340

IEEE TRANSACTIONS ON BIOMEDICAL CIRCUITS AND SYSTEMS, VOL. 4, NO. 6, DECEMBER 2010

A Miniature-Implantable RF-Wireless Active Glaucoma Intraocular Pressure Monitor

Eric Y. Chow, Member, IEEE, Arthur L. Chlebowski, Student Member, IEEE, and Pedro P. Irazoqui, Member, IEEE
AbstractGlaucoma is a detrimental disease that causes blindness in millions of people worldwide. There are numerous treatments to slow the condition but none are totally effective and all have signicant side effects. Currently, a continuous monitoring device is not available, but its development may open up new avenues for treatment. This work focuses on the design and fabrication of an active glaucoma intraocular pressure (IOP) monitor that is fully wireless and implantable. Major benets of an active IOP monitoring device include the potential to operate independently from an external device for extended periods of time and the possibility of developing a closed-loop monitoring and treatment system. The fully wireless operation is based off using gigahertz-frequency electromagnetic wave propagation, which allows for an orientation independent transfer of power and data over reasonable distances. Our system is comprised of a micro-electromechanical systems (MEMS) pressure sensor, a capacitive power storage array, an application-specic integrated circuit designed on the Texas Instruments (TI) 130 nm process, and a monopole antenna all assembled into a biocompatible liquid-crystal polymer-based tadpole-shaped package. Index TermsBiomedical applications of electromagnetic radiation, biomedical monitoring, biomedical telemetry, implantable biomedical devices.

Fig. 1. Glaucoma IOP monitor implant.

I. INTRODUCTION

LAUCOMA stems from a buildup of intraocular pressure (IOP) in the anterior chamber of the eye which mechanically constricts and damages the optic nerve, eventually leading to irreversible blindness. Primary open angle glaucoma is the most common type and results from occlusion of the drainage pathway presented by the trabecular meshwork and Schlemms Canal. The trabecular meshwork is an area of tissue in the eye responsible for draining aqueous humor from the anterior chamber, which is between the iris and the cornea. The aqueous humor then passes through a set of tubes known as Schlemms canal where it then ows into the circulatory system. Glaucoma is predicted to affect about 60.5 million people by 2010, of which 4.5 million will develop bilateral blindness [1].
Manuscript received May 25, 2010; revised September 14, 2010; accepted September 21, 2010. Date of publication November 09, 2010; date of current version November 24, 2010. This work was supported by SOLX Inc. This paper was recommended by Associate Editor H.-J. Yoo. E. Y. Chow was with the Center for Implantable Devices, Weldon School of Biomedical Engineering, Purdue University, West Lafayette, IN 47907 USA. He is now with Cyberonics Inc, Houston, TX 77058 USA (e-mail: eric.chow@cyberonics.com). A. L. Chlebowski and P. P. Irazoqui are with the Center for Implantable Devices, Weldon School of Biomedical Engineering at Purdue University, West Lafayette, IN 47907 USA (e-mail: achlebow@purdue.edu; pip@purdue.edu). Color versions of one or more of the gures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identier 10.1109/TBCAS.2010.2081364

Oftentimes in cases of open angle glaucoma, the individual has no symptoms or warning signs, and the disease gradually progresses leading to eventual blindness. Current clinical devices do not provide 24-h monitoring and may not be sufcient for at-risk patients due to the fact that IOP varies signicantly throughout the day and can be substantially greater at some times such as during intense activity or while sleeping [2], [3]. This paper focuses on the development of an implantable device, shown in Fig. 1, which can provide continuous measurements of the internal pressure of an eye. This data can be used to monitor the progression of the disease in glaucomatous patients and can hopefully help prolong vision and even prevent blindness. There are numerous continuous glaucoma IOP monitoring devices in the research-and-development phase. A popular technique is integration of a strain gauge pressure sensor into a soft contact lens. The Triggersh developed by Sensimed AG [4] has developed a method to make this approach completely wireless. The approach faces problems due to the fact that the nature and positioning of a contact lens causes the measurements to suffer from considerable noise and variability due to eye movement, blinking, and external environmental conditions, such as moisture, that could alter the form of the lens [5][7]. Internal glaucoma pressure monitors do not face those same problems and numerous works have described implantable passive LC resonant-based systems [8][15], which promote a device that is simple, robust, and somewhat small (maximum of 36 mm in its widest dimension). The main drawbacks of these passive devices are very limited functionality, only on-demand measurements, and the constant requirement of a nearby external data acquisition unit. Active implantable IOP monitors have also been explored and two of the leading active devices in the area are presented in [16] and [17]. These devices, which use an implantable lens approach, are relatively large, use inductive power coupling which limits the range and requires perfect alignment with an external powering coil, and still constantly requires a nearby external device.

