SENSORS
Implications of RFICs for Medical Instrumentation
Frederick G. Weiss
Although much of the focus regarding the use of radio-frequency integrated circuits (RFICs) lies in the realm of audio, video, and general data transmission, a variety of medical uses for RFIC-based sensor-transceivers are also viable.
Most of the electronics in today's medical instrumentation are packaged components mounted on circuit boards, which in turn are installed in instrument cases to be placed next to or connected to a patient. Advances in sensor technologies are beginning to make it possible for entire instruments to be integrated with a wireless communication channel on a single sensor-transceiver, providing more flexibility and control for patients and caregivers.
The implications of tiny, stand-alone sensor-transceivers for patient care are significant. For example, an implanted sensor can retrieve data from inside the body without wires or tubes penetrating the dermal (skin) barrier. This reduces the possibility of infection, which is frequently a greater contributor to morbidity and mortality than the original problem. At the other end of the spectrum, radio-frequency identification (RFID) bracelets can help prevent the abduction or switching of infants in hospital nurseries or track equipment to ensure availability.
Figure 1. Frequency-hopped ISM band.1
This article provides an overview of how stand-alone sensor-transceivers such as these can help practitioners deliver better care. The ideas presented grew out of discussions with practicing physicians in a variety of disciplines, as well as several manufacturers of cardiac monitoring equipment. Participants were asked to speculate on the impact an integrated sensor-transceiver might have on their specialty, particularly how these devices might solve monitoring control problems. Note that the scope of this article is necessarily limited and should not be taken as a definitive list of all possible medical applications of RFIC-based sensor-transceivers.
Medical Monitoring Considerations
Environment. Many medical situations use monitoring activities that may be characterized by their relative intensity and the amount of data being taken. In an emergency room, for example, a patient who has suffered severe trauma must be rapidly evaluated, stabilized, and treated. In this case, the patient is generally inert or immobile, so attaching leads or tubes for monitoring purposes does not usually impose undo restraint on a patient's mobility. The same is true for an operating room.
Once a postoperative patient is recovering, however, monitoring requirements are expected to decrease. It is usually necessary for patients to get out of bed and move around in order to regain strength. During recovery, it is often necessary to continue to monitor vital statistics, such as heart rate or blood pressure. As a patient's health improves, the restraints imposed by wire tethers to monitors become increasingly burdensome. Some patients, such as those with cardiac or diabetic problems, will continue to require periodic monitoring after discharge. These parameters could be monitored either by periodic visits to the physician's office or by self-monitoring and reporting abnormal readings.
Figure 2. A typical ECG waveform.
Physiologic Parameters. Physiologic parameters for managing patients fall into three general categories: mechanical, chemical, and electrical. Mechanical characteristics include temperature, heart rate, blood pressure, cardiac output, respiration rate, and bone density, to name a few. Chemical parameters include blood gases, hematocrit, and blood glucose level. Electrical signals are generated by the heart and the brain. It is important to note, however, that not all of these are candidates for monitoring solutions that use integrated sensor-transceivers.
In terms of the actual measurement process, some parameters—such as those from the heart and the brain—originate as electrical signals and need only amplification, filtering, or other conditioning to allow them to be measured. For mechanical parameters, the conversion process is generally straightforward and involves using familiar strain gauge– or diaphragm-based pressure or displacement transducers, or ultrasound or tomographic imaging systems. Measuring chemical content and composition is the most difficult, requiring either the use of mass- or light-spectrographic techniques, or chemical reactions in combination with sensor electrodes that result in a change in voltage, current, or conductance.
Transducer technologies are addressed for specific monitoring applications in later sections.
Sensor-Transceiver Architectures. Depending on the data being transmitted and the environment in which the device is to be used, several different transceiver architectures are worthy of consideration. The range of possibilities extends from a simple FM transmitter to a complex, frequency-hopped ISM-band transceiver as shown in Figure 1.1 Typical applications require an external circulator (or diplexer) and filters, but these are not insurmountable obstacles to integration. For an implanted unit such as the seizure disorder monitor described below, however, the need to implant the sensor between the skull and the scalp requires that it have a very thin cross section. Incorporating extra-IC components is much more difficult and represents a potential limitation to the application of the technology.
