Inductively Coupled Telemetry in Spinal Fusion Application Using Capacitive Strain Sensors

Telemetry is based on knowledge of various disciplines like Electronics, Measurement, Control and Communication along with their combination. This fact leads to a need of studying and understanding of these principles before the usage of Telemetry on selected problem solving. Spending time is however many times returned in form of obtained data or knowledge which telemetry system can provide. Usage of telemetry can be found in many areas from military through biomedical to real medical applications. Modern way to create a wireless sensors remotely connected to central system with artificial intelligence provide many new, sometimes unusual ways to get a knowledge about remote objects behaviour. This book is intended to present some new up to date accesses to telemetry problems solving by use of new sensors conceptions, new wireless transfer or communication techniques, data collection or processing techniques as well as several real use case scenarios describing model examples. Most of book chapters deals with many real cases of telemetry issues which can be used as a cookbooks for your own telemetry related problems. following:

that only responds to bending strain is also desirable for spinal fusion purpose. The strain sensor is expected to measure 1000 µε based on an adult of 200 pounds in a corpectomy model under bending with 2 stainless spinal fusion rods (6.4 mm in diameter and 50.8 mm long) implanted (Gibson, 2002). MEMS capacitive sensors using wireless data transmission have been evaluated in many applications such as humidity (DeHennis & Wise, 2005;Harpster et al., 2002;), temperature (DeHennis & Wise, 2005) and pressure sensing devices (Akar et al., 2001;Chatzandroulis et al., 2000;DeHennis & Wise, 2002, 2005Strong et al., 2002). The telemetry approach to monitor strain uses inductively coupled battery-less technology similar to the technology used in Radio Frequency IDentification (RFID) devices (Finkenzeller, 1999). Some examples of the early applications are shown in Table 1. The inductively coupled wireless system with sensing capability needs not only the working passive telemetry circuitry, but also both the sensor interface circuitry and the sensor themselves. A fully integrated implanted sensor system was realized (Chatzandroulis et al., 2000) with a capacitive pressure sensor and an application-specific integrated circuit ( A S I C ) c h i p t h a t c o n t r o l s R F m o d u l a t i o n and converts capacitance variations into frequency variations. Suster et al. developed a wireless strain detection with the transducer coil size of 3-inch coaxial to the interrogating reader (Suster et al., 2005). However, this transducer coil size is not desirable for spinal fusion implant. Research using this technique coupled with MEMS sensors has become widespread in biomedical applications. It is a promising approach for orthopedic implant sensors and the key is a highly sensitive capacitive sensor (Benzel et al., 2002). In the next sections, the highly sensitive MEMS bending strain sensor will be described in great detail followed by the system circuitry and the testing methods.

The MEMS strain sensors
This section focuses on the development and fabrication of the custom bending strain capacitive sensing element needed for the spinal fusion measurement implant (SFMI) applications. This application requires a high bending strain sensitivity with enough nominal capacitance to avoid loss due to parasitic capacitance, compatibility with an inductively powered circuit, and suitable dimensions for system packaging. The sensitive bending strain sensor is expected to be packaged in a housing container that attaches to the diameter spinal fusion rod. The distance between two vertebrae is about 25.4 mm in the lumber region, making the maximum length of the housing limited to approximately 12 mm long. Therefore, it is desirable that the sensor length be less than 10 mm. The housing is installed between two pedicle screws and needs to transfer the bending strain from the rod to the sensor as described in (Aebersold et al. 2007). The curved surface of the rod is compensated with the 2 mm thick plastic housing which conforms to the rod and is trimmed 1 mm down to provide a flat area of 2 mm x 10 mm for the sensor to mount. Certain characteristics were primarily considered when reviewing limited examples of previous parallel plate capacitive strain sensors in the literature. The basic concept of the capacitive strain sensor features a pair of metalized parallel plates with a dielectric gap. The sensing mechanism manifests itself in varying either the area of the plate, the gap between the plates, or the dielectric medium between the plates. A number of parallel plate sensor designs with a variable air gap were analyzed in the early 90's (Procter & Strong 1992). These sensors generally exhibited low nominal capacitance and sensitivity due to the large gap. In an attempt to increase the nominal capacitance in a non-air gap design, it was demonstrated by a sensor with a parallel plate structure and a thick-film dielectric material (Arshak et al., 1994). The dielectric film between the two plates was compressed during bending, thus expanding the film in area and decreasing the thickness from the perspective of the electrodes. These changes in the film geometry lead to a high gauge factor of 75-80 with a 15-25 µm gap based on a uniform model. The capacitive gauge factor is defined by the fractional change in capacitance with respect to strain. This thick-film dielectric produced both capacitive and resistive responses to strain making this approach electrically unique, but undesirable for the SFMI application due to power consumption. In another design, more effort was involved to invoke the change in permittivity of a dielectric material resulting in a gauge factor of 3.5 to 6, with a 150 µm gap (Arshak et al., 2000). This variable permittivity approach exhibits limited sensitivity that showed no dependency on its dimension (the gauge factor is constant and only depends on the "piezocapacitive" effect). This low gauge factor approach would require additional circuitry that is not desired for this implant design.