1932-4545/$26.00 2010 IEEE

CHOW et al.: A MINIATURE-IMPLANTABLE RF-WIRELESS ACTIVE GLAUCOMA INTRAOCULAR PRESSURE MONITOR

341

In this paper, we describe in detail the design of an active wireless implantable IOP monitoring system, introduced in [18], which consists of an application-specic integrated circuit (ASIC), microelectromechanical systems (MEMS) sensor, antenna, and capacitive powering array. The MEMS sensor is used to convert pressures into capacitance and is fed into the ASIC for data processing and modulation onto a high-frequency carrier for wireless transmission. The assembled device and liquid-crystal polymer (LCP)-based package is shown in Fig. 1 and is roughly the shape of a tadpole to facilitate trochanter-based surgical implantation in the anterior chamber of the eye. II. SYSTEM DESIGN Our glaucoma IOP monitoring device is a fully wireless active low-power measurement system. The primary advantages of this active system include the potential for a vast increase in functionality and the ability to operate for extended periods of time without a constant nearby external device which powers and/or extracts data. The ASIC is designed on the Texas Instruments (TI) 130 nm technology and consists of a sensor interface, voltage regulators and references, radio-frequency (RF) rectier for remote powering, and wireless transmitter. To support the wireless features, a monopole antenna is designed and integrated onto a Rogers Corporation LCP substrate. To allow for independent operation, a capacitor array that ts within the size constraints of our miniature package is used to provide power storage and eliminate the need for a continuous nearby external device. A. Capacitance-to-Time Conversion An ultra-low-power measurement technique is used in the sensor interface and utilizes a nanoamp current to charge the MEMS capacitor converting capacitance variations to time changes. The capacitance-to-time circuit is shown in Fig. 2 where the top plate of the MEMS sensor is connected to the cap pin and a PMOS current source supplies current to that pin. When the node voltage reaches the threshold of the Schmitt trigger, the output switches logic levels from high to low. The time between when the current source begins charging up the top plate and when it reaches the threshold is directly proportional to the capacitance of the sensor. This block is reset by setting the rst pin to a logic level high causing the to drain out all of the current from the NMOS transistor cap node. The full measurement circuit consists of two capacitance-totime circuits, as shown in Fig. 2. One circuit has the MEMS capacitive sensor and the other has a reference capacitor with a value that is equal to the base capacitance of the sensor. The current sources in the two capacitance-to-time circuits are matched and provide the equal currents to their corresponding capacitors. The MEMS sensor will always have an equal or larger capacitance compared to the reference capacitor resulting in a longer charge-up time. Digital logic is used to create a pulse that transitions to a logic level high after the reference capacitor charges up to the threshold of the Schmitt trigger. A short time later when the MEMS capacitor node reaches the threshold of its Schmitt trigger, the pulse switches back to a logic level low.

Fig. 2. Simplied schematic of the capacitance-to-time circuit.

The reference capacitor will be an off-chip component and is necessary to account for variations in the base capacitance of the MEMS sensor. In one measurement process, the circuit rst comes off of its reset state and the MEMS sensor and reference capacitors begin to charge up using a nanoamp current. The capacitance-to-time circuit is shown in Fig. 2 where the top plate of the MEMS capacitive sensor is connected to the drain of . The voltage on the the nanoamp PMOS current source MEMS capacitor is given by (1) where is the time in which current is being sourced and is the capacitance of the MEMS sensor. , the drain current of the PMOS current source in saturation and accounting for channel length modulation, is given by

(2) where is the technology parameter equal to , is , and the channel length modulation parameter of is the drain voltage of , which is equal to the voltage on the . This PMOS current source is used MEMS capacitor to output current on the order of a nanoamp but this current, as seen in (2), is initially greater and diminishes as the voltage on the MEMS sensor increases due to channel length modulation and eventually coming out of saturation towards the end of the cycle. The initial greater current levels allow the circuit to quickly charge-up the base capacitance of the sensor. As the end of the charge cycle is reached, the remaining capacitance, representing the pressure reading, is charged up with a smaller current value allowing for relatively larger time variations and thus improved sensitivity. This effect is advantageous when considering low-power, low-delay, and sufcient sensitivity of the circuit because it charges up the larger base capacitances quickly while achieving sufcient time changes for the smaller pressure-dependent capacitance-variation.

342

IEEE TRANSACTIONS ON BIOMEDICAL CIRCUITS AND SYSTEMS, VOL. 4, NO. 6, DECEMBER 2010

Using (1) for

in (2) the drain current is given by (3)

where the the constant

and overdrive squared terms are absorbed into . Using (3) in (1) and solving for gives (4)

reaches the threshold of the Schmitt trigger, When the output switches logic levels from high to low. Setting to this threshold in (4) gives the time it will take to charge the MEMS capacitor. The actual current source value of about 1 nA is determined by using (4) based on an optimization where the sensitivity of the digitization, measurement time, and power consumption are accounted for. This time between when the current source begins charging up the top plate and when it reaches the threshold is directly proportional to the capacitance of the MEMS sensor and, thus, the pressure. Since both capacitors are being charged with matched current sources, the reference capacitor reaches the threshold of the Schmitt trigger rst while the MEMS sensor takes a little more time to charge up to the same voltage. This time difference represents the MEMS capacitor variation and logic is used to convert the signals into a pulse. This pulse is then digitized and wirelessly transmitted or stored into memory in subsequent stages. B. Voltage Regulator and References The voltage regulator is supplied by the power capacitor array which can vary from 4 V down to 0 V. To maximize the usable charge, a common PMOS with operational amplier feedback topology is used, which allows for low dropout voltage operation [19], [20]. An NMOS-based operational amplier design with an NMOS current source is used in order to guarantee a constant overdrive voltage over the range of the supply. The regulator requires a reference that is stable across the voltage variation of the capacitor powering array and, thus, a modied form of the supply-independent reference circuit from [21] is used. Diode-connected transistors are used in both branches of this reference topology to achieve the desired voltage while minimizing area and reducing power consumption. To minimize the overall average power consumption of the system, a picowatt timer, similar to that presented in [22], designed to track 5-min intervals, can be used to gate the charge supply to the regulator. C. Digital Conversion Digitization is required for noise immunity and to allow for storage into on-chip memory. Digital conversion is done using an 8-b ripple carry counter with an asynchronous reset function that is designed to measure the corresponding width of the pulse, as shown in Fig. 3. The counter is controlled by a clock, generated using a current-starved ring-oscillator topology, whose frequency is determined based on the pulsewidths, specications of the MEMS sensor, desired pressure range of 50 mmHg, and required sensitivity of at least 0.5 mmHg. The pulse controls the clocking of the binary counter, and the falling edge of the pulse is used to stop and reset the counter and prepare it for the next