Application Areas
Several examples of potential medical monitoring applications using integrated sensor-transceivers are described in the following sections. Each discussion presents a brief overview of the physiology and function of the organ or system to be monitored, the present measurement method, and possible uses of integrated sensor-transceivers that could either improve measurement accuracy or treatment outcome. Improved outcome could include a reduction in morbidity or mortality, enhanced patient safety, or increased patient mobility. Specific application areas include:
- Cardiac monitoring.
- Seizure disorder monitoring and treatment.
- Blood glucose monitoring.
- Infant and pediatric sleep apnea monitoring.
- Infant and pediatric thermometry.
- Continuous blood pressure monitoring.
- Patient, caregiver, and equipment identification, location, and verification.
Cardiac Monitoring. The heartbeat is stimulated electrically by an internal pacemaker, with the beat proceeding as a propagating electrical depolarization of the muscles surrounding the atrial and ventricular chambers.2 The electrical disturbance resulting from this charge redistribution can be sensed by electrodes placed on the limbs and chest of a patient. A typical electrocardiogram (ECG) waveform (shown in Figure 2) may have a peak-to-peak amplitude as high as 8–10 mV. This relatively small signal is normally found riding on as much as 400 mV of noise arising from the body's muscle activity as well as induced signals from nearby lighting mains, radio transmitters, microwave ovens, etc. The frequency components of an ECG waveform are such that digitizing the signal requires a sampling rate on the order of 200 Hz, with an amplitude resolution of 12 bits required to capture the fine structure of the waveform.
In general, multiple views or "leads" of the heart are analyzed for diagnostic purposes. Each view is based on a particular combination of electrodes: right arm to left arm, right arm to left leg, right arm to the average of the left arm plus left leg, etc. As a patient recovers, the need for stationary monitoring is replaced by the need for mobility. In the hospital, a patient may carry a pocket-sized portable ECG monitor that connects to a central station using a low-powered RF link.3
A Holter monitor, which uses only a few electrodes and is worn by a patient after discharge, enables patients to periodically download an ECG to a physician's monitoring service by holding the monitor unit up to a telephone handset and activating an internal modem.4,5 Patients, however, are encumbered with wires and a small electronics package, typically the size of a beeper. Once continuous monitoring is no longer required, a patient can be supplied with a credit card–sized unit that can be held to the chest. Pushing a button makes a short record of the heart's activity that can be downloaded via a telephone line.6
The leads for each of these ECG monitors are typically attached to a patient's limbs or chest using adhesive patches coated with a conductive gel. Each lead has an electrode with a tab or snap connection for the wire. An RF-based sensor-transmitter was proposed that uses a front-end electronic equivalent to a single-channel ECG, coupled with a low-power transmitter in each patch (see Figure 3). A normal ECG measures the net signal between two chosen electrodes. Because a standard patch normally has only one wire attached, it must be modified to contain two electrodes between which the signal would be measured. It appears that electrode spacing as small as 6 mm provides adequate signal-to-noise ratio.
For an instrumented patch, the electronics must include all necessary amplification and digitizing circuitry, a transmitter (including frequency source, modulator, and power amplifier), a small loop antenna, and a battery, all fastened to the back of the patch. Once attached to a patient and energized, the unit transmits real-time data to either a direct monitoring system (such as in the hospital) or to a patient's store-and-forward unit.
The addition of pattern-recognition capability to the receiving unit would enable it to automatically alert a monitoring service or emergency response team if a patient's monitor were to indicate a severe problem such as an arrest. Currently, patients must manually trigger an alarm by pressing a button on a transmitter attached to a necklace.
