Effects of sterilization on the Tekscan digital pressure sensor

Medical Engineering & Physics 25 (2003) 775–780 www.elsevier.com/locate/medengphy

Technical note

Effects of sterilization on the Tekscan digital pressure sensor Howard J. Agins a,b,c, Valerie S. Harder a,b,c,∗, Eugene P. Lautenschlager c, James C. Kudrna a,b,c a Illinois Bone and Joint Institute, 2401 Ravine Way, Glenview, IL 60025, USA Department of Orthopaedic Surgery, Evanston Northwestern Healthcare, 1000 Central Street, Suite 880, Evanston, IL 60201, USA c Orthopaedic Biomaterials Laboratory, Northwestern University, 320 East Superior Street, Searle 2-401, Chicago, IL 60611, USA

b

Received 27 May 2002; received in revised form 9 April 2003; accepted 24 June 2003

Abstract Investigations into the effects of sterilization on a new biomechanical pressure sensor are necessary before contemplating in vivo use. Ten, designated Experimental, “K-Scan” digital pressure sensor arrays were sterilized with ethylene oxide gas (EtO), and their ability to accurately and reproducibly measure an applied load of 2225 N (500 lb) was assessed. Simultaneously, 10 un-sterilized sensor arrays, designated Control, were assessed. Each array was loaded 10 times inside a two-dimensional curved surface, and all arrays exhibited high reproducibility (coefficients of variation ⬍ 2.0%). Following sterilization, the Experimental sensors showed a 22.2% average decrease in recorded force, a statistically significant difference from the pre-sterile data (p ⬍ 0.002). However, when the Experimental sensors were re-calibrated post-sterilization, they showed only a 0.1% average decrease in recorded force, not a statistically significant difference (p ⬎ 0.05, b ⬍ 0.05). Following 1-week storage, trial 2 data of the Control sensors showed a less dramatic yet significant 3.4% average decrease in recorded force when compared to trial 1 data (p ⬍ 0.02). Control trial 2, once re-calibrated, showed a 0.5% average decrease in recorded force, not a statistically significant difference (p ⬎ 0.05, b ⬍ 0.05). Results suggest that, following EtO sterilization, accurate and reproducible pressure measurements can be obtained from KScan sensors when calibration is performed at time of use.  2003 IPEM. Published by Elsevier Ltd. All rights reserved. Keywords: Sterilization; Tekscan; Orthopaedic biomechanics; Calibration

1. Introduction Measurement of pressure distributions within articulating joints has been of interest to orthopaedic researchers for many years. Fuji Prescale pressure-sensitive film (Fuji Photo Co., Ltd, Tokyo, Japan), was first used over 20 years ago to measure in vitro contact pressures inside human joints [1] and is still being used today to measure in vitro pressures and areas for both physiologic [2–13] and artificial [14–17] human joints. A known disadvantage of Fuji film is the overestimation of contact area and stress when shear forces, caused by translational micro motion within a joint, lead to unintentional rupture of the dye-filled microcapsules [5]. Biome∗ Corresponding author. c/o Dr. James Kudrna, Illinois Bone and Joint Institute, 2401 Ravine Way, Glenview, IL 60025, USA. Tel.: +1-773-793-8526; fax: +1-847-998-9693. E-mail address: [email protected] (V.S. Harder).

chanical measurement error of Fuji film has been reported to be between 10% [1] and 15% [18]. Finite element modeling has shown that maximum contact pressures generated inside a joint using a piece of Fuji film could have an associated error of as much as 10– 20% [19]. Combining the error in measurement precision (10–15%) with the error due to the disruption of the joint congruity upon insertion of the Fuji film (10–20%), the contact pressures measured in biologic joints could have errors as high as 14–28% [19]. Tekscan digital pressure sensors (Tekscan, Inc., South Boston, MA, USA) are one of the newer available technologies for quantification of compressive loads. These thin film sensors are manufactured in many different sizes and shapes for a wide range of applications from industrial to medical. Researchers within the field of orthopaedics have used Tekscan sensors to quantify in vitro force and area distributions for both artificial [20,21] and physiologic [22] joints. One of Tekscan’s