The bending sensor theory
The mechanism of sensing pure bending on a test substrate is described in two folds: the capacitance and the strain condition imposed on the sensor, as illustrated in Fig. 2.
Assuming the bending sensor is attached to a steal cantilever with length L and thickness R in an elastic bending.
www.intechopen.com The capacitance from two parallel electrode plates is given by where A is the area, D is the distance between two parallel plates, ε 0 is the permittivity and ε r is the dielectric constant of the material between the plates. In order to measure the strain magnitude, a cantilever test substrate is utilized. For strain and capacitance calculations, it is assumed that the dimensions of the cantilever test substrate very large compared to the sensor and that the sensor is firmly affixed to the substrate. For a cantilever beam, the moment of inertia, I, is given by where, W is the beam width and T (or R as shown in Fig. 1) is the beam thickness. For a beam in uniaxial state of stress, the strain at any point on any surface under bending is given by a textbook (Hibbleer, 1997), where σ is the stress on the surface, E is the Young's modulus of the steel bar substrate, M is the bending moment, c is the distance from the neutral axis to the surface, F is the force www.intechopen.com applied at the free end of the beam and d is the sensor location from the free end of the beam. The sensor location on the beam is given by where L is the length of the cantilever substrate, L 4 and L 2 are the longitudinal boundaries that define the bottom beam of the sensor. Fig. 2a shows the sensor location on the bent cantilever test substrate. Fig. 2b is the side view of a bending condition of the sensor design depicted in Fig. 2c, showing the sensor's metal layer coordinates and the widened gap, D 0 +D(x). Figs. 2c also shows the details of the top and bottom electrode while under bending for the designs of interest. The initial sensor capacitance is given by where L 1 marks the beginning of the metal layer on the bottom electrode, L 2 not only represents the boundary of the sensor but also the end of the metal layers on both the bottom and top electrode beams and therefore, L 2 -L 1 =L 0 is the effective electrode length. With various designs, M 1 is a variable that represents the start of the metal layer on the top electrode beam and also ends the trace that connects the electrode to the pad on the bottom beam. Therefore, w 1 (L 2 -M 1 ) represents the area of the overlapping metal plates, w 2 (M 1 -L 1 ) the area of the metal trace, and D 0 the initial spacing between the plates (see Figs. 2b-2c). The first term represents the capacitance of the overlapping metal plates. The second term is the capacitance of the trace between the electrode and the pad. The third term, C p , is the parasitic capacitance of the metal traces between L 1 and L 3 combined with the planar pads between L 4 and L 5 . L 3 is also the pivot point where the gap starts and L 5 is the physical boundary of the top electrode beam. Capacitance calculations for planar pads indicate that the third term is 0.035 pF (Baxter, 1997). In order to estimate sensor sensitivity to strain, the capacitance change caused by an applied strain is calculated using standard beam equations. The sensor metal plate attached to the beam will follow the beam deflection while the initially parallel plate will remain straight under deformation. The deflection of a cantilever beam and the attached sensor metal plate is given by (Hibbleer, 1997), where the v(x), as seen in Fig. 2a, is the vertical displacement at position x on the beam. The initially parallel plate remains straight and its position is represented by a line tangent to the deformed beam at the pivot point of the sensor. The tangent line (see Fig. 2a where θ(x) is the slope at x and b is a constant determined by a boundary condition. The slope is determined from the first derivative of the deflection and given by At the sensor pivot point, L 3 , from Fig. 4b, the deflection of the two metal plates is equal, providing the boundary condition www.intechopen.com The constant b from (7) is solved by combining (6), (8) and (9) at point L 3 and becomes ( ) Therefore, the tangent line is expressed as The increased gap (see Fig. 2b) between the two electrode plates is a function in the xdirection and expressed as The capacitance change is determined by calculating the average distance between the two metal plates of the strain sensor. The average displacement, in addition to the initial gap, between main metal layers is expressed as 2 1 10 21 where M 1 is where the sensing portion of metal starts and L 2 where it ends. The capacitance due to the trace has an average displacement of D 2, which is expressed as 2 1 20 11 where the metal stops at L 1. Capacitance, due to beam deformation, C f , is given by Combining (3), (13), (14) and (15), yields Based on the equation above, the bending strain sensor is analytically formulated and to be compared with the fabricated MEMS sensor in the following section. www.intechopen.com

