- Email: [email protected]
Advanced System for Implant Stability Testing (ASIST)
Author’s Accepted Manuscript ADVANCED SYSTEM FOR STABILITY TESTING (ASIST)
IMPLANT
L. Westover, G. Faulkner, W. Hodgetts, D. Raboud www.elsevier.com/locate/jbiomech
PII: DOI: Reference:
S0021-9290(16)31060-0 http://dx.doi.org/10.1016/j.jbiomech.2016.09.043 BM7907
To appear in: Journal of Biomechanics Accepted date: 30 September 2016 Cite this article as: L. Westover, G. Faulkner, W. Hodgetts and D. Raboud, ADVANCED SYSTEM FOR IMPLANT STABILITY TESTING (ASIST), Journal of Biomechanics, http://dx.doi.org/10.1016/j.jbiomech.2016.09.043 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
L Westover et al.
ADVANCED SYSTEM FOR IMPLANT STABILITY TESTING (ASIST) L. Westover1*, G. Faulkner2, W. Hodgetts3, D. Raboud1 1
2
3
Department of Mechanical Engineering University of Alberta Edmonton, AB, Canada
Rehabilitation Research and Technology Development Glenrose Rehabilitation Hospital Edmonton, AB, Canada
Communication Sciences and Disorders, Rehabilitation Medicine University of Alberta Edmonton, AB, Canada
Original Article Keywords: bone anchored implants; implant stability; bone-implant interface
Word Count: 3999 *Corresponding Author: Lindsey Westover Department of Mechanical Engineering University of Alberta 10-203 Donadeo Innovation Centre for Engineering 9211 116 Street, Edmonton AB, T6G 1H9, Canada Phone: 780-221-7344 Fax: 780-492-2244 Email: [email protected]
1
L Westover et al.
ABSTRACT (143 Words)
This study presents the Advanced System for Implant Stability Testing (ASIST) which provides a non-invasive, quantitative measure of the stability of percutaneous implants used for craniofacial rehabilitation such as bone anchored hearing aids or dental implants. The ASIST uses an impact technique coupled with an analytical model which allows the measure to be independent of the system components. This paper presents a laboratory evaluation of the ASIST for the Oticon Medical Ponto and the Cochlear Baha Connect bone anchored hearing aid (BAHA) systems. There is minimal effect of abutment length on the ASIST Stability Coefficient (ASC) value, indicating that the method is able to isolate the interface properties from the overall system and the measurement is independent of attached components. Additionally, the ASIST was able to detect differences between different implant installations suggesting that it may be sensitive to changes in interface stiffness.
1
INTRODUCTION
Percutaneous implants are used in prosthetic reconstruction of the head and neck to improve function and aesthetics in various applications. Implants can be used to reconstruct dental arches, install bone anchored hearing aids (BAHA), and reconstruct craniofacial features after trauma or cancer treatment. The procedure works by installing a screw-shaped titanium implant into the bone. An attached abutment penetrates the skin surface and the device (e.g. the hearing processor or prosthetic tooth) is attached to the 2
L Westover et al.
abutment for use by the patient. Bone anchored implants rely on a structural integration between the implant and the bone, termed osseointegration (Brånemark et al., 1983; Brånemark et al., 1985; Tjellström et al., 1983). As a result, it is important to clinically monitor the integrity of the bone-implant interface in order to prescribe loading and identify the risk of failure.
Because most conventional medical diagnostics cannot be used with implants, several mechanical testing methods have been developed, relying on the principle that the osseointegration at the interface is related to the stiffness at this interface (Atsumi et al., 2007; Mathieu et al., 2014; Meredith, 1997; Swain et al., 2008a). A well-integrated implant is believed to be stiffer than a poorly integrated one.
To measure this stiffness, resonance frequency analysis (RFA) uses the idea that the resonance frequency of the system comprised of the implant, adjoining superstructure, and supporting bone is dependent on the stiffness at the bone-implant interface. Assuming the system components remain constant, changes in the resonance frequency indicate changes at the interface (Faulkner et al., 1999; Lin et al., 2010; Meredith et al., 1996; Meredith, 1997; Pérez et al., 2008; Swain et al., 2008a; Swain et al., 2008b).
The OsstellTM (Osstell, Göteborg, Sweden) is a commercially available instrument based on RFA (Sennerby and Meredith, 2008) that provides an implant stability quotient (ISQ), with higher numbers indicating greater stability. The OsstellTM is widely used clinically (Faber et al., 2012; Foghsgaard and Caye-Thomasen, 2014; McLarnon et al., 2012; 3
L Westover et al.