Fig. 3. (a) Block diagram of the digital conversion architecture. (b) Timing diagram of a sample digital conversion process showing waveforms at labeled nodes a, b, c, and d.

conversion. After the count sequence, the value of the counter is fed into a parallel-to-serial converter, designed using a standard 12-b shift register topology, to serially stream out the digital data either directly to the wireless transmitter or into on-chip memory. D. On-Chip FeRAM Memory To enable operation without a constant nearby external device, the system must integrate a memory storage unit. Nonvolatile memory is desirable for our ultra-low-power application because it can retain stored data while remaining unpowered. For our application, nonvolatile memory known as ferroelectric random-access memory (FeRAM) is used. FeRAM technology is based off the idea of using an electric eld to physically move a Zirconium/Titanate atom from one position within the molecule to another [23]. FeRAM has many benets over the much more common ash memory (a type of electrically erasable programmable read-only memory or EEPROM), including lower operational powers/voltages, faster write time, and a much higher limit for read-write cycles. These characteristics are advantageous for our application while the main drawback of FeRAM of lower storage densities is insignicant due to our small requirement for data storage capacity. FeRAM is supported on our selection of a specic Texas Instruments CMOS fabrication process and enabled us to integrate the technology with our ASIC on a monolithic substrate. Although the FeRAM and some control circuitry are implemented on-chip, additional control circuitry is currently implemented off-chip for addressing and reading/writing operations. This control circuitry also operates the system on a sleep-wake cycle to minimize power while storing data points every 5 min into on-chip FeRAM. In a wake period, the system rst reads an

CHOW et al.: A MINIATURE-IMPLANTABLE RF-WIRELESS ACTIVE GLAUCOMA INTRAOCULAR PRESSURE MONITOR

343

Fig. 4. Glaucoma IOP monitor implant in Ansoft Corporations human body 3-D model along with antenna radiation pattern simulation results.

address out of FeRAM, then digitizes and writes the MEMS capacitance value into memory, and nally writes an incremented address into memory. E. Antenna Design The wireless features of our glaucoma IOP monitor are enabled by an antenna, a 2.4-GHz transmitter for wireless data transfer, and a radio-frequency (RF) rectier for remote powering and recharging. The antenna is a monopole-like structure and utilizes the tail portion of the tadpole substrate. This antenna is modeled using Ansofts High Frequency Structural Simulator (HFSS) in free space and implanted in tissue, as shown in Fig. 4. For the implanted simulations, the Ansoft Human Body Model is used, which provides a reasonably accurate 3-D model composed of half a million tetrahedra with empirically measured material properties of permittivity and conductivity. HFSS is used to simulate the input impedance of the tadpole antenna and roughly model the antenna pattern, shown in Fig. 4, when accounting for tissue effects on electromagnetic wave propagation. The nal antenna topology is a curved monopole/whip antenna whose length is approximately a quarter wavelength at 2.4 GHz and the simulated gain for this structure is about 6 dBi. The antenna shape and dimensions are chosen rst based on clinical requirements and then adjusted and optimized for electromagnetic performance. Orogolomistician surgeries on live rabbits are performed to quantify the tissue effects on wireless ocular implants and show 45 dB power loss at 2.4 GHz [24]. F. Capacitive Power-Storage Array An ultra-low-power ASIC design is absolutely crucial in the design of the IOP monitoring system. Current miniature battery technologies are relatively large and would exceed the size constraints. To meet the clinical size requirements, high-dielectric ceramic surface-mount capacitors are used to provide a power storage unit. Miniature 0201 capacitors are used to meet the thickness constraint of 300 m and we chose the largest readily available capacitance of 0.47 F for our corresponding package. The size constraint of our device enables us to integrate 52 of these capacitors and this array has the potential to power the complete system over a 24-h period. For completely independent operation, on-chip memory is also required to store the pressure data. At the end of every 24-h monitoring period, an external device is brought nearby which wirelessly downloads the data and recharges the capacitor array.

Fig. 5. Cross-sectional and front view of the eye showing the target implant location.

G. Form Factor and Packaging In order to create the packaging to house the IC and all other electronics, a biocompatible hermetic package is designed using liquid-crystal polymer (LCP). When compared with other materials, including polyimide, parylene, and silicone, LCP has very ) and is exible, biocompatlow moisture absorption ( ible, has high reliability under harsh environments, and can be thermally bonded without adhesives [25][27]. Chronic 4 week animal studies are performed to assess the biocompatibility of the LCP in relation to other radio-transparent materials, such as low-temperature co-red ceramics, and alumina. Out of all the materials implanted in the 4-week studies conducted in rabbits, the LCP resulted in the least amount of brous encapsulation and inammation. The LCP even showed improved biocompatibility when compared with alumina, a commonly used biocompatible material for implantable devices [28], [29]. The encapsulation layer thickness around LCP was 43% less than that around alumina after 1 week, 75% less after 2 weeks, and 97% less after 4 weeks. LCP is also a exible material that accommodates the shape of the implantation location, which for the target ocular application is the anterior chamber of the eye. Two layers of LCP are bonded together through a thermal process, hermetically sealing the internal electrical components from the aqueous humor uid in the eye as well as other interstitial uids. The LCP is available with a layer of copper clad on each side allowing for the etching of traces using a mask photolithography process and a ammonium persulfate etchant. Long-term biocompatibility and stability tests on LCP are presented in [30] where inammation and other complications are examined over a 3-month period. Accelerated soak tests on LCP are also presented in [30] to evaluate long-term reliability through at temperatures from 37 C up to 75 C and show no degradation and good long-term biocompatibility. The nal design of the LCP packaging is in the shape of a tadpole with a tail enclosing an antenna and a head to house the electronics. This tadpole structure ts the clinical requirements and allows for a sufciently sized antenna for RF telemetry and powering and a miniature enclosure for all of the electrical components. When inserted properly, the tail will curve along the