An integrated ECG monitor and transmitter could also provide continuous monitoring of implanted pacemakers. Although pacemakers currently keep short records of a heart's activity, patients must still report to the physician periodically or transmit activity from a portable ECG unit over the telephone.7 The normal method of communicating with an implanted pacemaker is to put it into so-called magnet mode, which provides a very robust—but very slow—communications channel and therefore a limited record. As a result, the transmission of a long ECG record using this approach would be unreasonable. By integrating a small, low-power transmitter into the pacemaker, larger data records could be downloaded on demand, making it possible to record and transmit more information.
Seizure Disorder Monitoring and Treatment. The electrical activity of the brain is monitored by arrays of up to 128 electrodes placed on the outer surface of the skull, and the activity is recorded as an electroencephalogram (EEG). Such monitoring enables a physician to localize the source of the disruptive electrical activity that causes seizures.
Figure 3. Instrumented ECG patch.
The array of EEG electrodes measures the superposed signals of millions of neurons that have been transmitted through a lossy medium of flesh and bone. As a result, the signals found at the surface of the skull are very tiny—on the order of 10–20 µV peak-to-peak for normal brain activity, roughly three orders of magnitude smaller than ECG signals. A normal EEG shows little or no evidence of strongly correlated events. A brain with a seizure disorder, however, has one or more sites that generate strong electrical signals that can interfere with other brain functions and that presents as strongly correlated events on the EEG.
An intriguing application has been hypothesized for an integrated sensor-transceiver for the management of seizure disorders.8 This idea is highly speculative, but given the devastating impact of a seizure disorder, it is a promising application to explore. The concept works as follows: once the seizure focus has been identified using standard EEG techniques, a small sensor-transceiver is implanted just beneath the scalp close to the focal center. Several wires (on the order of four) are positioned beneath the skin surrounding the focus, allowing activity to be monitored with greater sensitivity than is possible using wires that merely contact the skin surface. Signals would be transmitted to an external processor programmed to recognize the onset of a seizure. In this case, the processor would command the sensor-transceiver to stimulate the center to counteract the signals generated by the focal site.
The ability to counteract a seizure using this approach is hypothetical, and it is unclear at this point whether it could even be accomplished as envisioned. For example, it is not known whether the onset of a seizure could be identified early enough and with sufficient accuracy for such intervention to be effective, nor is it clear how one might counter the activity of the seizure site.
A simple implantable monitor could also evaluate a patient's condition. In particular, it is difficult to obtain accurate reports of seizure frequency and severity based on patient reporting. An in-dwelling sensor, on the other hand, offers an objective record of seizure activity.
Blood Glucose Monitoring. Monitoring body chemistry for diagnostic purposes is relatively easy in an operating room or critical care facility because lines can be inserted directly into veins or arteries to obtain continuous samples. However, such procedures are risky because infection can occur at the entry points, and measures must also be taken to prevent clotting. An implantable sensor-transceiver could provide continuous monitoring of a blood-borne chemical and general overview of IC-based chemical sensors appeared recently.9
In a person with insulin-dependent diabetes, the control regimen entails periodic doses of insulin, and the blood glucose level must be monitored to adjust the insulin dose as well as to monitor the level of control. Currently, a diabetic must rely on periodic finger sticks to obtain blood for analysis. According to Anderson, "Every clinician's next dream" would be to have a means of noninvasively monitoring blood glucose levels.10 In a best-of-all-possible-worlds scenario, a continuous blood glucose monitor would be combined with a programmable controller and an implanted insulin pump to replicate the function of the pancreas. Development of an artificial pancreas has been the subject of considerable effort over the past few years, yet there has been no successful demonstration of a satisfactory substitute.
