Impedance Spectroscopy to Monitor Fracture Healing* Monica C. Lin, Safa T. Herfat, Chelsea S. Bahney, Meir Marmor, and Michel M. Maharbiz, Member, IEEE
Abstract— An estimated 7.9 million fracture injuries occur each year in the United States, of which a substantial fraction result in delayed or non-union. Current methods of monitoring fracture healing include taking x-rays and making clinical observations. However, x-ray confirmation of bone healing typically lags behind biologic healing, and physician assessment of healing is fraught with subjectivity. No standardized methods exist to assess the extent of healing that has taken place in a fracture. Without such knowledge, interventions to aid healing and prevent fracture non-union are often delayed, leading to increased morbidity and suffering to patients. We are developing an objective measurement tool that utilizes electrical impedance spectroscopy to distinguish between the various types of tissue present during the different stages of fracture healing. Preliminary measurements of cadaveric tissues reveal adequate spread in impedance measurements and differences in frequency response among different tissue types. Electrodes implanted in a simulated fracture created in an ex vivo cadaver model yield promising results for our system’s ability to differentiate between the stages of fracture healing.
I. INTRODUCTION Of the estimated 7.9 million fracture injuries that occur each year in the United States alone, 10% will fail to heal appropriately and result in delayed or non-union [1], with incidence of non-union rising to 46% when the fractures occur in conjunction with vascular injury [2]. Treatment of fractures costs the U.S. healthcare system $45 billion per year. In particular, multiple reoperations are often necessary to treat non-unions, and 51% of fracture patients do not return to work in 6 months [3]. This causes substantial disability to patients and represents a significant burden on the healthcare system. Determining when a fracture is healed is crucial to making correct clinical decisions for patients, but there are currently no standardized methods of assessing fracture union. Current available tools for assessing fracture healing include radiographic methods, serologic markers, and clinical *Research supported by the National Science Foundation under grant no. EFRI-1240380 and by the Center for Disruptive Musculoskeletal Innovations (IIP-1361975). M. C. Lin is with the Bioengineering Department at the University of California – Berkeley, CA 94720 and the University of California – San Francisco, CA 94158 USA (phone: 408-799-0730; email: [email protected]). S. T. Herfat, C. S. Bahney, and M. Marmor are with the Department of Orthopaedic Surgery at the University of California – San Francisco / San Francisco General Hospital Orthopaedic Trauma Institute, San Francisco, CA 94110 USA (emails: [email protected], [email protected], [email protected]). M. M. Maharbiz is with the Electrical Engineering & Computer Science Department and the Bioengineering Department at the University of California – Berkeley, CA 94720 USA (email: [email protected]).
978-1-4244-9270-1/15/$31.00 ©2015 IEEE
evaluation [4]. However, poor accuracy, unreliability, need for high doses of radiation, large expense, and/or subjectivity limit their clinical utility [4-7]. Clinically, fractures heal primarily through endochondral ossification, in which bone forms indirectly from a cartilage template. Healing occurs through four overlapping phases of repair, beginning with an inflammatory phase, followed by chondrogenesis of mesenchymal progenitors to form the early cartilage callus that matures into a hard callus of cancellous bone, and finally remodeling into healthy cortical bone. [1, 8, 9] These clearly defined stages of healing can be well characterized histologically, but they are not detectable by standard radiographic techniques. Electrical impedance spectroscopy (EIS) measures the dielectric properties of tissue as a function of frequency, and has been used for decades to characterize biological tissues [10-13] such as bone [14, 15]. Here, we apply this technique to monitoring fracture healing. First, we optimize a system to measure impedance in the fracture environment, then validate our system by measuring tissues present in healing fractures. We perform experiments to assess feasibility of our system to work within the current fracture treatment scheme, and present measurement data from a simulated fracture model created in a cadaver. II. SYSTEM OVERVIEW Previous work has been done in our group to study skin health using impedance spectroscopy [16], and we are now applying this technique to study fracture healing. The system consists of electrodes to contact the tissue of interest, which are routed to control hardware that interfaces with an LCR meter and laptop, allowing for automatic collection of impedance measurements across a range of frequencies. We used a Keysight Technologies E4980AL-100 Precision LCR meter with a 100 mV voltage sine wave output signal at frequencies of 20 Hz to 1 MHz to measure impedance magnitude and phase. The control hardware currently runs off of four AA batteries, and a set of measurements between two electrodes that incorporates a frequency sweep takes a little over one minute. To optimize our system for use in the fracture healing environment, we are designing specialized electrodes and have run experiments to understand how impedance data differ between tissues present in healing fractures, which will be detailed in this paper. Sensors will be placed at or near a fracture injury to gather as much information as possible about the tissues in the fracture gap. Multiple electrodes can be arrayed to probe the area at multiple locations to both spatially and temporally resolve the healing process.