Nelissen et al., 2016); however, it has several limitations. It uses SmartPegsTM which must be attached directly to either the implant or abutment and it does not consider the system components in the interpretation of the resonance frequency. As a result, the ISQ is not able to isolate the actual interface properties.
The Periotest® (Medizintechnik Gulden, Modautal, Germany) is another commercially available system originally developed for natural teeth, which uses an impact technique to assess the interface (Lukas and Schulte, 1990). This system has been applied to implants with limited success primarily because osseointegrated implants are considerably stiffer than natural teeth (Aparicio et al., 2006; Hobkirk and Wiskott, 2006). The Periotest® does not require attachment to the implant/abutment, so it can be used for any type of implant-abutment system. Similar to the OsstellTM however, the primary disadvantage of the Periotest® is that it does not take the system components into account, so again it is unclear how the measurement actually relates to the interface properties. Additionally, the Periotest® analysis uses a filtered acceleration signal which removes significant features in the system response (Swain et al., 2008a; Swain et al., 2008b).
To better understand the physical properties at the interface Swain et al. (2006; 2008a; 2008b) coupled the impact technique using the Periotest® handpiece with an analytical model of the implant-abutment system and the surrounding bone for percutaneous implants. Their approach, unlike the Periotest®, uses the raw acceleration signal and, with the inclusion of the analytical model, allows the measurement to be essentially 4
L Westover et al.
independent of the details of the implant-abutment system. This approach has been evaluated with in vitro experiments for BAHA implants (with a modified abutment) and dental implants as well as in vivo experiments for BAHA patients (Mo, 2009; Swain, 2006).
While the coupled impact technique and analytical model is a promising method, the previous work has several limitations including qualitative matching between the measured accelerometer signal and the analytical model, limited scope of the analytical models, and a damping model that does not adequately represent the physiological system.
We build on the ideas of the coupled impact technique and analytical model and develop a new Advanced System for Implant Stability Testing (ASIST) which provides a noninvasive, quantitative measure of implant stability that is independent of the system components. We address the limitations of the previous work and develop the analytical models for commercially available (clinically used) BAHA implant-abutment systems. Additionally, we implement a more physiologically appropriate damping model. This study presents a laboratory evaluation of the ASIST for the Oticon Medical Ponto and Cochlear Baha Connect systems.
5
L Westover et al.
2 2.1
METHODS Hardware and Measurement
The ASIST consists of a handpiece, the ASIST unit, a USB dongle, and a computer (Figure 1). The handpiece is extracted from the Periotest® Classic (Medizintechnik Gulden, Modautal, Germany) and contains the impact rod and accelerometer. The ASIST unit contains the circuitry to initiate the impact, record the accelerometer data, and transmit the strike data via Bluetooth to the USB dongle in the computer (Kamal et al., 2015). Similar to the Periotest®, an impact event with the ASIST consists of 16 strikes in 4 seconds.
2.2
Analytical Model
The BAHA implant system (Figure 2) is modeled as a four degree of freedom (4-DOF) system composed of two rigid bodies (implant with mass and moment of inertia and abutment with mass ! and moment of inertia particle with mass " = 9.4g (Swain et al., 2006).
! ).
,
The impact rod is modeled as a
The stiffness parameters in the model are a linear spring at the impact site between the impact rod and the abutment (# ) and a linear torsional spring between the implant and the abutment (#$ ). The bone-implant interface is modeled as a distributed stiffness % per unit area surrounding the implant. The coordinates used to describe the system are the
horizontal displacement of the impact rod (&' ), the horizontal displacement of the center
of the abutment at the height of impact (&( ), the rotation of the abutment ()' ), and the 6
L Westover et al.
rotation of the implant ()( ). Thus, &̈ ' represents the acceleration of the impact rod which is experimentally measured by the accelerometer.
The equations of motion for free vibration of the 4-DOF system are given by: [+]{&̈ } + [-]{&̇ } + [#]{&} = {0}
(1)
&' &( where {&} = 1) 2 as illustrated in Figure 2. ' )(
The modeled system is only valid while the impact rod and abutment are in contact. The impact is modeled as a free vibration initial value problem during this time. The initial conditions are zero displacement for all coordinates, an initial velocity of the impact rod (&̇ ' (0) = 67 ), and zero velocity for all other coordinates.
