Water leak detection sensors are installed in various places in the device. If these are activated, an alarm is issued and operation is stopped automatically, as in the case of a water leak outside the device. In brain temperature control using the surface watercooling method, warm water and cold water must be produced and stored at all times for operational control. Because a margin of approximately 500 W is required for the operation of the pump, electromagnetic valve, processor, and other components excluding heating and cooling, in our device two heaters with capacities of 250 W and 750 W are installed in the warm water tank, and only the 250 W heater is used when operating the compressor (600 W). Thus the power consumption limit of 1500 W in a clinical environment can be met.
3.
Operational Performance Tests
3.1 Test of the capacity to create warm water and cold water (1) Methods The water levels in the warm water tank and the cold water tank were set to 17.5 and 17.0l, and the process of producing warm water and cold water starting from room temperature was monitored.
21
Fig. 3. Production of warm and cold water before starting automatic control of brain temperature.
(2) Results Figure 3 shows the results. The temperatures of the warm water and cold water at the start were 25.8 ◦ C and 25.9 ◦ C, and room temperature was 25.0◦ C. The cold water was cooled by 21.0◦ C, reaching 4.9◦ C after 24 min, and the warm water was heated by 4.5 ◦ C, reaching 30.3 ◦ C. The compressor was off and the heater, running at a power consumption of 750 W, was on. During the following 17 min, the water temperature rose by 10.0 ◦ C, reaching 40.3 ◦ C. Thus we calculated that the average capacity for producing cold water required 1103 W of power, and the average capacity to produce warm water required 236 W with the compressor on and 741 W with the compressor off. 3.2
Fig. 4. Warming up and cooling down of the water for blankets by constant mixing of warm and cold water in the reservoir. The square, circle, and cross marks represent the temperatures of the warm and cold water and the blanket, respectively. the mixing tank temperature rose from 9.6 ◦ C to 29.2 ◦ C in 2 min, and the warm water tank temperature fell 4.5 ◦ C in 5 min 30 s. In warm water mixing at a flow rate of 10.00 ml/s, the mixing tank temperature rose from 9.0 ◦ C to 30.2 ◦ C in 2 min 30 s, and the warm water tank temperature fell to 4.8 ◦ C in 6 min. The temperatures of the warm water tank and the mixing tank approached 40 ◦ C, the set temperature for the warm water, in both cases. In cold water mixing at a flow rate of 27.78 ml/s, the mixing tank temperature fell from 38.4 ◦ C to 20.6 ◦ C in 3 min and the cold water tank temperature rose from 6.8 ◦ C to 13.1 ◦ C in 4 min 30 s. In cold water mixing at a flow rate of 10.00 ml/s, the mixing water tank temperature fell from 38.6 ◦ C to 20.0 ◦ C in 3 min, and the cold water tank temperature rose from 6.9 ◦ C to 13.3 ◦ C in 8 min. The temperatures of the cold water tank and the mixing tank approached 5 ◦ C, the set temperature for the cold water, in both cases. We calculated the heating and cooling capacity Wm of the mixing tank by Eq. (1). Figure 5 shows the time evolution of Wm . Cw and Pw are the specific heat and density of the water, Vm and ΔTm are the backward differences of the water level and the temperature in the mixing tank, and Δt is the 30-s measurement interval. ΔTm . (1) Wm = Cw ⋅ ρw ⋅ Vm ⋅ Δt
Test of capacity to warm or cool the mixing tank
(1) Methods Warm water at 40 ◦ C and cold water at 4 ◦ C were mixed at flow rates of 27.78 ml/s and 10.0 ml/s in the mixing tank at a water level of 2 l and at a room temperature of 24◦ C to 25◦ C, and the temperature changes in each tank were measured. The mixing flow rate of 27.78 ml/s represents the upper limit that can be set in the flow rate PI control unit. The device was set up so that the warm water tank and the cold water tank had 18.0 l of water added, and when 1.0 l had been used, that portion was returned to each tank from the buffer tank. (2) Results Figure 4 shows the temperature evolution of the warm water tank, cold water tank, and the mixing tank when mixing warm water and cold water at flow rates of 27.78 ml/s and 10.0 ml/s. In each graph, a square represents the warm water tank temperature; a cross, the cold water tank temperature; and a black circle, the mixing tank temperature. In warm water mixing at a flow rate of 27.78 ml/s,
22
Fig. 6. A block diagram of the blanket water temperature control algorithm, consisting of two blocks: the PI control system for warming and cooling efficiency, and the calculation system for mixing water flow.
ing tank; Th and Tc are the temperatures of the warm water tank and the cold water tank; and Fb is the the mixing flow rate of the warm water or the cold water. The proportional gain in PI control was set to 200, the integration time to 10, and the sampling period to 1 s. However, because the minimum flow rate that can be controlled in the flow rate control unit is 0.55 ml/s, the flow rate was set to 0 using the on/off electromagnetic valve when calculating a mixing flow rate below this level.