5138
steel) were used in this study. Measurements were taken in intact cortical bone.
Figure 1. a) Fracture model depicting sensors integrated within the treatment scheme and how the signal can travel within the fracture environment. b) Experimental setup with a bone plate secured by two bone screws in between electrodes.
III. FRACTURE MODEL For clinical application, these impedance sensors will ultimately be integrated into the existing management techniques of surgically-treated fractures. Target cases would include fractures stabilized by either internal or external fixation. Internally stabilized fractures involve a metal bone plate in which bone screws are drilled into healthy bone on either side of the fracture site to secure a plate in place and stabilize the fracture. Our sensors would be designed to mimic a bone screw and then drilled into the bone tissue at or immediately flanking the fracture site to measure impedance in or across the injury. Fig. 1 shows how the sensors fit within this treatment regime and how the signal will travel between the electrodes. External fixation is similar, except the metal plate is left outside the body and only the bone screws pierce the skin and are fixed to the uninjured areas of bone.
With the electrodes spaced 12mm apart, impedance measurements of the bone with and without a plate are virtually indistinguishable, signifying that there is something at the interface between the bone and the bone plate that makes it difficult for the current to pass from the bone to the plate. The bone plate can never be fully flush against the bone since bone is not entirely flat, so this poor electrical contact may prevent signal from traveling between these two structures. In other words, the impedance at the bone-plate interface, represented by Zb-p and Z’b-p, is high. Instead, the current should preferentially travel directly between the electrodes on the ends of the sensor screws, passing only through bone tissue and tissue filling the fracture gap. As another test, the electrodes were placed 36mm apart and bone was measured first without screws, and then with two bone screws inserted in between the electrodes. In this experiment, the impedance magnitude dropped after the screws were inserted. This result indicates that the bone screw is making good electrical contact to the bone, so current could travel through the bone screws and create a short circuit through the plate. This will only be an issue if the bone screws are placed too close to the sensors at the fracture site. However, in most clinical situations, bone screws are typically placed far from the site of injury to ensure they are secured in strong, healthy bone so the bone plate is adequately held in place against the fracture. Therefore, this is not a major concern, although this potential issue still needs to be considered in determining best use cases for our device.
Ideally, the signal will travel through biological tissue at and around the fracture site so the impedance measurement will reflect the changing tissues as the fracture heals. However, since there is a metal bone plate pressed up against the bone, there is a possibility that the current will short through this highly conductive path rather than travel through the bone tissue. To understand this, we conducted an experiment to determine how presence of a bone plate and bone screws would affect impedance measurements in a cadaveric model, with results shown in Fig. 2. Two Kirschner wires and a Syntheses 245.16 bone plate (stainless 5139
Figure 2. Bode diagrams of impedance magnitude and phase versus frequency showing the effect of a bone plate and bone screws on impedance measurements. The distance listed in the legend corresponds to the distance between the electrodes.
IV. RESULTS & DISCUSSION
to living tissue, these measurements can still demonstrate differences in impedance between the various tissue types.