344

IEEE TRANSACTIONS ON BIOMEDICAL CIRCUITS AND SYSTEMS, VOL. 4, NO. 6, DECEMBER 2010

Fig. 6. Simplied schematics and block diagram of just the wireless telemetry and powering blocks in the implanted and external devices.

rim of the anterior chamber, as shown in Fig. 5, and not interfere with the sight of the patient. The chosen packaging shape also facilitates implantation via a trochanter technique. On the two ends of the tadpole are rings that are used to insert the implant. The trochanter is comprised of a hook connected to an extendable wire, and when the tadpole is attached to the tool, the tail is inserted into the hollow core of the trochanter leaving the head exposed. Upon making an incision 3 mm in diameter, corresponding to the maximum width of the implant, the head of the device, which is connected to the hook, and tip of the trochanter are inserted into the anterior chamber. Next the hook extends from the trochanter and advances the implant in a circular unwinding fashion into the anterior chamber of the eye. Once the tail of the tadpole has completely exited the tool, the two rings are used to precisely place the implant in the target location. The trochanter hook and extended wire are then extracted from the incision in the eye, and the incision is sealed using ophthalmic stitches. H. Wireless Transmitter A 2.4-GHz wireless transmitter is designed on the Texas Instruments 130-nm CMOS process and custom made to feed the tadpole antenna, provide enough power to ensure successful data reception when implanted, and operate at low levels of current consumption. To allow for wireless telemetry, the data are modulated onto a 2.4-GHz carrier through an on-off keying (OOK) modulation scheme. OOK is used as a simple low-power modulation scheme, which, on average, consumes about half the power of a constant carrier scheme. A complementary cross-

coupled transmitter (TX) topology, shown in Fig. 6, is used for its feasibility of on-chip implementation, relatively low phase noise, and low power consumption for comparable performance with other topologies. The oscillator is turned off during the off cycles by controlling the NMOS current source bias voltage. This technique inuences the maximum data-rate due to a nite settling time but has the advantage of reducing the average power consumption of the system which is weighted heavier in this target application. A buffer is necessary to isolate the oscillator from the external connection to the antenna. Although other power-amplier types can achieve greater efciencies, a self-biased class AB power amplier is chosen for this work because it provides a good compromise between efciency, low phase noise, and minimization of die area. I. RF Wireless Powering Wireless powering of the system is performed using a highfrequency wave which feeds an RF rectier circuit through the antenna. This initial version of this RF powering link operates a 3.65 GHz which is a lightly-licensed band for unrestricted protocols commonly used for WiMAX and WiFi applications [31]. This frequency is sufciently removed from the 2.4-GHz telemetry band and allows for implementation of an off-chip frequency diplexer consisting of bandstop lters to isolate the RF powering and the telemetry path. A CockcroftWalton multiplier, often referred to as the Greinacher multiplier, is used in this paper as the fundamental block of the RF rectier circuit. The multiplier produces sufciently high voltages with relatively low input power levels when compared to various other structures including the PMOS

CHOW et al.: A MINIATURE-IMPLANTABLE RF-WIRELESS ACTIVE GLAUCOMA INTRAOCULAR PRESSURE MONITOR

345

voltage multiplier, full-wave diode rectier, and gate cross-connected bridge rectier [13]. Schottky diodes are chosen for our RF rectier due to their high-frequency performance and low forward-bias voltage. This low voltage (150 mV 300 mV) allows for rectication at lower input biases which is desirable for low-power operation. Capacitance values for a short-term (24 h) power-storage unit are chosen so that the time constant is signicantly larger than the interval period of the pulsed RF power signal. This large time constant ensures low ripple in the output dc voltage, which places less burden on the subsequent voltage regulators. Two stages are implemented to further increase the output voltage to maximize the charge delivered to our capacitive powering array. Increasing the number of stages quickly approaches diminishing returns due to the current drain of later stages, so a complementary design shown in Fig. 6, is also used to achieve a doubling of the output voltage. The complementary architecture also yields additional advantages such as a smaller input impedance and less reected harmonics due to the symmetry [32]. The overall efciency of the RF rectier circuit is given by (5)

Fig. 7. Optical microscope picture of ASIC.

is the output voltage, is the diode saturation curwhere is the output current, is the rst order modied rent, is the thermal Bessel function, is the number of stages, and voltage for the diodes. J. External Receiver For this application, the implant has much more stringent requirements than the external device, especially in terms of complexity and power consumption. The transceiver link is therefore optimized as an asymmetric link where the minimal power draw of the implanted transmitter is priority and the complexity is all pushed over to the external receiver side. A custom-designed receiver architecture is used to decode the unlocked-carrier, noncoherent, 2.4 GHz OOK-modulated transmission from the implant. The receiver, whose block diagram is shown in Fig. 6, consists of an antenna, low-noise amplier (LNA), downconverter, intermediate-frequency (IF) ampliers and lters, a demodulation and digitization block, and a computer interface circuit. After reception from the antenna, passing through the L-match network, and RF amplication through the LNA, the signal is downconverted through a noncoherent, single-conversion, low-IF architecture [33]. After downconversion, IF amplication and ltering/isolation are performed. Two MAXIM IC LNAs (MAX2650) are used which provide 20 dB of gain at 120 MHz (the MAX2650 LNAs actually have a at gain response from dc to 900 MHz, which allows for exibility in the IF value). A T-network 3-pole Butterworth band-pass lter is designed with a 120-MHz center frequency and 20-MHz pass-band optimized for the transmitter frequency jitter resulting from the unlocked nature of the low-power transmitter. Note that matching networks are designed and implemented between each component to maximize the power transfer and, thus, signal integrity from stage to stage.