Fig. 5. Efficiency of warming and cooling of water in blankets by constant mixing of warm and cold water.
The maximum production capacity for warm water is 1000 W, equivalent to the maximum power consumption of the warm water tank heater. For room temperature of about 25◦ C, the maximum cold water production capacity can be estimated to be about 1000 W, equivalent to slightly less than twice the compressor power consumption. Therefore, based on the principles of the device, if 1000 W is exceeded, the initial set temperature cannot be maintained in the warm water tank or the cold water tank. However, at any mixing level, a 1000 W is not exceeded up to 1 min after the start. This appears to be because of the transient increase in the temperature of the cold water tank during cold water mixing and the transient decrease in the temperature of the warm water tank at the mixing temperature shown in Fig. 4. Wm drops sharply thereafter and is estimated to be 35 W to 50 W.
(2) Results Figure 7 shows the control dynamics. When the target was 30 ◦ C, the mixing tank temperature, which had been at 19.7 ◦ C, reached 30.0 ◦ C after 63 s and converged to within ±0.3 ◦ C after 77 s. When the target was 20 ◦ C, the temperature dropped from 29.5 ◦ C to 20.0 ◦ C in 71 s, and converted to within ±0.5 ◦ C after 130 s and to within ±0.3 ◦ C after 221 s.
3.3 Test of mixing tank temperature control (1) Methods The maximum flow rate for warm water and cold water was set to 10.00 ml/s, and experimental control was performed with the mixing tank at 20 ◦ C or 30 ◦ C. The control algorithm used is shown in the block diagram in Fig. 6. This algorithm is composed of a PI control feedback loop representing the control system for the heating and cooling capacity Wm in the mixing tank, and a block that calculates the mixing flow rate for warm water and cold water. Tms in the figure is the target temperature in the mix-
Fig. 7. Dynamics of water temperature in the blanket during PI control from 20 ◦ C to 30 ◦ C and from 30 ◦ C to 20 ◦ C.
23
3.4
simple, and control without physiological problems can be realized. T in the figure is the temperature and R is the reference value. The tilde indicates the change in a variable from its initial value. The superscripts “model” and “solut” denote the values of the variables in the model and the theoretical solutions, and the subscripts “water” and “brain” denote the water temperature and brain temperature.
Test to simulate brain temperature control
(1) Control algorithm Water-surface cooling has the advantage of being noninvasive, in contrast to blood cooling or intravascular cooling. However, the time constant for the brain cooling response with respect to the body surface temperature is about 24 h in an adult, so that brain temperature control is difficult [10]. Further, blood flow to the skin drops with decreasing body surface temperature, and thus control becomes more difficult due to a decrease in gain and an increase in the time constant [11]. In brain hypothermia performed under general anesthesia, fluid transfusion, blood transfusion, and drug administration are performed constantly under artificial respiration. During such medical measures, control may be disrupted not only by a simple disturbance but also by changes in the patient’s characteristics. Changes in the patient’s status that occur along with therapy may also make control more difficult. In addition, not only are there substantial individual differences in patient characteristics, but the true order of the transfer function is also unclear. Given these conditions, precise and flexible brain temperature control using a control method with fixed control rules is difficult. Therefore, considering the system in Ref. [6] with two degree of freedom, combining optimal control and adaptive control as shown in Fig. 8, we use an algorithm based on this system in our device. Figure 8 shows a block diagram of the algorithm. The algorithm consists of a control loop for modeling the characteristics and a control loop for the patient. The characteristic model approximates the average brain temperature response characteristics with respect to the blanket temperature using a first-order delay, and functions as a reference model. Thus setting the control parameters is