A. Impedance Measurements of Fracture Tissues Fracture healing occurs through four stages that are characterized by different tissue types. Stage one, the inflammatory phase, commences immediately after the injury with the formation of a hematoma. The hematoma is formed to stop the bleeding and contain the fracture debris after the bone break. The hematoma is a critical step in initiating healing and the tissue composition begins largely as coagulated blood that is remodeled into a fibrous scaffold. Stage two is characterized by a soft callus that bridges the fracture gap and is primarily composed of cartilage. During stage three, blood vessels invade the cartilage so the tissue becomes calcified and eventually converts into cancellous bone. Finally, the fracture heals fully in stage four by remodeling the cancellous bone into cortical bone that nearly perfectly resembles the original tissue in both form and function. [1,8,9]
The impedance magnitude measurements trend as expected, with readings steadily increasing from stage one through stage four. The largest spread in the data is found at frequencies between 5 and 15 kHz. Importantly, the shapes of these plots as a function of frequency vary amongst the tissues, with the dominant pole shifting to the right (higher frequency) as the tissues progress through the healing phases. The poles and zeros that describe the frequency responses of impedance of these tissues fall out of transfer function fits to their respective Bode plots. Ultimately, a parameter used to distinguish between the different tissue types may be a combination of information gathered from impedance magnitude and phase, as well as from transfer function fits. This will allow us to objectively classify a fracture within one of the four stages of healing, and track the progression of recovery over time.
To best replicate these four phases of fracture healing, impedance measurements were taken of blood, coagulated blood, cartilage, cancellous bone, and cortical bone to validate the ability of our system to distinguish between these various tissue types present in fracture healing using impedance spectroscopy. Gold-plated copper electrodes, 300 um in diameter, were used to measure these tissues, which were extracted from a cadaver. Each set of measurements consisted of 5 impedance readings at each of 18 frequencies spanning 20 Hz to 1 MHz. For each tissue type, 5 sets of measurements were taken, with mean and standard deviation calculated at each frequency and plotted in Fig. 3. Despite the lack of water and other fluids in cadaveric tissue as compared
B. Ex Vivo Simulated Fracture We simulated a fracture ex vivo in a cadaveric tibia to understand our system in the context of a human injury. We first created a complete fracture in the center of a tibia bone extracted from a cadaver, and fixed it in place with an external fixator, resulting in a 5mm fracture gap. The external fixator pins, made of inert stainless steel, were screwed into the bone 28 mm apart and used as the electrodes in this study. Small amounts of cartilage and cancellous bone were individually stuffed into the fracture gap, and 5 sets of impedance measurements were taken for each of 18 frequencies from 20 Hz to 1 MHz. In addition, a
Figure 4. Bode diagrams of impedance magnitude and phase for measurements across a simulated fracture plotted as a function of frequency.
Figure 3. Bode diagrams of impedance magnitude and phase for various fracture tissue types plotted as a function of frequency.
5140
heterogeneous mixture of cartilage and cancellous bone was stuffed into the fracture gap as a comparison. Analysis of the data, shown in Fig. 4, shows clear differences between the impedance across a fracture gap filled with cartilage and one filled with cancellous bone, with the graph for the mixture falling in between the two tissues as expected. Cartilage and cancellous bone placed in a gap created in cortical bone represent stage two and stage three of the fracture healing process, respectively. Since cortical bone is of higher impedance than the other tissues present in the fracture gap, it is critical that impedance measurements taken across the fracture reflect the tissues in the gap and are not completely dominated by the cortical bone around the fracture. Data collected from this simulated fracture indicate that our system can at least distinguish between injuries at stage two versus stage three of the healing process. This will enable tracking of fracture healing over time, and allow physicians to spot when a fracture does not progress through the different stages of healing at the anticipated rate. This will allow for early intervention to prevent high risk fractures from failing to heal in an acceptable time frame.