Demodulation of the OOK-modulated signal is performed through the use of a Linear Technology (LT5534) RF power detector. The monolithic power detector is congured to operate on a 3-V source and detect signals down to 60 dBm over a frequency range of 50 MHz to 3 GHz. Input impedance is matched of 4.25 dB. Logic high occurs to 120 MHz, providing an during 120 MHz, and logic low occurs when low RF energy is present in the detection band. The 500-MHz passive low-pass lter is used prior to the power detector to decrease the high frequency out-of-band noise. The resulting digital data is then buffered and sent to a universal asynchronous receiver/transmitter (UART), converting from digital-to-universal serial bus (USB), which enables interfacing with a PC. An FTDI USB module (FT4232H) is used to read the serial digital data-stream into the computer, and the USB chip is congured to use the serial peripheral interface bus protocol (SPI) to read data in. Data streamed through the SPI are then digitally ltered and processed on a PC. III. MEASUREMENTS AND RESULTS The IOP monitoring application specic integrated circuit (ASIC), shown in Fig. 7, is fabricated through the Texas Instruments (TI) 130-nm CMOS process. A. Wireless Telemetry and Powering The fabricated 2.4-GHz transmitter consumes 812 A from a 1.5-V regulated supply and outputs 45 dBm. The on-chipfully-symmetric differential inductor with a patterned ground shield has a measured of 10.36 at 2.4 GHz. The data rate is inversely proportional to the system average power consumption when using a bursty transmission technique since faster wireless telemetry reduces the transmitter on-time for a given amount of data. The nite startup time of the on-chip oscillator, simulated to be about 150 ns, limits the maximum date rate. Through empirical testing, an optimal point in free-space is found at a data rate of 8 Mb/s when considering maximizing the data rate and minimizing the bit-error rate (BER). It is important to note that even though the actual measured values of BER depend on distance, the trend of BER versus data-rate is independent of distance. At 8 Mb/s and a free-space distance of 50 cm, the transceiver link has a sensitivity of 80 dBm and a measured BER of

346

IEEE TRANSACTIONS ON BIOMEDICAL CIRCUITS AND SYSTEMS, VOL. 4, NO. 6, DECEMBER 2010

Fig. 8. RF powering IC power conversion efciency (PCE).

7.5 , when tested over 1 b. The transceiver link is also characterized through in vivo studies where across a 50-cm telemetry-link distance, the BER at 8 Mb/s is on the order of . In the current implementation, a standard communication protocol is not used and the receiver simply listens to data and takes it as is. As a result, errors are likely in a noncontrolled environment, and the next revision of this system utilizes a standard communication protocol and incorporates error detection and correction schemes. The RF powering module is integrated with the full IOP monitoring system and measurements, plotted in Fig. 8, and show that with a 20 mW (13 dBm) input, the rectier has a power conversion efciency (PCE) of 11.3% and produces 1.5 V across a 1-k load which is sufcient for biasing standard CMOS process nodes and supplying adequate current for most low-power designs [34]. Friis path loss formula estimates a 10 dB reduction in power across 10 cm, and coupled with the 45 dB power loss though ocular tissue in an implanted setting necessitates a 0.50.6 W transmit power from the external source. Note, that this derivation does not take into account the loss/gain of the implanted antenna. In our particular setup, the 6-dBi loss of the implanted antenna in conjunction with the external horn antenna with gain of 8 dBi, results in a 0.8 W0.9 W power output requirement from the external source into the antenna. Considering the power density seen at the surface of the eye when the horn is placed 10 cm in front of the face and using the density of 1.0071 g/cm for vitreous humor, the specic-absorption-rate (SAR) seen in the eye is calculated to be about 3.988 W/kg [35]. According to the IEEE standard for safety levels (IEEE C95.1-2005), the basic requirement for localized exposure of the general public is 2 W/kg while for people in controlled environments, the limit is 10 W/kg [36]. Currently, this system is within the controlled environment specication but exceeds the general public requirement. Improvement in rectier efciency and decreased power consumption of the system enables it necessary to meet the general public specication and to increase operating distances and/or achieve greater implantation depths. B. Pressure Measurement For the initial testing of the capacitance-to-time measurement block, variable capacitors are used to set the base and sensor capacitances. The capacitances are set using an automated stepper motor integrated with a custom-made plastic screwdriver to precisely set the values. The capacitances are veried using an Ag-