(2) Methods For ethical reasons, performance testing of a device under development on actual patients is not allowed. Thus we used a mannequin [7] that represented the average thermal characteristics of an adult male, and performed simulation experiments to control the temperature of its head. This mannequin was equipped with tubes and pads in its body to simulate blood circulation, and a mechanism to run water through it using a pump. Metabolic heat could also be simulated by way of heaters located at various sites. Because the mannequin is made of a water-containing gel, its specific heat, density, and heat transfer rate are similar to those of physiologic tissue, and the response of the head temperature to the surface temperature is close to that of a living body. In the test, we turned on the pump to simulate circulation and the heater to simulate metabolic heat, and then left the mannequin at room temperature for 12 h, until an equilibrium state was reached. Then, using a sampling period of 60 s for brain temperature control and a sampling period of 1 s for mixing tank temperature control, we performed experiments to simulate brain temperature control in accordance with the following target schedule; this schedule reduces only the period during which control is sustained, which is several days in reality. (I) Cooling phase: cooling the mannequin head to 33 ◦ C in 7.5 h (II) Sustained phase: maintaining the mannequin head at 33 ◦ C for 8.5 h (III) Warming phase: warming the mannequin head to 35 ◦ C in 8 h (IV) Monitoring phase: keeping the mannequin head at 35 ◦ C for 4 h In particular, the following actions were performed in sequence in order to confirm the control performance with respect to a disturbance or change in heat characteristics in the sustained phase. (A) The circulating pump in the mannequin was turned off for 30 min. (B) The output of the heater in the mannequin head was increased by 20% for 5 min. (C) The room temperature was adjusted to 28 ◦ C to ◦ 30 C using an air conditioner.
Fig. 8. Block diagram of the brain temperature control algorithm, consisting of two feedback loops: for optimal control of the characteristic model, and for adaptive control of an actual patient.
24
(D) The blanket on the mannequin was removed for 30 min. (E) The heater output outside the mannequin head was increased by 20%. (A) represents cardiac arrest; (B), brain hypermetabolism; (C), a rise in the intensive care unit room temperature; (D), removal of the blanket, for example, during cleaning; and (E), hypermetabolism in the body. Because the mannequin pump is located outside the mannequin, the water flowing in place of blood is directly affected by the room temperature. Therefore, (C) includes factors representing not only a disturbance from the body surface that occurs in a clinical situation but also a disturbance via the blood. Moreover, (B), (D), and (E) correspond to situations that may occur clinically. Because cardiac arrest that escapes notice for 30 min or a rise in the intensive care unit room temperature to 30 ◦ C are ordinarily impossible, (A) and (C) are extreme actions. Measurement of the head temperature was performed by inserting a platinum resistor, and the signal was input to the PC associated with our device via a signal converter and an AD converter.
Fig. 9. Control of head temperature of mannequin. In the upper illustration, the white line represents the head temperature and the black line represents the target temperature. In the middle, the white line represents the water temperature in the blanket and the black line represents the solution calculated by the control algorithm. The bottom illustration shows room temperature.
(3) Results Figure 9 shows the results of the control process. The black line and the white line in the upper graph represent the set target temperature and the measured temperature of the mannequin’s head. The black line and the white line in the middle graph represent the calculated solution for the temperature of the mixing tank and the actual temperature in the mixing tank. The black line in the lower graph represents the evolution of the room temperature. In the cooling phase, a control error of –0.45 ◦ C occurred 30 min after the start of control; it then converged to ±0.15 ◦ C after 90 min. In the sustained phase, the temperature of the mannequin’s head rose to 28 ◦ C to 30 ◦ C 10 min after the start of an adjustment to room temperature, and after an error not exceeding 0.69 ◦ C, it converged to within ±0.15 ◦ C after 11.5 h. During this time, the temperature of the mixing tank did not track the calculated solution, and was estimated to be approximately 8.0 ◦ C. The control errors in the warming phase and the monitoring phase were ±0.15 ◦ C and ±0.10 ◦ C, respectively. A solution in the range of –33.3 ◦ C to 36.5 ◦ C was estimated for the temperature of the mixing tank, but in reality the range was estimated to be 8.0 ◦ C to 36.5 ◦ C, and the solution could not be tracked when the temperature was below 8.0 ◦ C. Except for this time range during which tracking could not be performed, the calculated rms error of the temperature of the mixing tank was 0.23 ◦ C. The room temperature was estimated to be in the range of 22 ◦ C to 27 ◦ C except for the time period when performing action (C).