As an intermediate solution, an implanted blood glucose monitor coupled to a transmitter would at least allow a diabetic to continuously monitor blood glucose levels without having to constantly prick a finger. Figure 4 shows a conceptual diagram of a blood glucose monitor, which would be attached to an artery on a permanent or semipermanent basis.11 The sensor would consist of a surface acoustic wave (SAW) resonator, the surface of which is coated with a thin layer of monoclonal antibody sensitive to the glucose molecule. Depending on the concentration of glucose in the bloodstream, the mass loading of the SAW resonator would change, leading to a change in the oscillation frequency. Because the quality factor (Q) of a SAW resonator is inherently very high, the device should be capable of very high resolution, even when Q is reduced due to the extra mass loading of the monoclonal antibody film and the blood that contacts the resonator surface.
Figure 4. Implantable continuous blood glucose monitor.
One problem associated with an implanted sensor is the possibility that the body will reject it as a foreign object. In the case of a tap into a vein or artery, the tissue in the surrounding area may become inflamed, or the opening may literally become walled off as a layer of fibrous tissue grows over it. A greater danger, however, is the possibility of clots forming in an attempt to wall off the opening: when detached, these clots can drift in the bloodstream until they become lodged in a vein or artery supplying the brain, which leads to a stroke. Hence, such in-dwelling sensors must be made of inert substances, and considerable work has already been done on artificial heart valves and other hardware purposely left behind during surgery. A further possibility might be to impregnate the sensor's housing material with a steroid that slowly diffuses out of the package matrix to minimize inflammation. This technique is used with cardiac pacemaker electrodes and could well work in this situation.12
Pediatric Sleep Apnea Monitoring. Pediatric sleep apnea is a breathing disorder in which a child's central nervous system fails to stimulate the body to take a breath during sleep. As a result, long intervals without a breath can occur, with respiration recommencing in response to a backup stimulus triggered by the lack of oxygen.
Present monitoring practice involves the use of a crib or bedside monitor, with wires connected to patch electrodes placed on the child's chest. Most such units monitor for muscle activity associated with breathing; some add a simple ECG. When breathing stops, an audible alarm sounds and wakes the child, at which time breathing resumes in response to a different controller.
With this approach, other muscle activity can yield false alarms; further, in ECG-augmented monitoring, by the time the heart rate drops in response to the hypoxic (low-oxygen) state, brain damage may have already occurred. Finally, children can choke on the loose wires. An integrated sensor-transceiver in this application would eliminate the choking hazard because no wires would be involved. Because existing techniques use patch electrodes as sensors, the instrumented ECG-patch electrode proposed earlier would apply equally well here. Nonetheless, there are inherent problems associated with trying to infer respiration from muscle activity or waiting until the heart rate drops to sound an alarm. Whittaker points out that the only sure way to monitor respiration is to measure airflow.13 Although this may seem self-evident, it adds a level of complexity to implementation.
The most reliable monitor would be a simple, lightweight, soft, compliant, comfortable mask that would cover the child's nose and mouth and contain a simple airflow-sensing switch. Air volume and flow rate don't need to be measured explicitly; rather, it is sufficient to determine that respiration has occurred within a preset timeout interval. Failure to breathe would result in an audible alarm, just as with existing monitors.
Figure 5. Implantable continuous blood pressure monitor.
A different approach, which is somewhat inferential but might be less intrusive to the child, would be to use a sensor similar to the instrumented ECG patch, but with a sensitive microphone instead of electrodes. The patch would attach to the chest and would sense the sound of air passing through the trachea; if desired, a backup ECG function could be retained. The detection process might be as simple as using a set of hardware or software filters to identify the spectrum of airflow-related sounds. A more sophisticated system might also discern airflow-related sounds from background noise with similar spectral content. Clearly, the more difficult the detection process, the more inferential the measurement becomes, and the less advantage a sensor would have over conventional techniques.
Pediatric Thermometry. Unlike the previous applications described, a pediatric thermometry application for integrated sensor-transmitters would provide a convenience rather than have a fundamental impact on patient morbidity or mortality. Also unlike applications mentioned earlier, pediatric thermometry represents a larger market, because these devices have much more consumer potential. Such devices address problems associated with spiking a fever: by the time a child becomes sufficiently uncomfortable to wake up, a considerable amount of time is required to reduce the fever and control the situation.