Strain sensor fabrication
The sensor fabrication process is illustrated in Fig. 3. The materials include borosilicate glass (Pyrex Corning 7740, 500 µm thick) and silicon wafers (p-type, (100), 1-10 ohm-cm, double side polished, 310 µm thick). Fabrication began with clean glass and silicon substrates as shown in Figs. 3a and 3c. An electrode, traces, and a pair of contact pads were patterned onto the glass substrate by sputtering 0.02 µm chromium for adhesion layer followed by 0.2 µm of gold. The metal trace leading to the bonding area makes electrical contact with the silicon side electrode after anodic bonding. Wet etching was used to pattern the metal (Fig. 3b).The silicon wafer was wet oxidized (Fig. 3d), patterned using photolithography and etched with buffered oxide etch (BOE) solution to form an oxide mask for silicon surface machining. The wafer was etched using potassium hydroxide (KOH) at 85°C (approximately 0.7 µm / minute) to form recessed features and created the initial gap spacing. The etching mask was removed using BOE leaving two www.intechopen.com silicon islands, which function as anchor platforms for the anodic bonding interface, as seen in Fig. 3e. An electrode and trace were then sputtered and patterned onto the silicon using the previously described metallization process. The small contact area on the raised anchor connected the pad on the glass plate with the electrode on the silicon plate via the traces, as seen in Fig. 3f. The glass and silicon wafers were stacked with the metal surfaces facing each other and visually aligned using a mask aligner. Methanol was used to temporarily maintain alignment. The substrates were anodically bonded at 450 o C on a grounded hotplate using a pointed probe to selectively place a -1000 V source on the glass, as shown in Fig. 3g. A gap is created between the electrodes on glass and silicon. This technique of selectively applying the electric field and bonding pressure prevented the recessed spaces from bonding to each other due to thermally induced warpage and electrostatic attraction. An automated dicing saw equipped with a 250 µm thick diamond blade was used to separate the individual sensor die from the bonded wafers. The silicon substrate was diced nearly through at the area above the contact pads. This was accomplished by limiting the depth of the cut and using the dicing alignment marks previously patterned on the silicon. Cuts to individually remove the sensors were similarly made from the silicon and glass substrate leaving approximately 30 µm of each substrate's depth (Fig. 3g). Care was taken to avoid chipping and prevent debris from filling the sensor gap. The sensors were separated from the wafer manually by flexing them to break the remaining thin substrate (Fig. 3h). Sensors with less than 3 µm gap have been fabricated, but with unreliable capacitance values and low yield. It is because of the collapsing of the two electrodes during the anodic bonding process. In an effort to maintain high nominal capacitance, preserve sensitivity and promote linearity, a sensor with an electrode area of 2 mm x 4 mm, and a gap of 3 µm was fabricated for final SFMI prototyping. This sensor was tested on a spinal fusion rod with a near-linear response, as shown in Fig. 4. where dC/C is the fractional change of capacitance and ε is the strain. Using a linear fit of the differential capacitance data graphed in Fig 5, the gauge factor was ploted and calculated to be 252 for 0 to 1000 µε. This value is extremly high in comparison to the current literature (Arshak 1994(Arshak , 2000Proctor & Strong 1992). By comparison, piezoresistive gauges typically provide a gauge factor less than 200 (Fraden, 1995) even at the cost of high temperature sensitivity.