1350-4533/$30.00  2003 IPEM. Published by Elsevier Ltd. All rights reserved. doi:10.1016/S1350-4533(03)00119-X

776

H.J. Agins et al. / Medical Engineering & Physics 25 (2003) 775–780

sensors, K-Scan, was designed to fit inside both the medial and lateral compartments of the human knee. The K-Scan sensors have been found to yield more accurate contact pressure and area distribution data than Fuji film [20,23]. Advantages of the K-Scan sensor include its ability to record multiple sets of data both statically and dynamically, in both two and three dimensions, with smaller standard deviations as compared to Fuji film [23]. In addition, interfacing directly with the I-Scan software and having the ability to convert all data to ASCII format, simplifies data management and reduces user error and bias. Utilization of this and other sensing technologies in vivo for orthopaedic applications will require functionality across curved surfaces and accuracy after sterilization. The central purpose of this study was to quantify the effects of ethylene oxide (EtO) gas sterilization on the K-Scan sensor and to determine the sensor’s ability to output accurate and reproducible measurements of force on a two-dimensional curved surface with and without re-calibration.

2. Materials and methods The following study used 10 K-Scan sensors (model 4000/0121T1/1500). Each sensor had two sensing arrays, for a total of 20 arrays available for testing. These sensors measured average pressure distributions within a physiologic range from 0 to approximately 10.3 MPa (1500 psi). Each K-Scan sensor array measured 0.1 mm thick, with 572 individual sensing points on a matrix of intersecting conductive ink lines; sensels. Each sensel had an area of 1.6 mm2, for a total array sensing area of 920 mm2. Each sensel was a variable resistor and relayed voltage signals to the computer for analog to digital conversion. The associated software (I-Scan version 4.231, Tekscan, Inc., South Boston, MA, USA) converted the digital signal (integers from 0 to 255) to a pressure value via a unique calibration equation calculated for each sensor array. The striped matrix of row and column electrodes are comprised of an electrically conductive silver ink and are printed directly onto two separate polyester films such as Mylar (DuPont, Wilmington, DE). As disclosed by Maness et al. in US Patent No. 4856993, a proprietary pressure-sensitive resistive ink, likely composed of a carbon–molybdenum disulfide in an acrylic binder, is applied in a thin coating over the entire surface of the film containing the row electrodes. On the other film, containing the column electrodes, the pressure-sensitive resistive ink is applied in a striped fashion to coat each column electrode while leaving the spaces between the electrodes free of ink. The two films are placed together with the row and column electrodes at right angles to each other, with the resistive coatings facing one

another. Finally, an adhesive material is applied along the edges to seal the films together [24]. The electrically conductive silver ink that is printed on the Mylar film in row and column lines is contained between the two Mylar film layers from the sensing pad all the way down the arm leading to the handle that connects the sensor to the computer. At the junction of the sensor and the handle, the electrode contact points are exposed so a physical connection can be made for transmission of the electrical signals. Initially, all sensor arrays were conditioned, calibrated and loaded using an Instron (model #1114, Instron, Canton, MA, USA). Each sensor array was placed between a solid acrylic rod (51 mm diameter 25 mm length of the curved contact face) and a concave mated acrylic surface (radius of curvature = 26 mm) as the base (Fig. 1); subjecting the sensor to a 2D curved compressive load with an approximate contact area of 650 mm2 (1 in2). This chosen radius of curvature was based on measurements from previous studies looking at radii of curvature for physiologic human joints [25–28]. A thin, flexible, nylon fabric was placed between the sensor array and the concave acrylic base to ensure an even distribution of load maintained well within the sensor array’s available sensing area. Once positioned between the 2D mated acrylic surfaces, the sensor arrays were conditioned by applying 2250 N five times prior to calibration. During conditioning, the position of the sensor array was adjusted until all applied load was contacting

Fig. 1. A schematic (not to scale) of the calibration and experimental loading apparatus. 2D mated acrylic surfaces with a radius of curvature = 26 mm and a load bearing area = 650 mm2.