ilent 4284A Precision LCR Meter. The output of the capacitance-to-time measurement circuit is captured by probing the output of just that block after a buffering stage. The data show that the greatest change in pulse-width is when the difference between base and sensor capacitances is below a few picofarads. The measured pulse width change with a small difference be. Theretween base and sensor capacitances is 2.2316 fore, the optimal choice for the value of the base capacitance is the capacitance value of the sensor which corresponds to the low end of our pressure range. This choice of base capacitance minimizes its capacitive difference with that of the sensor and produces the greatest change in pulse width for pressure variations, and minimizes the time required to complete a measurement. The measured rate of pulse-width change of 2.2316 is actually higher than the simulated rate of 1.324 . This increased rate-of-change does not necessarily mean a more optimal performance since greater times also result in longer duration for a measurement and, thus, more power consumption. This difference between empirical and simulated data is primarily due to process variations resulting in a difference in the current source output. The sensor interface is then integrated with the MEMS capacitor, base capacitance, and tested with the wireless enabled. The MEMS capacitors are rst quantied through measurements done in a pressure regulated chamber and the capacitance variation (averaged over 4 sensors) is plotted in Fig. 9. The base capacitance is 5.3 pF, the average sensitivity is 6.64 fF/mmHg (with a maximum variation of about 9.7 fF/mmHg). For testing and validation, the output of the capacitance-to-time circuit is fed directly into the wireless transmitter. An external receiver is used to receive the wireless signal and perform demodulation to recover the data. Empirical tests involved enclosing the fully functional wireless prototypes inside a custom-built pressure chamber while the wireless power source and data receiver are positioned externally. Pressure measurements are taken by sweeping pressure within the chamber while wirelessly supplying power to the device and receiving the RF telemetry. Empirical measurements of the prototypes are plotted in over Fig. 10(a) and show a rate-of-change of 47.4 a 050 mmHg range with a correlation of capacitance to pulse width of 140.084 fF/ms. To capture 23.7 /0.5 mmHg changes and thus achieve a pressure resolution of at least 0.5 mmHg, the digitalization module is clocked at 42.2 kHz. The on-chip-based oscillator has a nite jitter which adds to some measurement errors; however, this on-chip implementation has the benet of reducing external component count and these discrepancies are alleviated through averaging. Measured pressures, plotted in Fig. 10(b), show the same trend as simulated and actual pressures and after multiplication with a calibration constant of . This cal0.593, there is an average error of ibration constant based on measured versus simulated results is just an example and unique calibration constants need to be calculated on a situational basis to account for biological, process, and device variations. The nal product will be calibrated with the Goldmann applanation tonometer, the reference standard for measuring IOP [37]. Process variations are the major cause for the discrepancy between measured and simulated pulse-width changes. An additional effect contributing to the discrepancy

CHOW et al.: A MINIATURE-IMPLANTABLE RF-WIRELESS ACTIVE GLAUCOMA INTRAOCULAR PRESSURE MONITOR

347

TABLE I MEASURED AND SIMULATED POWER CONSUMPTION OF ASIC COMPONENTS

Fig. 9. Measurement of the MEMS capacitance variation with pressure.

TABLE II ULTRA-LOW-POWER ACTIVE SYSTEMS FOR IOP MONITORING APPLICATIONS

Fig. 10. Measured performance (a) of the capacitance-to-time measurement circuit and (b) full monitor with and without calibration alongside the simulated results as a function of actual pressure.

is the nite series resistance resulting in leakage through the capacitors especially if this leakage is signicantly different between the MEMS capacitor and base capacitor. C. Comparison With Existing Devices The measured power consumptions of each block are tabulated in Table I alongside Spectre simulated results. Most of the blocks consumed slightly more power than the simulations predicted due to process variations and unmodeled parasitics from the die, bonding process, and test boards. The wireless transmitter is actually able to operate at much lower currents than that shown in simulations. This current is controlled by a bias voltage which is reduced to a value that minimizes power consumption but still ensures the output of a stable oscillation.

The system, excluding the wireless transmitter, is measured to consume 202.43 W during its 1-ms wake period. From simulations, the FeRAM requires 180 pC and 39.75 9 s for the 1 read and 2 write operations, which results in 270-nW average power consumption over the 1-ms measurement cycle. A 24- F capacitive power storage array enables continuous independent operation and when fully charged, stores 48 of usable charge. Using a picowatt timer, similar to that presented in [22], to track the 5-min intervals, the sleep-mode charge consumption over a 24-h period is calculated to be 86.4 nC. After subtracting the sleep-mode charge consumption, the remaining capacitor charge is able to supply the 202.7 W wake period power draw for 360 cycles, which allows for recordings every 5 min throughout one day. The average power consumption of the device over this 24-h period is simulated to be 675.67 pW. At the end of a 24-h period, an external device is brought nearby to wirelessly download the data and recharge the device. A comparison with published state-of-the-art low-power active IOP monitors, tabulated in Table II, shows comparable power consumptions during wake mode; however, since our system architecture, along with the FeRAM, allows for sleep-wake cycles, the average power consumption is more than 200 000 times less than other IOP systems.

348

IEEE TRANSACTIONS ON BIOMEDICAL CIRCUITS AND SYSTEMS, VOL. 4, NO. 6, DECEMBER 2010

IV. CONCLUSION The system presented in this paper provides an active IOP monitoring system that operates over a 24-h period and records pressure data every 5 min. A MEMS sensor captures the pressure data and a custom-designed ASIC processes and digitizes these measurements. These data can either be transmitted out directly to an external device or stored into on-chip memory (FeRAM). If on-chip memory storage is utilized for independent operation, at the end of the day, an external device is brought nearby to download the pressure data and recharge the implant. The implant has onboard power storage as well as wireless recharging and telemetry capabilities. This work describes the design and assembly of a monitoring device that can detect elevated levels of IOP for the diagnosis and treatment of glaucoma.

ACKNOWLEDGMENT The authors would like to thank Texas Instruments for their fabrications services, and microFab Bremen for their supply of MEMS sensors.