Table 1 summarizes the average power consumed by producing warm water and cold water, and the duty ratios in the heater and compressor. The average power consumed in producing warm water and cold water was highest during the cooling phase. In all phases it was of the order of 600 W. The time to produce cold water was approximately 90% of the control period in each phase; the time to produce warm water at 250 W was 15% to 36%. The time to produce warm water, in which a total of 1000 W was consumed considering the 750 W in the heater, was 0.2% to 3.8%.
Table 1. Consumed power and running time of heater and compressor for production of warm and cold water Ratio of Ratio of Ratio of Average of heating time heating time cooling time power at power at power at power consumption for heating consumption consumption consumption and cooling of 250 W (h) of 1000 W of 600 W (h) (h) (W) Cooling Hypothermia Rewarming Normothermia Total
25
676.4 616.2 604.6 648.6 634.1
0.363 0.173 0.157 0.207 0.226
0.027 0.038 0.021 0.015 0.027
0.931 0.890 0.906 0.977 0.918
4. 4.1
warm water and cold water are faster at a mixing flow rate of 27.77 ml/s than at 10.00 ml/s. However, this difference is small compared to the flow rate difference. This is because as the mixing flow rate increases, the flow rate of water returned to the warm water and cold water tanks also rises, and the warm water and cold water tanks can no longer be maintained at their set temperatures, resulting in a large difference in the warming and cooling capacity of the mixing tank, as shown in Fig. 5. This result confirmed that there was no difference in performance when the peak mixing flow rate was reduced to 10.00 ml/s. A flow rate of 10.00 ml/s is approximately 36% of 27.77 ml/s; this ratio was determined taking into consideration the ratio of 1500 W, the maximum power that can be consumed in a clinical environment, to 4 kW, the maximum power consumed in a conventional device.
Discussion
Necessity and methods for conserving power
The brain temperature control algorithm of our device may raise and lower the mixing tank temperature repeatedly to achieve operation appropriate to changes over time and individual differences in patient characteristics. Large quantities of warm and cold water are then used, so that a significant reserve margin is needed for the power consumption in producing warm water and cold water. In earlier devices, a maximum of 4 kW could be consumed when producing warm water and cold water, requiring special power supply construction in an intensive care unit [7]. Thus, unless the overall power consumption of the device is reduced to below 1500 W, clinical use cannot be justified. Methods for conserving power include reducing the peak power consumption of the heater and compressor, smoothing the temperature changes by increasing the water levels in the warm and cold water tanks, conserving the warm water and cold water mixing level by increasing the sampling period for brain temperature control and reducing the peak mixing flow rate, and adjusting the control parameters accordingly.
4.2
(3) Mixing tank temperature control performance If the capacity to produce warm water and cold water is reduced, then the temperature in the warm water tank drops and that in the cold water tank rises due to water returned from the buffer tank, and the capacity to warm and cool the mixing tank is reduced. Consequently mixing tank control and brain temperature control performance deteriorate. The warming and cooling rates in the mixing tank immediately after the start of the test were 0.16 ◦ C/s and 0.13 ◦ C/s, as can be seen in Fig. 7. This rate is sufficiently high for practical use if the sampling period for brain temperature control is 60 s. The error after convergence is estimated to be the same as or less than that of a conventional device. Thus the validity of the proportional gain, integral time, and mixing tank temperature control sampling period that we used was also confirmed.
Effect of conserving power on operational performance
(1) Capacity to produce warm water and cold water Our device reduced the peak power consumption of the heater and compressor to less than half that of a conventional device and increased the stored water level in the warm water tank and the cold water tank by a factor of 1.5. This increases the time for the warm water and the cold water to reach the set temperature. It is confirmed that this time was approximately 40 min, based on Fig. 3. Approximately 30 min to 1 h is required from contact of the transport of a patient to the start of brain hypothermia, which is a clinically acceptable time period. In surface cooling, a method in which cooling is started within about 10 min by using rapid transfusion of electrolytes at approximately 4 ◦ C has already been considered [12]. If this method is combined with our method, then the time taken to produce warm water and cold water in our device can be increased.