The mass [+], stiffness [#], and damping [-] matrices can be obtained for this system
using the method of influence coefficients (Den Hartog, 1985). The matrix [+] is given by: " ⎡ 0 [+] = ⎢⎢ ⎢0 ⎣0 where:
0 + ! ! <>! − ( + ! )ℎ − (A − <> )
0 ! <>! − ( + ! )ℎ >! )( + ℎ( ! + ! (ℎ − < ℎ(A − <> )
0 ⎤ − (A − <> ) ⎥ (2) ℎ(A − <> ) ⎥⎥ > )( ⎦ + (A − <
A = length of the implant
<>! = distance from the bottom of the abutment to the center of mass
7
L Westover et al.
<> = distance from the bottom of the implant to the center of mass
ℎ = distance from the bottom of the abutment to the striking location The stiffness matrix [#] is:
# ⎡ ⎢−# [#] = ⎢ 0 ⎢ ⎣ 0
−# # + 4%E7 (A − AF ) −4%E7 (A − AF )ℎ −2%E7 (A( − A(F )
0 −4%E7 (A − AF )ℎ #$ + 4%E7 (A − AF )ℎ(
0 ⎤ −2%E7 (A( − A(F ) ⎥ −#$ + 2%E7 (A( − A(F )ℎ⎥ (3) ⎥ H −#$ + 2%E7 (A( − A(F )ℎ #$ + I %E7 (AI − AIF ) ⎦
where AF is the portion of the implant above the bone surface (i.e. the distributed stiffness
at the bone-implant interface only acts along the length JK = A − AF ) and E7 is the
implant radius.
One of the major differences in the analytical model compared to Swain et al. (2008a) is the damping assumption. We use a general viscous damping model, while Swain et al. (2008a) used a proportional damping model. In the BAHA implant-abutment system, proportional damping is not representative of the physiological conditions where considerable damping is expected to occur at the bone-implant interface with relatively minimal damping in the rest of the system. Our model assumes that damping is distributed along the bone-implant interface (similar to the interface stiffness) and is negligible elsewhere. In particular (unlike proportional damping) there is no damping at the spring locations # and #$ .
The damping matrix [-] is given by: 8
L Westover et al.
0 ⎡0 ⎢ [-] = ⎢0 ⎢ ⎣0
0 4LE7 (A − AF ) −4LE7 (A − AF )ℎ −2LE7 (A( − A(F )
0 −4LE7 (A − AF )ℎ 4LE7 (A − AF )ℎ( 2LE7 (A( − A(F )ℎ
0 −2LE7 (A( − A(F )⎤ ⎥ 2LE7 (A( − A(F )ℎ ⎥ ⎥ H LE7 (AI − AIF ) ⎦ I
(4)
where L is the viscous damping coefficient per unit area surrounding the implant. The damping at the interface includes the effects of damping within the bone and the connection between the bone and the implant at their interface.
2.3
Estimating the Interface Properties
A custom Mathematica program (Wolfram Mathematica 10, Champaign, IL, USA) was developed for the ASIST analysis.
2.3.1 Measured Acceleration Signals – Curve Fit Approximation To correlate the measured signal with the acceleration response of the analytical model, it is necessary to extract relevant information from the measured signal. This is done by fitting a curve to the measured data using least squares minimization. This intermediate analysis procedure allows the damping in the measured signal to be estimated and provides a measure of the amplitude of the signal which can be used to normalize the acceleration response. Normalization allows proper matching between the measured data and the modeled signal because the model prediction has units of acceleration (m/s ( ) while the measured data has units of voltage (Swain et al., 2008a).
9
L Westover et al.
Since the measured signals show two prominent modes of vibration, the curve fit model is based on a simplified 2-DOF vibration response which provides an adequate representation of the data. The model parameters are determined by minimizing the sum of the squared error between the curve fit model and the measured data. An example of the measured accelerometer data and the resulting curve fit are shown in Figure 3.
2.3.2 Matching the Analytical Model to the Measured Acceleration
The analytical model consists of three stiffness parameters: #$ , # , and %. #$ and # are
determined as described in Section 2.6.2. The interface stiffness is, in general, unknown and is determined from the experimental/numerical procedure by matching the model predicted acceleration with the measured acceleration during an impact.