REFERENCES [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11]
V. CONCLUSION We have developed a system that utilizes impedance spectroscopy to study fractures and has shown promise in its ability detect the different tissue compositions that are expected to be at the fracture site during the progression of bone healing. Experiments have been conducted to ensure that presence of a bone plate and screws in the area of injury for fracture stabilization will not compromise our measurements. We have demonstrated that our system can distinguish between the different tissues present in healing fractures. Measurements taken across a simulated fracture further highlight the functionality of our sensors to classify fractures as being in one of the four stages of healing, which will provide physicians with additional quantitative information to help direct treatment. These experiments establish feasibility of our system to detect differences in electrical properties of the tissue at the fracture site as an injury heals, which can be validated in an in vivo model. Ultimately, understanding where a patient’s injury is within the healing process can diagnose delayed healing at an earlier stage and allow for timely intervention to prevent problem fractures from progressing to non-union.
[12] [13] [14] [15] [16]
ACKNOWLEDGMENT M. C. L. acknowledges Sarah Swisher, Amy Liao, and Kaylee Mann for their contribution to the control board and software development. We acknowledge support from the Berkeley Sensor and Actuator Center, the Swarm Lab at UC Berkeley, and the Orthopaedic Trauma Institute at San Francisco General Hospital. This work was supported by the National Science Foundation under grant no. EFRI-1240380 and by the Center for Disruptive Musculoskeletal Innovations (IIP-1361975). M. C. L. was supported by a National Science Foundation Graduate Research Fellowship. 5141
M. L. Schenker, et al., “Fracture Repair and Bone Grafting,” Orthopaedic Knowledge Update, pp. 15-25, 2014. K. F. Dickson, et al., “Importance of the blood supply in the healing of tibial fractures,” Contemp Orthop, vol. 30, no. 6, pp. 489-493, 1995. American Academy of Orthopaedic Surgeons, “The Burden of Musculoskeletal Diseases in the United States”, Rosemont, IL, 2008. S. Morshed, “Current Options for Determining Fracture Union,” Advances in Medicine, pp. 1-12, 2014. S. P. Whiley, et al., “Evaluating fracture healing using digital X-ray image analysis,” CME, vol. 29, no. 3, pp. 102-106, 2011. V. C. Protopappas, et al., “Ultrasonic Monitoring of Bone Fracture Healing,” Transactions on Ultrasonics, Ferroelectrics, and Frequency Control, vol. 55, no. 6, pp. 1243-1255, 2008. L. E., Claes and C. A. Heigele, “Magnitudes of local stress and strain along bony surfaces predict the course and type of fracture healing,” Journal of Biomechanics, pp. 255-266, 1999. B. McKibbin, et al., “The Biology of Fracture Healing in Long Bones,” Journal of Bone and Joint Surgery, pp. 150-162, 1978. H. M. Frost, et al., “The Biology of Fracture Healing: An Overview for Clinicians, Part I,” Clinical Orthopaedics and Related Research, pp. 283-293, 1989. R. D. Stoy, K. R. Foster, and H. P. Schwan, “Dielectric properties of mammalian tissues from 0.1 to 100 MHz: a summary of recent data,” Phys Med Biol, vol. 27, no. 4, pp. 501-513, 1982. H. Jongschaap, et al., “Electrical impedance tomography: a review of current literature,” European Journal of Radiology, vol. 18, pp. 165174, 1994. D. A. Dean, et al., “Electrical Impedance Spectroscopy Study of Biological Tissues,” J Electrostat, vol. 66, no. 3-4, pp. 165-177, 2008. S. Gabriel, R. W. Lau, and C. Gabriel, “The dielectric properties of biological tissues: II,” Phys Med Biol, vol. 41, pp. 2251-2269, 1996. I. V. Ciuchi, et al., “Impedance spectroscopy characterization of bone tissues,” Journal of Advanced Research in Physics, vol. 1, no. 1, pp. 15, 2010. S. Schaur, B. Jakoby, and G. Kroureif, “Position-dependent characterization of bone tissue with electrical impedance spectroscopy,” IEEE, pp. 1-4, 2012. S. L. Swisher, et al., “Impedance sensing device enables early detection of pressure ulcers in vivo,” Nature Communications, vol. 6, 2015.