(4) Brain temperature control performance In a conventional device, the brain temperature control sampling period is 12 s. In our device it is 60 s in order to conserve power. If the sampling period is lengthened in this way, then not only does the response to the harmonic component of a disturbance drop, but adaptability to patient characteristics is also reduced. The large control error immediately after the start of cooling observed in the simulation test appears to have been due to the unavoidable occurrence of adaptive operation when control starts. The later convergence occurs because this phenomenon is transient, showing that control stability can be achieved. The control error that occurred after action (C) was caused by the fact that calculations were performed using a mixing tank temperature that cannot be realized in our device. The results showed that if this kind of
(2) Capacity for warming and cooling the mixing tank As can be seen in Fig. 4, the changes in the mixing tank temperature immediately after the start of mixing of
26
disturbance alone is introduced, our device might encounter control difficulties. The rms control error excluding the period immediately after the start of cooling and after action (C) was 0.12 ◦ C. Compared to a simulation test of a conventional device [7], the period in which the mixing tank temperature changes continues to be 20 min to 30 min. The amplitude is reduced from about 10 ◦ C to about 2 ◦ C. The reason is that in the past, the temperature was averaged over 1 min and used for control calculations, while in the present test the average was taken over 5 min, so that the effects of measurement error and noise were reduced. The decrease in amplitude of the temperature changes in the mixing tank represents a reduction in the amount of warm water and cold water used for mixing. Therefore, the increase in the control sampling period had an effect on energy conservation. However, the rms control error increased above the value of 0.042 ◦ C [7] for a conventional device. In the case of an actual patient the control error would increase further, as indicated by experiments in clinical application testing. In a clinical setting, the brain temperature management range should be about ±0.1 ◦ C [1]. This level of precision can be achieved by small adjustments in the sampling period. This is because the power consumed in producing warm water and cold water is of the order of 600 W on average, and when the approximately 500 W required to operate the pump, proportional control valve, electromagnetic valve, and PLC and PC is subtracted from 1500 W, there is a margin of over 300 W. If the margin is used effectively, small adjustments in the sampling period are sufficient, and our device has the capacity to satisfy the clinical requirements even for precise brain temperature control.
In our device, consideration was given to maintaining the safety of the patient and health care workers by means of a fail-safe mechanism that stops after automatic absorption of water while issuing an alarm when a leak is detected, the selection of piping materials, and the installation of an antimicrobial light in the mixing tank circulation path. Various settings and displays can be provided on the touch panel screen. Thus the operability and display performance are improved compared to a conventional device. In addition, methods of measuring brain temperature clinically now include methods of measuring the tympanic membrane temperature, the intranasal temperature, the esophageal temperature, bladder temperature, the rectal temperature, and blood temperature, as replacements for the method in which a measurement catheter is inserted into the brain. Our device is equipped with an external terminal for acquiring such measurement data. Thus brain temperature control feedback can be obtained by simply connecting signal cables from various temperature measurement devices. In particular, the tympanic membrane temperature measurement based on a tympanic membrane contact temperature measurement probe [13] previously developed by the present authors and colleagues is relevant here. Thus the developed device takes into consideration the various conditions for clinical use mentioned in Section 2.1. However, whether health care workers can use it smoothly and without stress is a topic for evaluation in future clinical testing.
5.
Conclusions
We developed an automatic brain temperature control device using the surface water-cooling method suited to the conditions of clinical use. In particular, as regards the reduction power consumption, an important issue for clinical use, various operating performance tests verified the viability of our device. Thus our device is well suited as a prototype for clinical use. However, to create an actual product, further size reduction, complete heat insulation of the interior of the device, and increased efficiency of power usage are required.
4.3 Clinical use of our device In our device, the control unit proper, the unit for producing warm water, and the unit to create cold water are combined into one, making it easier to use than a conventional device [7]. Furthermore, it fits into the less than 60 cm clinically allowed for width. Therefore, it is suitable for clinical use in terms of shape and size. As regards power consumption, the required clinical precision of brain temperature control can be achieved at less than 1500 W, as described. In the tests described here, the compressor usage time was kept to about 90%, apparently because the uninsulated buffer tank was warmed by the room temperature on that day and by waste heat from the pump. Therefore, further reduction of power consumption by fully insulating the interior of the device remains an issue. In addition, ideas for using a heater and compressor with a higher capacity than in the present device by installing a secondary battery to store extra power is also a topic for the future.
Acknowledgments This research was performed with support from a Science and Research Grant-in-Aid for Basic Research (A) (topic no.: 19200042) from the Ministry of Education, Science, Sports, and Culture from 2007 through 2009. The fabrication of our device was substantially supported by Mr Toshio Ohashi, chairman and director of the board of Industry Network Co., Ltd., and by Mr Hiroshi Ando, president of CIS Corporation. We take this opportunity to thank both.
27
REFERENCES 7.