H.J. Agins et al. / Medical Engineering & Physics 25 (2003) 775–780

area within the dimensions of the sensing array with no load being lost to the surrounding acrylic base. Using a one-point calibration equation (linear regression through (0, 0)) at an instantaneous applied load of 2225 N, each individual array was calibrated. The software detected the active contact area and calculated the calibration equation based on the average pressure distributed across the sensing array, thus generating a unique calibration equation for each sensor array. Once calibration was completed, the sensor array was not removed from or adjusted within the loading apparatus, thus maintaining identical contact area from calibration to experimental runs. For each experimental run, a 2225 N load was applied, and an instantaneous snapshot of the pressure distribution pattern was recorded and saved. The load was removed and re-applied 10 times for each array. A random sample of half the sensors were designated as the Control group, and the remaining half were designated as the Experimental group to be sterilized. Following completion of all initial testing, the 10 sensor arrays designated as the Experimental group were subjected to a 12 h EtO gas sterilization process consisting of three phases: conditioning, sterilization, and exhaust. The 10 sensor arrays in the Control group were stored for a period of 1 week at 22.5 ± 0.5 °C. Post-sterilization and storage, the Experimental sensor arrays and the Control sensor arrays were re-tested, duplicating the conditioning, calibration, and experimental loading protocols. Thus, each sensor array had a second calibration equation through which post-sterilization (Experimental) and the trial 2 (Control) data were fit. 2.1. Statistical analysis

777

trical lead failure or crimping. Each sensor was fully functional after tests were completed. Within the 10 repeated measurements on each sensor array, the output values of force were consistent. Neither the Control nor the Experimental sensor arrays significantly lost or gained sensitivity during the repeated trial runs (coefficients of variation ⬍ 2.0%). Data collected from the post-EtO sterilized Experimental sensors were fitted to both the pre-sterile calibration equation and the post-sterile calibration equation. The average force from the data fit to the pre-sterile calibration equation was 1732.9 ± 341.7 N (mean ± standard deviation) and when fit to the post-sterile calibration equation was 2227.0 ± 23.5 N (Fig. 2). This average 22.2% change in force was statistically significantly different well above a 99% confidence level (p = 0.0017). Data collected from the trial 2 Control sensors were fitted to both the trial 1 calibration equation and the trial 2 calibration equation. The average force from the data fit through the trial 1 calibration equation was 2146.3 ± 89.6 N and when fit through the trial 2 calibration equation was 2221.2 ± 17.9 N (Fig. 2). This considerably smaller change of 3.4% in force was still statistically significant above a 95% confidence level (p = 0.0162). Data collected from the Experimental sensor arrays during the initial (pre-sterilization) testing were fitted to the pre-sterile calibration equation, and data collected after EtO sterilization were fitted to the post-sterile calibration equation. The average force in the pre-sterile calibrated data set was 2230.3 ± 11.8 N and in the poststerile calibrated data set was 2227.0 ± 23.5 N (Fig. 3). This 0.1% difference in force was not statistically significant (p = 0.719). Analysis of b-error between the two

Coefficients of variation (100 × standard deviation / mean) were calculated to test for repeatability across the 10 measurements taken on each sensor array. Paired ttest analyses were conducted to determine statistical significance between average force values for each group of sensors. Statistical differences were significant at pⱕ0.05. If insignificant results (p ⬎ 0.05) were obtained from any comparison, the data sets were subjected to a Power analysis to determine if a Beta (b) error had been committed. b-error was calculated, assuming less than a 5% clinically acceptable change in force, and statistical similarity was accepted for any bⱕ0.05, making the Power ([1⫺b] × 100)ⱖ95%). 3. Results Visual inspection of the Experimental group sensors following EtO gas sterilization showed no signs of damage or difference in appearance when compared to the Control group sensors. No sensors were lost due to elec-

Fig. 2. The average forces and standard deviations of Experimental post-EtO data and Control trial 2 data. Comparison of data fit to the original calibration vs. same data fit to the new calibration. In both the Control and Experimental groups, the original calibration curves produced statistically significantly (p ⬍ 0.05) lower force than the new calibration curves.