Abstract— An estimated 7.9 million fracture injuries occur each year in the United States, of which a substantial fraction result in delayed or non-union. Current methods of monitoring fracture healing include taking x-rays and making clinical observations. However, x-ray confirmation of bone healing typically lags behind biologic healing, and physician assessment of healing is fraught with subjectivity. No standardized methods exist to assess the extent of healing that has taken place in a fracture. Without such knowledge, interventions to aid healing and prevent fracture non-union are often delayed, leading to increased morbidity and suffering to patients. We are developing an objective measurement tool that utilizes electrical impedance spectroscopy to distinguish between the various types of tissue present during the different stages of fracture healing. Preliminary measurements of cadaveric tissues reveal adequate spread in impedance measurements and differences in frequency response among different tissue types. Electrodes implanted in a simulated fracture created in an ex vivo cadaver model yield promising results for our system’s ability to differentiate between the stages of fracture healing.
I. INTRODUCTION Of the estimated 7.9 million fracture injuries that occur each year in the United States alone, 10% will fail to heal appropriately and result in delayed or non-union [1], with incidence of non-union rising to 46% when the fractures occur in conjunction with vascular injury [2]. Treatment of fractures costs the U.S. healthcare system $45 billion per year. In particular, multiple reoperations are often necessary to treat non-unions, and 51% of fracture patients do not return to work in 6 months [3]. This causes substantial disability to patients and represents a significant burden on the healthcare system. Determining when a fracture is healed is crucial to making correct clinical decisions for patients, but there are currently no standardized methods of assessing fracture union. Current available tools for assessing fracture healing include radiographic methods, serologic markers, and clinical *Research supported by the National Science Foundation under grant no. EFRI-1240380 and by the Center for Disruptive Musculoskeletal Innovations (IIP-1361975). M. C. Lin is with the Bioengineering Department at the University of California – Berkeley, CA 94720 and the University of California – San Francisco, CA 94158 USA (phone: 408-799-0730; email: [email protected]). S. T. Herfat, C. S. Bahney, and M. Marmor are with the Department of Orthopaedic Surgery at the University of California – San Francisco / San Francisco General Hospital Orthopaedic Trauma Institute, San Francisco, CA 94110 USA (emails: [email protected], [email protected], [email protected]). M. M. Maharbiz is with the Electrical Engineering & Computer Science Department and the Bioengineering Department at the University of California – Berkeley, CA 94720 USA (email: [email protected]).
978-1-4244-9270-1/15/$31.00 ©2015 IEEE
evaluation [4]. However, poor accuracy, unreliability, need for high doses of radiation, large expense, and/or subjectivity limit their clinical utility [4-7]. Clinically, fractures heal primarily through endochondral ossification, in which bone forms indirectly from a cartilage template. Healing occurs through four overlapping phases of repair, beginning with an inflammatory phase, followed by chondrogenesis of mesenchymal progenitors to form the early cartilage callus that matures into a hard callus of cancellous bone, and finally remodeling into healthy cortical bone. [1, 8, 9] These clearly defined stages of healing can be well characterized histologically, but they are not detectable by standard radiographic techniques. Electrical impedance spectroscopy (EIS) measures the dielectric properties of tissue as a function of frequency, and has been used for decades to characterize biological tissues [10-13] such as bone [14, 15]. Here, we apply this technique to monitoring fracture healing. First, we optimize a system to measure impedance in the fracture environment, then validate our system by measuring tissues present in healing fractures. We perform experiments to assess feasibility of our system to work within the current fracture treatment scheme, and present measurement data from a simulated fracture model created in a cadaver. II. SYSTEM OVERVIEW Previous work has been done in our group to study skin health using impedance spectroscopy [16], and we are now applying this technique to study fracture healing. The system consists of electrodes to contact the tissue of interest, which are routed to control hardware that interfaces with an LCR meter and laptop, allowing for automatic collection of impedance measurements across a range of frequencies. We used a Keysight Technologies E4980AL-100 Precision LCR meter with a 100 mV voltage sine wave output signal at frequencies of 20 Hz to 1 MHz to measure impedance magnitude and phase. The control hardware currently runs off of four AA batteries, and a set of measurements between two electrodes that incorporates a frequency sweep takes a little over one minute. To optimize our system for use in the fracture healing environment, we are designing specialized electrodes and have run experiments to understand how impedance data differ between tissues present in healing fractures, which will be detailed in this paper. Sensors will be placed at or near a fracture injury to gather as much information as possible about the tissues in the fracture gap. Multiple electrodes can be arrayed to probe the area at multiple locations to both spatially and temporally resolve the healing process.