1. Asai Y ed. Special edition: Current and future outlook for brain hypothermia. ICU CCU 2003;27(8):723– 770. 2. The Hypothermia after Cardiac Arrest Study Group. Mild therapeutic hypothermia to improve the neurologic outcome after cardiac arrest. N Eng J Med 2002;346(8):549–556. 3. Bernard SA, Gray TW, Buist MD, Jones BM, Silvester W, Gutteridge G, Smith K. Treatment of comatose survivors of out-of-hospital cardiac arrest with induced hypothermia. N Eng J Med 2002;346(8):557–563. 4. Shankaran S, Laptook AR, Ehrenkranz RA, Tyson JE, MacDonald SA, Donovan EF, Fanaroff AA, Poole WK, Wright LL, Higgins RD, Finer NN, Carlo WA, Duara S, Oh W, Cotton M, Stevenson DK, Stoll BJ, Lemons JA, Guillet R, Jobe AH. Whole-body hypothermia for neonates with hypoxicischemic encephalopathy. N Eng J Med 2005;353(15): 1574–1584. 5. Gluckman PD, Wyatt JS, Azzopardi D, Ballard R, Edward D, Ferriero DM, Polin RA, Robertson CM, Thresen M, Whitelaw A, Gunn AJ. Selective head cooling with mild systemic hypothermia after neonatal encephalopathy: Multicare randomized trial. Lancet 2005;365:663–670. 6. Utsuki T, Wakatsuki T, Wakamatsu H. Model reference control of brain temperature with 2 degrees of
8.
9.
10. 11.
12.
13.
freedom for brain hypothermia treatment. Trans Soc Instr Contr Eng 2006;42(6)683–690. (in Japanese) Utsuki T, Wakatsuki T, Wakamatsu H. Development of automatic controller of water temperature by surface water-cooling for brain hypothermia treatment. Trans Jpn Soc Med Biol Eng 2007;45(1):1–10. (in Japanese) Wakamatsu H, Utsuki T, Mitaka C, Ohno K. Clinical system engineering of long-term automatic thermal control during brain hypothermia under changing conditions. Tech Health Care 2010;18(3):181–201. Wakamatsu H, Gaohua L, Utsuki T. Requirements for the automatic control system of brain temperature for adult patients. Brain Death Resuscit 2005;17(1):139– 145. (in Japanese) Hayashi N. Brain Hypothermia, Springer, Tokyo; pp. 121–151, 2000. Khan F, Spence VA, Belch JJ. Cutaneous vascular responses and thermoregulation in relation to age. Clin Sci 1992;82(5)521–528. Kimb F, Olsufka M, Longstretch WT Jr., Maynard C, Carlbom D, Deem S, Kudenchuk P, Copass MK, Cobb LA. Pilot randomized clinical trial of prehospital induction of mild hypothermia in out-of-hospital cardiac arrest patients with rapid infusion of 4◦ C normal saline. Circulation 2007;115:3064–3070. Open Patent Application: A Device to Detect Eardrum Temperature; Inventors: Hideo Wakamatsu, and Tomohiko Utsuki; submitted by: Tokyo Medical and Dental University; Application No.: 2011-147506.
AUTHOR
Tomohiko Utsuki (nonmember) completed the program in medicine at Tsukuba University in 1996, receiving a bachelor’s degree. He completed the latter half of the doctoral program at the Graduate School of Health Care Sciences at Tokyo Medical and Dental University in 2006. After becoming an adjunct instructor and specially appointed assistant professor, he became a specially appointed lecturer in research at the same university in 2009. He is now engaged in research on automatic control of biophysiological states. He is a member of SICE, JSMBE, the Japan Association of Cerebral Resuscitation and Brain Death, and the Japanese Association of Brain Hypothermia. Dr. of public health.
28
AUTHOR (continued)
Hidetoshi Wakamatsu (nonmember), completed the M.E. program at the Graduate School of Yokohama National University in 1972, then became a lecturer at Tokyo Medical and Dental University. After becoming an assistant professor at Ashikaga Institute of Technology and subsequently a professor in the Faculty of Engineering of Fukui University, he became a professor at the Graduate School of Health Care Sciences of Tokyo Medical and Dental University. He was a Deutsche Akademische Austauschdienst exchange student in Germany between 1973 and 1975. He was a visiting researcher at Erlangen-N¨urnberg University School of Medicine, as well as a visiting professor at Oregon State University in 1993, at Pusan National University in 1995, at Huazhong University of Science and Technology in 1996, and at Tianjin University. D.Eng. (University of Tokyo).
29