REFERENCES
[1] H. A. Quigley and A. T. Broman, The number of people with glaucoma worldwide in 2010 and 2020, Brit. J Ophthalmol., vol. 90, pp. 262267, 2006. [2] R. C. N. Sampaolesi, C. A. De Carvalho, and R. Reca, Diurnal variation of intraocular pressure in healthy, suspected and glaucomatous eyes, Bibl Ophthalmol., vol. 74, pp. 123, 1968. [3] J. T. Wilensky, Diurnal variations in intraocular pressure, Trans. Amer. Ophthalmol. Soc., vol. 89, pp. 757790, 1991. [4] M. Leonardi, P. Leuenberger, D. Bertrand, A. Bertsch, and P. Renaud, First steps toward noninvasive intraocular pressure monitoring with a sensing contact lens 2004, vol. 45, pp. 31133117. [5] M. Leonardi, P. Leuenberger, D. Bertrand, A. Bertsch, and P. Renaud, A soft contact lens with a MEMS strain gage embedded for intraocular pressure monitoring, in Proc. 12th Int. Conf. Transducers, Solid-State Sensors, Actuators and Microsystems, 2003, vol. 2, pp. 10431046. [6] M. Leonardi, P. Leuenberger, D. Bertrand, A. Bertsch, and P. Renaud, First Steps Toward Noninvas. Intraocul. Press. Monitor. Sens. Contact Lens, vol. 45, pp. 31133117, 2004. [7] M. Leonardi, E. M. Pitchon, A. Bertsch, P. Renaud, and A. Mermoud, Wireless contact lens sensor for intraocular pressure monitoring: Assessment on enucleated pig eyes, Acta Ophthalmol., 2008. [8] C. Po-Jui, D. C. Rodger, S. Saati, M. S. Humayun, and T. Yu-Chong, Microfabricated implantable parylene-based wireless passive intraocular pressure sensors, J. Microelectromechan. Syst.,, vol. 17, pp. 13421351, 2008. [9] G. V. R. Puers, D. De Bruyker, R. Puers, and G. Vandevoorde, Electrodeposited copper inductors for intraocular pressure telemetry, J. Micromechan. Microeng., vol. 10, pp. 124129, 2000. [10] J.-B. Y. a. E. Y. E.-C. Park, Hermetically sealed inductor-capacitor (LC) resonator for remote pressure monitoring, Jpn. J. Appl. Phys., vol. 37, pp. 71247128, 1998. [11] A. DeHennis and K. D. Wise, A double-sided single-chip wireless pressure sensor, in Proc. 15th IEEE Int. Conf. Micro Electro Mechanical Systems, 2002, pp. 252255. [12] S. Lizon-Martinez, R. Giannetti, J. L. Rodriguez-Marrero, and B. Tellini, Design of a system for continuous intraocular pressure monitoring, IEEE Trans. Instrum. Meas.,, vol. 54, no. 4, pp. 15341540, Aug. 2005. [13] K. Van Schuylenbergh and R. Puers, Passive telemetry by harmonics detection, in Proc. IEEE 18th Annu. Int. Conf. Engineering in Medicine and Biology Soc. Bridging Disciplines for Biomedicine, 1996, vol. 1, pp. 299300.