778

H.J. Agins et al. / Medical Engineering & Physics 25 (2003) 775–780

Fig. 3. Average forces and standard deviations of trial 1 vs. trial 2 of the Control sensors and pre-EtO vs. post-EtO of the Experimental sensors. Data sets were fit to their appropriate calibration curves, and force comparisons proved to be statistically similar (p ⬎ 0.2, Power ⬎ 99%).

average force data sets for the Experimental group produced a Power ⬎ 99%. Similarly, for the Control group sensor arrays, the data collected during the first trial run were fitted to the trial 1 calibration equation and the data collected after 1 week storage were fitted to the trial 2 calibration equation. The average force resulting from the trial 1 calibrated data set was 2232.1 ± 12.0 N and from the trial 2 calibrated data set was 2221.2 ± 17.9 N (Fig. 3). This 0.5% difference was not statistically significant (p = 0.239). Analysis of b-error between the two average force data sets for the Control group produced a Power ⬎ 99%. 4. Discussion Orthopaedic biomechanics research of pressure distributions has already progressed from in vitro to in vivo investigation. To the best of our knowledge, there have been two published accounts of in vivo investigation into the distribution of pressure within joints using pressuresensitive material [29,30], and both were conducted during total knee arthroplasty. Fuji Prescale film was used first to obtain an intraoperative qualitative assessment of the pressure distribution prior to soft tissue balancing [29]. A different research group used Tekscan’s digital pressure sensors to quantify contact pressures during implant alignment and ligament balancing [30]. Both studies reported to use EtO gas sterilization, but neither commented on the effects the sterilization process may have had on the accuracy of their measurement tool. The purpose of this study was to determine the accuracy and reproducibility of the digital data generated by the K-Scan sensor following EtO gas sterilization. The results of this study suggest that the K-Scan pressure

sensor becomes less sensitive following sterilization, thus resulting in a decrease in the amount of force measured. Data collected from the 10 Experimental sensor arrays post-sterilization and then refitted to their earlier pre-sterile calibration equations indicate that the EtO gas sterilization causes the sensors to be, on average, 22% less responsive. However, if re-calibrated following sterilization, the K-Scan sensor is able to reproduce an accurate force reading. When the same post-sterile data are fit to a post-sterilization calibration equation, the resultant force is statistically equivalent to the magnitude of applied load. This fact supports the notion that this new, thin film, pressure-sensing system is a viable option for use in vivo. EtO is a colorless, toxic, reactive and highly flammable gas and is classified as a mutagenic, carcinogenic and reproductive hazard. Metals acting as catalysts for the decomposition of EtO (including copper, silver, mercury, magnesium and their alloys) should not be used as parts of the sterilizing equipment because they may react with impurities in the gas to form explosive compounds. Potassium, tin, zinc, aluminum and iron oxides tend to accelerate the polymerization of EtO. Equipment for storage and handling of EtO is generally fabricated from stainless steels so to eliminate the potential for rust, which can catalyze EtO polymerization. EtO rapidly attacks and degrades most organic polymers and elastomers, and therefore, alternative materials are used to manufacture the o-rings, gaskets and packaging materials that come into contact with the EtO gas during the sterilization process. EtO gas is used in sterilization processes because it is a substance that penetrates plastic material. Mylar, the plastic material used in the fabrication of the Tekscan sensor, is reported to not degrade by standard sterilization procedures involving EtO gas exposure. Information was not available concerning the adhesive used to seal the two Mylar layers together, and therefore, its reactivity with EtO gas was not known at this time. As noted in Section 2, the electrically conductive silver ink is contained between the two Mylar film layers from the sensing pad all the way down the arm leading to the handle that connects the sensor to the computer. However, at the junction of the sensor and the handle, the electrode contact points are exposed to air so we feel this may be the location on the sensor most affected by the EtO gas during sterilization. We also hypothesize that a thin oxidation film could form on these exposed contact points while in storage and in contact with air. This could help to explain the slight yet significant decrease in sensitivity noted in the Control sensors as well. The mechanism for this decrease in the sensitivity of the K-Scan sensor following exposure to the EtO gas sterilization process is not yet known and needs further investigation. Liggins et al., studying the effect of EtO