5138
steel) were used in this study. Measurements were taken in intact cortical bone.
Figure 1. a) Fracture model depicting sensors integrated within the treatment scheme and how the signal can travel within the fracture environment. b) Experimental setup with a bone plate secured by two bone screws in between electrodes.
III. FRACTURE MODEL For clinical application, these impedance sensors will ultimately be integrated into the existing management techniques of surgically-treated fractures. Target cases would include fractures stabilized by either internal or external fixation. Internally stabilized fractures involve a metal bone plate in which bone screws are drilled into healthy bone on either side of the fracture site to secure a plate in place and stabilize the fracture. Our sensors would be designed to mimic a bone screw and then drilled into the bone tissue at or immediately flanking the fracture site to measure impedance in or across the injury. Fig. 1 shows how the sensors fit within this treatment regime and how the signal will travel between the electrodes. External fixation is similar, except the metal plate is left outside the body and only the bone screws pierce the skin and are fixed to the uninjured areas of bone.
With the electrodes spaced 12mm apart, impedance measurements of the bone with and without a plate are virtually indistinguishable, signifying that there is something at the interface between the bone and the bone plate that makes it difficult for the current to pass from the bone to the plate. The bone plate can never be fully flush against the bone since bone is not entirely flat, so this poor electrical contact may prevent signal from traveling between these two structures. In other words, the impedance at the bone-plate interface, represented by Zb-p and Z’b-p, is high. Instead, the current should preferentially travel directly between the electrodes on the ends of the sensor screws, passing only through bone tissue and tissue filling the fracture gap. As another test, the electrodes were placed 36mm apart and bone was measured first without screws, and then with two bone screws inserted in between the electrodes. In this experiment, the impedance magnitude dropped after the screws were inserted. This result indicates that the bone screw is making good electrical contact to the bone, so current could travel through the bone screws and create a short circuit through the plate. This will only be an issue if the bone screws are placed too close to the sensors at the fracture site. However, in most clinical situations, bone screws are typically placed far from the site of injury to ensure they are secured in strong, healthy bone so the bone plate is adequately held in place against the fracture. Therefore, this is not a major concern, although this potential issue still needs to be considered in determining best use cases for our device.
Ideally, the signal will travel through biological tissue at and around the fracture site so the impedance measurement will reflect the changing tissues as the fracture heals. However, since there is a metal bone plate pressed up against the bone, there is a possibility that the current will short through this highly conductive path rather than travel through the bone tissue. To understand this, we conducted an experiment to determine how presence of a bone plate and bone screws would affect impedance measurements in a cadaveric model, with results shown in Fig. 2. Two Kirschner wires and a Syntheses 245.16 bone plate (stainless 5139
Figure 2. Bode diagrams of impedance magnitude and phase versus frequency showing the effect of a bone plate and bone screws on impedance measurements. The distance listed in the legend corresponds to the distance between the electrodes.
IV. RESULTS & DISCUSSION
to living tissue, these measurements can still demonstrate differences in impedance between the various tissue types.