[14] L. R. Y. Backlund, B. Hok, and B. Svedbergh, Passive silicone transensor intended for biomedical, remote pressure monitoring, Sens. Actuators, A, pp. 5861, 1990. [15] Y. B. L. Rosengren, T. Sjostrom, B. Hok, and B. Svedbergh, A system for wireless intra-ocular pressure measurements using a silicon micromachined sensor, J. Micromech. Microeng., vol. 2, pp. 202204, 1992. [16] K. Stangel, S. Kolnsberg, D. Hammerschmidt, B. J. Hosticka, H. K. Trieu, and W. Mokwa, A programmable intraocular CMOS pressure sensor system implant, IEEE J. Solid-State Circuits, vol. 36, no. 7, pp. 10941100, Jul. 2001. [17] T. Eggers, J. Draeger, K. Hille, C. Marschner, P. Stegmaier, J. Binder, and R. Laur, Wireless intra-ocular pressure monitoring system integrated into an articial lens, in Proc. 1st Annu. Int. Conf. Microtechnologies in Medicine and Biology, 2000, pp. 466469. [18] E. Y. Chow, S. Chakraborty, W. J. Chappell, and P. P. Irazoqui, Mixed-signal integrated circuits for self-contained sub-cubic millimeter biomedical implants, presented at the Int. Solid State Circuits Conf., San Francisco, CA, 2010. [19] G. W. den Besten and B. Nauta, Embedded 5 V-to-3.3 V voltage regulator for supplying digital ICs in 3.3 V CMOS technology, IEEE J. Solid-State Circuits, vol. 33, no. 7, pp. 956962, Jul. 1998. [20] T. Furuyama, Y. Watanabe, T. Ohsawa, and S. Watanabe, A new on-chip converter for submicrometer high-density DRAMs, IEEE J. Solid-State Circuits, vol. SSC-22, no. 3, pp. 437441, Jun. 1987. [21] A. M. Sodagar and K. Naja, A wide-range supply-independent CMOS voltage reference for telemetry-powering applications, in Proc. 9th Int. Conf. Electronics, Circuits and Systems, 2002, vol. 1, pp. 401404. [22] L. Yu-Shiang, D. Sylvester, and D. Blaauw, A sub-pW timer using gate leakage for ultra low-power sub-Hz monitoring systems, in Proc. IEEE Custom Integrated Circuits Conf., 2007, pp. 397400. [23] E. M. Philofsky, FRAM-the ultimate memory, in Proc. IEEE 6th Biennial Int. Nonvolatile Memory Technology Conf., 1996, pp. 99104. [24] E. Y. Chow, C.-L. Yang, A. Chlebowski, S. Moon, W. J. Chappell, and P. P. Irazoqui, Implantable wireless telemetry boards for in vivo transocular transmission, IEEE Trans. Microw. Theory Tech., , vol. 56, no. 12, pp. 32003208, Dec. 2008. [25] Rogers Corp., R/Flex 3000 Series liquid crystal polymer circuit materials fabrication guidelines 2005. [Online]. Available: http://www. rogerscorp.com/ [26] K. Jayaraj and B. Farrell, Liquid crystal polymers and their role in electronic packaging, Adv Microelectron., vol. 25, pp. 1518, 1998. [27] X. Wang, J. Engel, and C. Liu, Liquid crystal polymer (LCP) for MEMS: Process and applications, J. Micromech. Microeng., vol. 13, pp. 628633, 2003. [28] C. Chouard and P. Pialoux, Biocompatibility of cochlear implants, Bull. Acad. Natl. Med., vol. 179, pp. 549555, 1995. [29] D. R. Merrill, M. Bikson, and J. Jefferys, Electrical stimulation of excitable tissue: Design of efcacious and safe protocols, J. Neurosci. Meth., vol. 141, pp. 171198, 2005. [30] S. W. Lee, J.-M. Seo, S. Ha, E. T. Kim, H. Chung, and S. J. Kim, Development of microelectrode arrays for articial retinal implants using liquid crystal polymers, Invest. Ophthalmol. Vis. Sci., vol. 50, 2009. [31] CFR, Part 90 Federal Communications Commission, 2007. [32] J.-P. Curty, N. Joehl, F. Krummenacher, and C. Dehollain, A model for -powered rectier analysis and design, IEEE Trans. Circuits Syst. I, Reg. Papers, vol. 52, no. 12, pp. 27712779, Dec. 2005. [33] W. Tomasi, Electronic Communication Systems, 4th ed. Upper Saddle River, NJ: Prentice-Hall, 2001. [34] International Technology Roadmap for Semiconductors, ITRS 2008 update. 2008. [35] N. Soman and R. Banerjee, Articial vitreous replacements, BioMed. Mater. Eng., vol. 13, pp. 5974, 2003. [36] IEEE Standard for Safety Levels with Respect to Human Exposure to Radio Frequency Electromagnetic Fields, 3 kHz to 300 GHz, IEEE Std C95.1, IEEE Int. Comm. Electromagnetic Safety (SCC39), 2005. [37] A. Kotecha, E. T. White, J. M. Shewry, and D. F. Garway-Heath, The relative effects of corneal thickness and age on Goldmann applanation tonometry and dynamic contour tonometry, Brit. J. Ophthalmol., vol. 89, pp. 15721575, 2005. [38] W. Mokwa and U. Schnakenberg, Micro-transponder systems for medical applications, IEEE Trans. Instrum. Meas., vol. 50, no. 6, pp. 15511555, Sep. 2001.

CHOW et al.: A MINIATURE-IMPLANTABLE RF-WIRELESS ACTIVE GLAUCOMA INTRAOCULAR PRESSURE MONITOR

349

Eric Y. Chow (M07) received the B.S. degree in electrical and computer engineering from Cornell University, Ithaca, NY, in 2005 and the M.S. and Ph.D. degrees in electrical and computer engineering from Purdue University, West Lafayette, IN, in 2007 and 2009, respectively. He has also had four internships with Intel Corp. in the digital and analog wireless communications groups. During his graduate studies and Postdoctoral Research Associate position at Purdue University, he developed miniature-implantable devices for the treatment of glaucoma and cardiac disease. Currently, he is a Senior Research Scientist in the Advanced Technology Development group at Cyberonics, Inc., where he is developing application-specic integrated circuits (ASICs), RF communications, and wireless charging capabilities for the next-generation implantable device for the treatment of epilepsy. His interests include RF and low-power ASIC design, antenna design, wireless powering, system integration and the development of miniature-implantable devices, and in vivo experimentation for biomedical applications.

Arthur L. Chlebowski (S09) received the B.S. degree in biomedical engineering from Purdue University, West Lafayette, IN, in 2007, where he is currently pursuing the Ph.D. degree in biomedical engineering. During his undergraduate career, he developed sealing methods for biocompatible packages and is currently working on biocompatibility testing, advanced packaging for biomedical implants, and the study of the correlation between glaucoma and blindness.

Pedro. P. Irazoqui (M05) received the B.Sc. and M.Sc. degrees in electrical engineering from the University of New Hampshire, Durham, in 1997 and 1999, respectively, and the Ph.D. degree in neuroengineering from the University of California, Los Angeles, in 2003 for work on the design, manufacture, and packaging of implantable integrated circuits for wireless neural recording. Together, with three partners, he helped found and was Vice-President of IC development at Triangle Biosystems, Inc., Research Triangle Park, NC. Currently, he is an Assistant Professor at the Weldon School of Biomedical Engineering at Purdue University, West Lafayette, IN, where his lab is pursuing research into a modular approach to the design of biological implants in general and neural prosthetic devices in particular. These devices are being applied to the clinical treatment of physiological disorders, using miniature, wireless, and implantable systems. Specic research and clinical applications explored in his lab include glaucoma, epilepsy, neural regeneration, and cardiac disease. He has been Associate Editor of the IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING since late 2006. Prof. Irazoqui received the Best Teacher Award from the Weldon School of Biomedical Engineering in 2006, the Early Career Award from the Wallace H. Coulter Foundation in 2007, and the Marion B. Scott Excellence in Teaching Award from Tau Beta Pi in 2008.