A numerical procedure was implemented to match the curve fit model from the measured data to the analytical model of the system. Matching is done by minimizing the Euclidean norm between the model predicted acceleration and the curve fit approximation of the measured signal over the time of the strike. Thus the interface stiffness per unit area is estimated by finding the value of % that minimizes: N = O∑VKW'QR(SK ) − RT (SK )U
(
(5)
where:
R(SK ) = the model predicted acceleration of the impact rod at time SK
10
L Westover et al.
RT (SK ) = the curve fit approximation of the measured acceleration at time SK
N = the Euclidean norm (measure of the distance between the two functions over the strike)
Figure 4 shows an example of the measured signal and the matched acceleration response from the analytical model.
2.3.3 Outcome Measure The primary outcome measure is the ASIST Stability Coefficient (ASC) which is a nondimensional value directly related to the effective interface stiffness (over the implant area) and normalized to a nominal stiffness at the impact site (# ). The interface stiffness per unit area (%) is determined from the numerical procedure described above. Subsequently, the effective stiffness in the direction of the impact (#XYY ) over the interface is calculated as: #XYY = 4%E7 JK
(6)
where JK = A − AT is the length of the threaded portion of the implant. #XYY represents
the total effective stiffness of the bone-implant connection (N/m). The ASC relates this effective interface stiffness to a nominal stiffness at the impact site, providing a nondimensional stiffness ratio. The ASC is calculated from:
11
L Westover et al.
Z\- =
^_`` ^a
=
Hbcd ef ^a
with # = 4 × 10i N/m
(7)
The normalization procedure is done simply for ease of interpretation in a clinical setting.
2.4
Application to Oticon Medical Ponto Implant-Abutment System
Four sizes of Oticon Medical Ponto BAHA implants were obtained from the manufacturer (Oticon Medical, Göteborg, Sweden) (Table 1). Three lengths of Ponto abutments were obtained: 6mm, 9mm, and 12mm. Three samples of each implant and one sample of each abutment were used.
The center of mass and mass moment of inertia for the implants and abutments were calculated based on approximate, simplified geometry (Table 1). Figure 2 (not to scale) shows a representative example of the simplified geometry. The implant, abutment, and connecting screw are all made from titanium.
2.5
Application to Cochlear Baha Connect Implant-Abutment System
One type of Cochlear Baha Connect implant (BI300) was obtained from the manufacturer (Cochlear Bone Anchored Solutions, Göteborg, Sweden). Four different lengths of the BA400 abutment were obtained (nominal lengths: 6mm, 8mm, 10mm, 12mm). One sample of the implant and one sample of each abutment were used. To use the analytical model for the Cochlear system, the mass and geometry properties must be modified (Table 2). The implant, abutment, and connecting screw are all made from titanium with 12
L Westover et al.
portions of the implant coated with TiOblastTM and portions of the abutment coated with hydroxyapatite (DermaLockTM Technology) (www.cochlear.com).
2.6
Development and Experimental Evaluation
2.6.1 Experimental Testing Implants were installed into discs of either photoelastic FRB-20 (Measurements Group Inc., Raleigh, NC) or 3D printed PLA (Cubify, 3D Systems Inc.) to simulate a wellintegrated implant in bone. PLA (polylactic acid) is a thermoplastic commonly used in 3D printing. Two samples of each Oticon Medical implant type were installed in FRB-20 and one sample was installed in PLA. An adhesive (5-minute epoxy for the FRB-20 samples and super glue for the PLA samples) was used to ensure as uniform an interface as possible between the implant and the material (Swain et al., 2008a; Swain et al., 2008b). The different materials and adhesives were used to simulate a range of interface stiffness values (i.e. the FRB-20 samples were expected to be stiffer than the PLA samples). Each of the Oticon Medical implant samples were tested with the three corresponding abutments (6mm, 9mm, 12mm).
One implant sample was prepared for the Cochlear Baha Connect System. The BI300 implant was installed in a disc of FRB-20, secured with super glue, and tested with each of the four BA400 abutments (6mm, 8mm, 10mm 12mm).
13
L Westover et al.
The disc containing an implant was clamped to a platform and an abutment was attached with 25Ncm of torque as per manufacturer instructions (Figure 5). The ASIST handpiece was held in a custom built adjustable stand (Swain et al., 2008a) to minimize variations in handpiece position. Five ASIST measurements were taken with each implant-abutment combination.
2.6.2 Determining model parameters
In order to objectively determine the interface stiffness %, all other parameters in the
system should be known. Thus, we must determine appropriate values for the stiffness parameters #$ and # .