The transducer circuit
The transducer circuit is an inductively coupled, load modulated design similar in concept to Radio Frequency IDentification tags used for inventory and security. The 125 kHz magnetic field sourced by the interrogating reader induces a voltage in the LC tank of the implant. The LC tank (1 cm diameter, 600 turns) is resonant at 125 kHz. The AC voltage is then rectified, filtered, and regulated using a low quiescent power regulator. A supply of 3VDC, 28 µA is required to power the oscillator circuit described above and a frequency divider circuit composed of flip-flops. The oscillator frequency is divided to less than 1/20 the carrier frequency so that detection is simplified. The frequency divider also buffers the oscillator signal so that it can drive the gate of a MOSFET placed across the LC tank. The MOSFET acts a load that modulates the 125 kHz tank output at the divider output frequency. A diagram that shows the functionality can be seen in Fig. 6(a). As it is functioned as an oscillator in Fig. 6(a), a capacitance to frequency (c-f) converter is used to convert the strain signal to a signal that can be transfered wirelessly. The c-f circuit is comprised of a pair of CMOS inverters (Lancaster, 1997) and can be seen in detail in Fig.  6(b). The oscillator produces a periodical voltage due to the charging and discharging of the RC across a threshold of an inverter input. The frequency of the RC oscillator is expressed as F=2.3R t C, where C is the variable capacitor, or a capacitive strain sensor, and R t is the matched resistance of the oscillator. The resistor R I is to unload the RC network from clamping effect of protection diodes in the inverter. It will also result in a nearly square duty cycle and make the frequency less dependent of power supply variations. R I is normally set about 10 times as higher as R t to minimize the effect of the protection diodes. The oscillator oscillates empirically at 20 kHz with 20 pF capacitor and 100 kOhm resistor. www.intechopen.com

The interrogating reader
The interogating reader operating on 12 VDC, 175 mA provides the 125 kHz magnetic field for the implant, as illustrate in Fig. 7. The reader antenna is 24 cm in diameter and is tuned to 125 kHz. An EM Microelectronic (Marin, Switzerland), EM4095 IC contains an on-chip oscillator, antenna driver, and a demodulation circuit. The output of the demodulator is measured using an Agilent 53131A counter and logged with a computer based data acquisition system. Fig. 7. The block diagram of the power reader.

Detection region
In the region of detection, see Fig. 8, the implant receives enough power to operate from the magnetic field sourced by the reader. There is no degradation of strain sensing performance if sufficient DC power is available from the regulator. However at the far end of the region, planar and axial alignment becomes very important. With distance from the reader, inductive coupling is reduced thereby reducing the AC voltage across the LC tank and thus the modulated signal amplitude. The data signal also relies on the same low coupling between the implant and the reader. It is necessary to have a sensitive reader to detect the implant at the far end of the region. Fig. 8. The detection region is within the cone shape.

The testing methods
The corpectomy model was used to evaluate the strain measurement system prior to spine testing. This model has similar bending to a four-point bending model therefore the rod strain does not change significantly along the length of the rod. A single metal foil strain gauge reference was attached to the rod adjacent to the housing for comparison to the transducer system. Comparing the metal foil reference gauge to the spinal fusion sensor placed in close proximity is justified. Figure 9 shows the sensor assembly on a housing without a container cover. Fig. 9. The transducer sensor system on a housing that attaches to a stainless rod before a hermetically sealed container is assembled. www.intechopen.com The strain measurement system was tested using a corpectomy model designed as a simplified mechanical analog of a spine section and then will be tested using a human spine. Figure 10 shows a free body diagram of the forces and bending moments applied to the spinal rods through the pedicle screws due to loading of the hard plastic blocks. Note that when the spinal fusion rods are fastened to the fixtures, often the initial strain is introduced and recalibrated. On the first test, the corpectomy apparatus was assembled inside a clear acrylic water tank without water on the MTS machine for application of the load; the sensor system was oriented facing out to be detected. The MTS machine's dynamic motion only changes 3 mm in the detection distance between the interrogating reader and the strain sensor. The read range was limited to 10 cm due to the reader design used and the sensor coil design constraints. A more optimal reader would increase the range. For other applications where a larger coil diameter is acceptable the range would be increased as well. The data acquisition is obtained with LabView software, an interfacing program designed to transfer and record live data between instruments and computer by National Instrument. The strain information is recorded by the commecial metal foil strain gauge through a strain www.intechopen.com indicator (Measurement group model P-3500). The frequency output from the wireless strain sensor system measurement is obtained through the digital universal counter (Agilent 53131A). In normal use, the response time for the strain indicator is 0.5 second, and the universal counter about 0.2 second. However, with the Labview interface the response time for the universal counter changes to 2.3 seconds but does not change for the strain indicator. The lag in time in the acquisition of the frequency data from the sensor system will cause false information when recording a drastic dynamic motion. In order to have a consistent and corresponding readings for the referred strain and the frequency output, a delay or pause in the live measurement is needed. Therefore, the MTS machine is programmed to pause 5 seconds at every increment of load for the frequency output to stabilize. Both the strain and the frequency data then are taken at the same time frame as the Material Test Machine. The result of the telemetry strain sensor detection in Fig. 11 shows good linearity with R 2 = 0.99. The gauge factor is calculated to be 249 up to 1000 micro-strain, with R 2 =0.96 as see in Fig. 12. Similar gauge factor of 252 in the capacitance mode was measured in section 2. The frequency mode also exhibits high gauge factor without any amplification. Fig. 11. The demodulate frequency responses from the sensor transducer shows a linear response due to bending test of the corpectomy model. www.intechopen.com