As with sleep apnea monitoring, wired solutions pose a choking hazard; therefore, an integrated temperature sensor-transmitter offers a simple means of avoiding wires. A system could be developed in which an elastic armband holds the sensor-transmitter in place in the armpit, which provides a compromise between an oral temperature (which is difficult to obtain during sleep) and cooler arm or leg measurements (which give a less accurate indication of core temperature). The receiving unit is fashioned like a typical aural baby monitor and has a threshold adjustment to set the alert point. Both aural and temperature monitors could be designed to transmit to a common mobile unit carried by a parent.
Continuous Blood Pressure Monitoring. Under certain conditions, it is necessary to monitor blood pressure continuously rather than every few minutes as provided by the conventional method, which uses a blood pressure cuff, manometer, and stethoscope. As an alternative, a microelectromechanical system (MEMS) pressure sensor combined with an integrated transmitter could—when attached to an artery—provide continuous, real-time monitoring of blood pressure (see Figure 5).14 A battery and loop antenna in an inert housing would complete the package.
Figure 6. Soft tissue loss measured with a small test cell.
The use of invasive blood pressure sensors would be problematic. Current noninvasive methods for continuously monitoring blood pressure are based on electrical measurements at the surface of the skin and may prove to give adequate performance. A MEMS-based sensor is, nonetheless, under active development, and the concept serves as an indicator of what is possible: the integration of micromachined sensors with transceiver electronics might be used for other pressure measurement applications or to deliver tiny, controlled amounts of medication using a MEMS diaphragm as a pump (a microimplanted infusion [intrathecal] pump).15
Patient, Caregiver, and Equipment Monitoring. Another application of monolithic RFICs—one with potentially very high volume—lies in the realm of RFID tags to track patients, caregivers, and even equipment in hospitals. In fact, this growing area is already reaching a level of maturity.16,17
Given the increasing number of patients in hospitals and the implications of mistakes in the administration of treatment or medication, improving methods to verify patient medication or treatment will decrease the likelihood of errors. Identification bands containing an RFID tag that could automatically provide name and ID number, flag known allergies, and transmit other critical data, could both automate and safeguard patient care and recordkeeping.
Beyond simple documentation, however, RFID tags can be used to monitor the location and identification of high-risk patients, such as newborns or patients with limited cognitive abilities. Pagers equipped with RFID-like transponders could be used to locate physicians for emergencies. Similarly, RFID-tagged equipment can be readily located for calibration and maintenance or when needed for a procedure.
Other Considerations
Economic and practical considerations must be considered when contemplating the integration of RFIC-based devices into medical electronics.
Economic Considerations. No matter how advanced or advantageous a device might be, it is not a viable venture if someone won't pay for it. In today's healthcare environment, it is often the government or private insurers and managed care organizations that control the fee structure and determine what procedures will be covered.
An interesting case is the ECG monitor and emergency notification system proposed in the section on cardiac monitoring. Such a system, which used an OEM cell-phone module as the RF link coupled to a small six-lead ECG module, was actually developed and built. A patient would wear the unit on an arm or in a small holster, and the system would communicate directly with the monitoring service from anywhere within range of a cellular base station. This proposed system was well-received by caregivers. It was abandoned, however, because insurers would not pay for its use.18
Practical Considerations. Some practical hurdles must be overcome in order to successfully integrate RF-based sensor-transceivers into medical electronics. Specifically, the increasing number of electronic devices in hospitals leads to growing concern over electromagnetic interference, and new devices must not only provide robust operation but must also be designed so that they do not interfere with other devices.19
Another concern relates to signal attenuation due to the lossy nature of tissue and bone, which determines the power required to communicate reliably with an external system. In an effort to gain insight into the effects, and because no data were readily available, a small test cell (see inset of Figure 6) was designed and built to measure soft tissue loss. With the cell inserted between the ports of an HP85046 S-Parameter test set connected to an HP8753B network analyzer, the transmission was measured both with and without a subject's nose inserted between the electrodes of the cell. The plots of Figure 6 show the results. For frequencies in the megahertz range, the loss is virtually zero. By 3 GHz, the loss increases to 30 dB of attenuation.