H.J. Agins et al. / Medical Engineering & Physics 25 (2003) 775–780

gas sterilization on Fuji Prescale pressure-sensitive film, reported an error (23%) in their pressure measurements if their Control group film calibration equations were used to interpret their sterilized film data [31]. Their research concluded that the pressure readings of the sterilized films were statistically significantly different from the pressure readings of the control films. In addition, they reported that the sterilized Prescale films produced greater stain densities than the control films, but they were unable to detect any qualitative differences between the films using a scanning electron microscope. It has not yet been determined how the EtO gas affects the Prescale film or the K-Scan sensor. This study also demonstrated a gradual loss of sensitivity in the Control group sensors between trials 1 and 2 (⫺3%), considerably less than the Experimental group (⫺22%) but still statistically significant. This suggests that the K-Scan sensor must be re-calibrated if it is to be re-used regardless of sterilization. This significant degradation in sensitivity of the Control sensors occurred over a 1-week period. The specific length of time associated with loss of sensitivity was not determined by this study. Current investigation into degradation of sensitivity as it relates to time is being conducted. This study provides information on the feasibility of using sterile K-Scan sensors to quantify joint forces in vivo given the sensor is calibrated post-sterilization. This leads to the question of whether or not a sterile calibration device is necessary for in vivo tests. The I-Scan software saves digitized data separate from generated calibration equations, so that the data may be fit to any calibration equation at any time. Therefore, we believe it may be possible to sterilize the K-Scan sensor, utilize its capability to record real time digitized measurements in a sterile environment, and then convert the digital data to values of force using a calibration equation generated immediately following in vivo experimentation in a nonsterile calibration device outside of the operating room. In support of the K-Scan sensor’s high degree of reproducibility across subsequent measurements, it may be possible to calibrate immediately after use rather than before use, thus avoiding the need for a sterile calibration device. Current investigation into the accuracy of post-experimental calibration equations is being conducted. In conclusion, digital data were reproducible across the 10 repeated loads on a 2D curved surface conducted during both loading protocols of Control and Experimental sensor arrays. Calibration equations generated at the time of use are paramount for the accurate translation of the digital signal into force quantities because EtO sterilization and the passing of time cause the sensors to be less responsive.

779

References [1] Fukubayashi T, Kurosawa H. The contact area and pressure distribution pattern of the knee: a study of normal and osteoarthritic knees. Acta Orthop Scand 1980;51:871–9. [2] Bay BK, Hamel AJ, Olson SA, Sharkey NA. Statically equivalent load and support conditions produce different hip joint contact pressures and periacetabular strains. J Biomech 1997;30:193–6. [3] Bruns J, Volkmer M, Luessenhop S. Pressure distribution at the knee joint: influence of varus and valgus deviation without and with ligament dissection. Arch Orthop Traum Su 1993;133:12–9. [4] Bruns J, Volkmer M, Luessenhop S. Pressure distribution in the knee joint. Influence of flexion with and without ligament dissection. Arch Orthop Traum Su 1994;113(4):204–9. [5] Baratz ME, Fu FH, Mengato R. Meniscal tears: the effect of meniscectomy and of repair on intraarticular contact areas and stress in the human knee. Am J Sport Med 1986;14(4):270–5. [6] Riegger-Krugh C, Gerhart TN, Powers WR, Hayes WC. Tibiofemoral contact pressures in degenerative joint disease. Clin Orthop 1998;348:233–45. [7] Paletta GA, Manning T, Snell E, Parker R, Bergfeld J. The effect of allograft meniscal replacement on intraarticular contact area and pressures in the human knee. Am J Sport Med 1997;25(5):692–8. [8] Alhalki MM, Maury LH, Howell SM. Contact mechanics of the medial tibial plateau after implantation of a medial meniscal allograft. Am J Sport Med 2000;28(3):370–6. [9] Ihn JC, Kim SJ, Park IH. In vitro study of contact area and pressure distribution in the human knee after partial and total meniscectomy. Int Orthop 1993;17:214–8. [10] Lee SB, Itoi E, O’Driscoll SW, An KN. Contact geometry at the undersurface of the acromion with and without a rotator cuff tear. Arthroscopy 2001;17(4):365–72. [11] Warner JP, Bowen MK, Deng XH, Hannafin JA, Arnoczky SP, Warren RF. Articular contact patterns of the normal glenohumeral joint. J Shoulder Elb Surg 1998;7(4):381–8. [12] Von Eisenhart-Rothe R, Adam C, Steinlechner M, Muller-Gerbl M, Eckstein F. Quantitative determination of joint incongruity and pressure distribution during simulated gait and cartilage thickness in the human hip joint. J Orthop Res 1999;17:532–9. [13] Konrath GA, Hamel AJ, Sharkey NA, Bay BK, Olson SA. Biomechanical consequences of anterior column fracture of the acetabulum. J Orthop Trauma 1998;12(8):547–52. [14] Szivek JA, Anderson PL, Benjamin JB. Average and peak contact stress distribution evaluation of total knee arthroplasties. J Arthroplasty 1996;11(8):367–75. [15] Szivek JA, Cutignola L, Volz RG. Tibio-femoral contact stress and stress distribution evaluation of total knee replacements. J Arthroplasty 1995;10(4):480–91. [16] Zdero R, Fenton PV, Rudan J, Bryant JT. Fuji film and ultrasound measurement of total knee arthroplasty contact areas. J Arthroplasty 2001;16(3):367–75. [17] Liggins AB, Finlay JB. Recording contact areas and pressures in joint interfaces. In: Little EG, editor. Experimental mechanics: technology transfer between high tech engineering and biomechanics. Amsterdam: Elsevier Science Publishers; 1992. p. 71–88. [18] Hale JE, Brown TD. Contact stress gradient detection limits of Pressensor film. J Biomech Eng 1992;114:352–7. [19] Wu JZ, Herzog W, Epstein M. Effects of inserting a Presensor film into articular joints on the actual contact mechanics. J Biomech Eng 1998;120:655–9. [20] Harris ML, Morberg WM, Walsh WR. An improved method for measuring tibiofemoral contact areas in total knee arthroplasty: a comparison of K-scan sensor and Fuji-film. J Biomech 1999;32:951–8. [21] Matsuda S, Ishinishi T, Whiteside LA. Contact stresses with an