Water tank test
For the next application in water tank test, the sensor is expected to be in a fluidic environment. To protect from the influence of water or water vapor, the system was hermetically sealed in a container made of nylon and painted heavily with silicone and polyethylene. The container protected the device during corpectomy testing where it is submerged in a water tank used to simulate the body and during spine testing where it is in contact with tissue. Caution has been taken for the temperature difference between water and ambient. A temperature dependent test on the sensor shows that the frequency drift from 1752 to 1778 Hz from 0 o C to 22.2 o C and vice versa with no strain at the rate of about 1.8 Hz/ o C. The interrgating reader was not moved at about 10 cm from the sensor system, through water, glass and air. The difference of the readings before the introduction of water and after was unnoticeable. The corpectomy model in water test was successful in showing the repeatability for the cyclic loads. It aslo suggestes that measurements were possible in conditions present in-vivo. The tests of the corpectomy model in water tank were successful in showing that measurements were possible in conditions present in-vivo.

Excise spine test
A discectomy was performed on an excised spine from a cadaver and was constrained and loaded in a MTS system to simulate a 113.4 kg (250 lb) patient. The excised spine was potted using a liquid lead/bismuth alloy (Cerro Products, Bellefont, PA) and attached to a MTS Bionix mechanical testing system. A custom-built test apparatus was used to apply anatomic loads to the spine. The intact spine was tested by applying loads ranging from 100 to 200 N in an increasing and decrease manner with a 5-second period of pause at every increment to accommodate the detection speed of the system. The disc between L3 and L4 was surgically removed to simulate an unstable spine prior to fusion surgery, and the test sequence repeated. The comparison of results before and after the disk is removed are shown in Fig. 13. With the same loading steps on the spine from MTS, the spinal fusion rods experienced less strain increment in the spine with the disk than that in the spine with the disk removed.

www.intechopen.com
From the wireless transmision data along with the referred metal foil strain gauge, it is suggested that the spinal fusion rods showed that roughly one third of the load were shared by the intact spine after the spine is fused. The telemetry system clearly shows the rigidity of the intact spine.

Conclusion
A telemetry system using a capacitive strain sensor has been developed for the detection of spinal fusion monitoring. The strain sensor was made using MEMS process with high sensitivity and reduced size that serves the purpose for strain detection on the spinal fusion rod. The cantilever structure of the sensor is composed of two parallel plates, galss and silicon, respectively, with a narrow gap D 0 (3 μm) and a conjoint end. The bending strain sensor has the characteristics of high nominal capacitance (20 pF), high sensitivity, and compact dimensions (2 mm x 7 mm x 0.8 mm). It utilizes a variable gap configuration comprised of silicon and glass beams that are bonded at one end and open at the opposing end. This type of structure has been tested to withstand a strain range of 0 to 1000 µε. The inductive link between the implant circuit and the reader was sufficient for supplying power to the implant circuit and extracting data at 10 cm distance. A specific sensor has a linear gauge factor of 252 in the capacitive domain and 249 in the frequency domain. Measurements made through air and water with a corpectomy model produced a linear response consistent with a metal foil reference gauge. The strain measurement system was also tested with the corpectomy model designed as a simplified mechanical analog of a spine section and was then tested using a human spine. For the biomechanics application, the sensor is expected to be in a fluidic environment. The tests of the corpectomy model placed in water tank were successful in showing that measurements were possible in conditions present in-vivo. The read range was limited to 10 cm due to the reader design used and the sensor coil design constraints. Finally, a test performed using a human spine showed the wireless implant detected strain roughly one third of the load were shared by the intact spine after the spine is fused. Telemetry is based on knowledge of various disciplines like Electronics, Measurement, Control and Communication along with their combination. This fact leads to a need of studying and understanding of these principles before the usage of Telemetry on selected problem solving. Spending time is however many times returned in form of obtained data or knowledge which telemetry system can provide. Usage of telemetry can be found in many areas from military through biomedical to real medical applications. Modern way to create a wireless sensors remotely connected to central system with artificial intelligence provide many new, sometimes unusual ways to get a knowledge about remote objects behaviour. This book is intended to present some new up to date accesses to telemetry problems solving by use of new sensors conceptions, new wireless transfer or communication techniques, data collection or processing techniques as well as several real use case scenarios describing model examples. Most of book chapters deals with many real cases of telemetry issues which can be used as a cookbooks for your own telemetry related problems.