A. Impedance Measurements of Fracture Tissues Fracture healing occurs through four stages that are characterized by different tissue types. Stage one, the inflammatory phase, commences immediately after the injury with the formation of a hematoma. The hematoma is formed to stop the bleeding and contain the fracture debris after the bone break. The hematoma is a critical step in initiating healing and the tissue composition begins largely as coagulated blood that is remodeled into a fibrous scaffold. Stage two is characterized by a soft callus that bridges the fracture gap and is primarily composed of cartilage. During stage three, blood vessels invade the cartilage so the tissue becomes calcified and eventually converts into cancellous bone. Finally, the fracture heals fully in stage four by remodeling the cancellous bone into cortical bone that nearly perfectly resembles the original tissue in both form and function. [1,8,9]
The impedance magnitude measurements trend as expected, with readings steadily increasing from stage one through stage four. The largest spread in the data is found at frequencies between 5 and 15 kHz. Importantly, the shapes of these plots as a function of frequency vary amongst the tissues, with the dominant pole shifting to the right (higher frequency) as the tissues progress through the healing phases. The poles and zeros that describe the frequency responses of impedance of these tissues fall out of transfer function fits to their respective Bode plots. Ultimately, a parameter used to distinguish between the different tissue types may be a combination of information gathered from impedance magnitude and phase, as well as from transfer function fits. This will allow us to objectively classify a fracture within one of the four stages of healing, and track the progression of recovery over time.
To best replicate these four phases of fracture healing, impedance measurements were taken of blood, coagulated blood, cartilage, cancellous bone, and cortical bone to validate the ability of our system to distinguish between these various tissue types present in fracture healing using impedance spectroscopy. Gold-plated copper electrodes, 300 um in diameter, were used to measure these tissues, which were extracted from a cadaver. Each set of measurements consisted of 5 impedance readings at each of 18 frequencies spanning 20 Hz to 1 MHz. For each tissue type, 5 sets of measurements were taken, with mean and standard deviation calculated at each frequency and plotted in Fig. 3. Despite the lack of water and other fluids in cadaveric tissue as compared
B. Ex Vivo Simulated Fracture We simulated a fracture ex vivo in a cadaveric tibia to understand our system in the context of a human injury. We first created a complete fracture in the center of a tibia bone extracted from a cadaver, and fixed it in place with an external fixator, resulting in a 5mm fracture gap. The external fixator pins, made of inert stainless steel, were screwed into the bone 28 mm apart and used as the electrodes in this study. Small amounts of cartilage and cancellous bone were individually stuffed into the fracture gap, and 5 sets of impedance measurements were taken for each of 18 frequencies from 20 Hz to 1 MHz. In addition, a
Figure 4. Bode diagrams of impedance magnitude and phase for measurements across a simulated fracture plotted as a function of frequency.
Figure 3. Bode diagrams of impedance magnitude and phase for various fracture tissue types plotted as a function of frequency.
5140
heterogeneous mixture of cartilage and cancellous bone was stuffed into the fracture gap as a comparison. Analysis of the data, shown in Fig. 4, shows clear differences between the impedance across a fracture gap filled with cartilage and one filled with cancellous bone, with the graph for the mixture falling in between the two tissues as expected. Cartilage and cancellous bone placed in a gap created in cortical bone represent stage two and stage three of the fracture healing process, respectively. Since cortical bone is of higher impedance than the other tissues present in the fracture gap, it is critical that impedance measurements taken across the fracture reflect the tissues in the gap and are not completely dominated by the cortical bone around the fracture. Data collected from this simulated fracture indicate that our system can at least distinguish between injuries at stage two versus stage three of the healing process. This will enable tracking of fracture healing over time, and allow physicians to spot when a fracture does not progress through the different stages of healing at the anticipated rate. This will allow for early intervention to prevent high risk fractures from failing to heal in an acceptable time frame.