780

[22]

[23]

[24]

[25]

H.J. Agins et al. / Medical Engineering & Physics 25 (2003) 775–780

unresurfaced patella in total knee arthroplasty: the effect of femoral component design. Orthopedics 2000;23(3):213–8. Cooper PS, Nowak MD, Shaer J. Calcaneocuboid joint pressures with lateral column lengthening (evans) procedure. Foot Ankle Int 1997;18(4):199–205. Pavlovic JL, Takahashi Y, Bechtold RB, Gustilo RB, Kyle RF. Can the Tekscan sensor accurately measure dynamic pressures in the knee joint? In: Conference Proceedings, 17th Annual American Society of Biomechanics Meeting, Iowa City, IA. 1993. p. 135–6. Maness WL, Golden RF, Benjamin MH, Podoloff RM, Pressure and contact sensor system for measuring dental occlusion. US Patent No. 4856993; 1989. Iannotti JP, Gabriel JP, Schneck SL, Evans BG, Misra S. The normal glenohumeral relationships. An anatomical study of one hundred and forty shoulders. J Bone Joint Surg Am 1992;74(4):491–500.

[26] Yoshioka Y, Siu D, Cooke TD. The anatomy and functional axes of the femur. J Bone Joint Surg Am 1987;69(6):873–80. [27] Nuno N, Ahmed AM. Sagittal profile of the femoral condyles and its application to femorotibial contact analysis. J Biomech Eng 2001;123(1):18–26. [28] Kapandji IA. In: 2nd ed. The physiology of the joints, vol. 2. Edinburgh, London: E & S Livingstone; 1970. p. 86–7. [29] Takahashi T, Wada Y, Yamamoto H. Soft-tissue balancing with pressure distribution during total knee arthroplasty. J Bone Joint Surg Br 1997;79(2):235–9. [30] Wallace AL, Harris ML, Walsh WR, Bruce WM. Intraoperative assessment of tibiofemoral contact stresses in total knee arthroplasty. J Arthroplasty 1998;13:923–7. [31] Liggins AB, Hardie WR, Finlay JB. Sterilization of Fuji pressuresensitive film. Med Eng Phys 1994;16:496–500.