A final concern involves power, both in terms of the power supply and the radiated power from the antenna. Particularly in the case of an implanted device, the battery must provide adequate operating life. At the same time, the radiated power must not harm surrounding tissues, so keeping output power low (and amplifier efficiency high) maximizes battery life. Batteries currently used for implanted cardiac pacemakers and defibrillators provide years of service. In principle, these should provide a reasonable life for an integrated sensor-transceiver. However, because sensor-transceivers are smaller than pacemakers, they would not allow as much room for a battery, which would limit the available power.
Conclusion
The ability to combine sensors, baseband signal processing, and RF transceiver circuitry into a single integrated circuit opens up a variety of medical monitoring possibilities. Areas of cardiology, neurophysiology and internal medicine stand to reap potentially significant benefits in terms of enhanced monitoring capability, patient convenience, and patient safety. The use of RFID-like tags in hospitals can minimize mistakes in the administration of medication, guard against patient misidentification, and track equipment. Home healthcare applications also represent a potentially large consumer market for devices using integrated sensor-transceivers.
References
1. A Rofougaran et al, "A Single-Chip 900-MHz Spread-Spectrum Wireless Transceiver in 1 µm CMOS Part I: Architecture and Transmitter Design," IEEE Journal of Solid-State Circuits 33, no. 4 (1998): 515–534.
2. FH Netter, "Heart," in The CIBA Collection of Medical Illustrations 7 (New Jersey: CIBA Pharmaceutical Co., 1986).
3. The APEX S Transmitter, by Marquette Medical Systems (Milwaukie, WI), is a good example of a portable wireless ECG monitor designed for the hospital environment (http://www.mei.com). The HP Medical Products Group (Andover, MA) offers a broad line of in-hospital monitoring products (see the "Healthcare" category on http://www. p.com).
4. J Mersch and K Cook, "Technology Whose Time Has Come: The Stanford Transtelephonic Arrythmia Network," Stanford Nurse 18, no. 1 (1996).
5. The King of Hearts arrythmia monitor and Lifesigns System ECG/blood pressure/blood oxygen saturation monitor, both by Instromedix (Hillsboro, OR), are examples of portable ECG monitoring instruments.
6. Heartcard by Instromedix (Hillsboro, OR).
7. Medtronic Inc. (http//www.med tronic.com) offers a good introduction to pacemaker technology and the practical issues associated with pacemaker use.
8. N So, neurophysiologist (private communication).
9. HT Nagle et al, "Special Report on Electronic Noses," IEEE Spectrum 35, no. 9 (1998): 22–38.
10. T Anderson, internist (private communication).
11. K. Bates, Pangea Medical Inc. (private communication).
12. Medtronic Inc. (http://www.mei.com).
13. K Whittaker, pediatrician (private communication).
14. E Hynes, Analog Devices Inc. (private communication).
15. Medtronic implantable pumps infuse medication directly into the spinal fluid from within the abdominal cavity.
16. E Worthman, "From 125 Hz to Microwave—The Technology of RFID," RF Design (1998): 38.
17. C Matsumoto, "'Indoor' Positioning System Hits Market," Electronic Engineering Times (1998): 58.
18. C Davis, BioSight Inc. (private communication).
19. A Berger, Sparling Corp. (private communication). Sparling Corp. specializes in the design of hospitals.
Frederick G. Weiss, MSEE, PhD, was formerly a senior design engineer at the Northwest Laboratories of Analog Devices (Beaverton, OR). He has been involved for 20 years in the design of inegrated circuits for analog, mixed-signal, data-conversion, and RF applications. He has gained insight into the medical world from his wife, who is an internist. Comments on this article should be directed to Scott Wayne at Analog Devices (Wilmington, MA).
Back to the Fall 2000 Table of Contents