REFERENCES [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11]
V. CONCLUSION We have developed a system that utilizes impedance spectroscopy to study fractures and has shown promise in its ability detect the different tissue compositions that are expected to be at the fracture site during the progression of bone healing. Experiments have been conducted to ensure that presence of a bone plate and screws in the area of injury for fracture stabilization will not compromise our measurements. We have demonstrated that our system can distinguish between the different tissues present in healing fractures. Measurements taken across a simulated fracture further highlight the functionality of our sensors to classify fractures as being in one of the four stages of healing, which will provide physicians with additional quantitative information to help direct treatment. These experiments establish feasibility of our system to detect differences in electrical properties of the tissue at the fracture site as an injury heals, which can be validated in an in vivo model. Ultimately, understanding where a patient’s injury is within the healing process can diagnose delayed healing at an earlier stage and allow for timely intervention to prevent problem fractures from progressing to non-union.
[12] [13] [14] [15] [16]
ACKNOWLEDGMENT M. C. L. acknowledges Sarah Swisher, Amy Liao, and Kaylee Mann for their contribution to the control board and software development. We acknowledge support from the Berkeley Sensor and Actuator Center, the Swarm Lab at UC Berkeley, and the Orthopaedic Trauma Institute at San Francisco General Hospital. This work was supported by the National Science Foundation under grant no. EFRI-1240380 and by the Center for Disruptive Musculoskeletal Innovations (IIP-1361975). M. C. L. was supported by a National Science Foundation Graduate Research Fellowship. 5141
M. L. Schenker, et al., “Fracture Repair and Bone Grafting,” Orthopaedic Knowledge Update, pp. 15-25, 2014. K. F. Dickson, et al., “Importance of the blood supply in the healing of tibial fractures,” Contemp Orthop, vol. 30, no. 6, pp. 489-493, 1995. American Academy of Orthopaedic Surgeons, “The Burden of Musculoskeletal Diseases in the United States”, Rosemont, IL, 2008. S. Morshed, “Current Options for Determining Fracture Union,” Advances in Medicine, pp. 1-12, 2014. S. P. Whiley, et al., “Evaluating fracture healing using digital X-ray image analysis,” CME, vol. 29, no. 3, pp. 102-106, 2011. V. C. Protopappas, et al., “Ultrasonic Monitoring of Bone Fracture Healing,” Transactions on Ultrasonics, Ferroelectrics, and Frequency Control, vol. 55, no. 6, pp. 1243-1255, 2008. L. E., Claes and C. A. Heigele, “Magnitudes of local stress and strain along bony surfaces predict the course and type of fracture healing,” Journal of Biomechanics, pp. 255-266, 1999. B. McKibbin, et al., “The Biology of Fracture Healing in Long Bones,” Journal of Bone and Joint Surgery, pp. 150-162, 1978. H. M. Frost, et al., “The Biology of Fracture Healing: An Overview for Clinicians, Part I,” Clinical Orthopaedics and Related Research, pp. 283-293, 1989. R. D. Stoy, K. R. Foster, and H. P. Schwan, “Dielectric properties of mammalian tissues from 0.1 to 100 MHz: a summary of recent data,” Phys Med Biol, vol. 27, no. 4, pp. 501-513, 1982. H. Jongschaap, et al., “Electrical impedance tomography: a review of current literature,” European Journal of Radiology, vol. 18, pp. 165174, 1994. D. A. Dean, et al., “Electrical Impedance Spectroscopy Study of Biological Tissues,” J Electrostat, vol. 66, no. 3-4, pp. 165-177, 2008. S. Gabriel, R. W. Lau, and C. Gabriel, “The dielectric properties of biological tissues: II,” Phys Med Biol, vol. 41, pp. 2251-2269, 1996. I. V. Ciuchi, et al., “Impedance spectroscopy characterization of bone tissues,” Journal of Advanced Research in Physics, vol. 1, no. 1, pp. 15, 2010. S. Schaur, B. Jakoby, and G. Kroureif, “Position-dependent characterization of bone tissue with electrical impedance spectroscopy,” IEEE, pp. 1-4, 2012. S. L. Swisher, et al., “Impedance sensing device enables early detection of pressure ulcers in vivo,” Nature Communications, vol. 6, 2015.