Similar to Swain et al. (2008a), the maximum torsional stiffness for each system (#$ jkl ) is determined by assuming the abutment acts as a cantilever beam fixed at the implantabutment connection. It can be shown that #$ jkl is dependent on the abutment length. The actual torsional stiffness is assumed to be less than the maximum value: #$ = n#$ jkl
(7)
The torsional stiffness coefficient (α) is assumed to be constant within each BAHA system.
14
L Westover et al.
The impact stiffness # represents the local deformation of the impact rod and the
abutment at the location of the strike. # is also assumed to be constant across abutments from the same system.
The developed model allows the matching (minimization of Equation (5)) to be done
using multivariable minimization. For example, the combination of # and % in the analytical model resulting in the best match with the given measured acceleration can be determined. This multivariable minimization was used to determine appropriate average values for the model parameters # and n that are applicable across the range of cases
tested for each BAHA system. Once these appropriate values were determined, they were fixed in the model, leaving % as the only unknown stiffness parameter.
Appropriate values for # and n were determined for the Oticon Medical system using the numerical model (# = 2.90 × 10i N/m, n = 0.26). Similarly, the appropriate values for the Cochlear system were found to be # = 4.74 × 10i N/m and n = 0.50.
For any patient with a given implant-abutment system, all geometric and inertia
parameters (Tables 1-2) as well as the stiffness parameters # and #$ are known. Thus
the only unknown in the analytical model is the interface stiffness %, which is the outcome variable of interest (ASC is directly calculated from %) and is determined from
the ASIST analysis procedure (Section 2.3).
15
L Westover et al.
2.6.3 ASIST Evaluation The measured data was analyzed with the model parameters chosen as described above. For each measurement, the interface stiffness and ASC value were determined.
3
RESULTS
The results from a representative Oticon Medical implant sample (4.50mm diameter, 4mm length) with three abutment lengths are shown in Figure 6. The differences in the acceleration signal with each abutment are clearly visible in the figure. These differences are expected and illustrate that changing the abutment changes the system and thus changes the measured acceleration. The shorter abutments result in a higher natural frequency than the longer abutments. There is a good fit between the measured
acceleration signal and the analytical model in all cases with average E ( values from 0.97-0.98
(averaged
over
repeated
measurements
for
each
implant-abutment
combination). Additionally, despite the significant change in the measured response, the calculated ASC values are within 4.0 points (range: 37.9-41.9) for this example.
The average ASC values for each Oticon Medical implant-abutment combination are shown in Figure 7. For this system, average ASC values ranged from 7.6-78.1 (average over repeated measurements). Considering each group of bars in Figure 7 (each implant sample with the three abutments), the effect of abutment length on the ASC value was found to be small as the average difference in ASC due to changes in abutment length is 2.9 (8.0% of ASC). In general, samples with higher ASC values have a slightly larger variation due to changing abutment length (Figure 7). At higher values of the interface 16
L Westover et al.
stiffness and higher natural frequencies, small changes in the frequency response can result in larger changes in the ASC. The sensitivity of the ASC at higher frequencies should not be clinically important because implants with higher ASC values would be considered stable and would not be of clinical concern.
Comparing the different groups of bars in each subfigure in Figure 7, there are clear differences in the ASC values determined for each implant sample, suggesting that the measure is sensitive to changes in the support conditions. Figure 8 shows a representative example for two different samples of the same Oticon Medical implant type (4.50mm diameter, 4mm length). The first sample is installed in FRB-20 and the second sample is installed in PLA. Both samples in have a 9mm abutment attached. There are clear differences in the measured acceleration response with the PLA sample having a lower fundamental frequency than the FRB-20 sample. In this case, this results in a 78% reduction in the ASC value (ASC=37.9 for FRB-20, ASC=8.3 for PLA). In both cases there is a good fit between the measured acceleration and the analytical model with E ( values of 0.98 and 0.95 for the stiffer and softer samples respectively.
Figure 9 shows the results for the Cochlear implant with four different abutments. There is a good fit between the measured acceleration signal and the analytical model in all cases (E ( between 0.94-0.98). The difference in ASC due to changing abutment length is 2.8 (9.5% of ASC). The range of ASC values for the four different abutments is 26.631.8, noting that only one implant sample was tested for the Cochlear system.
17
L Westover et al.
4
DISCUSSION
This study presents the Advanced System for Implant Stability Testing (ASIST) for stability measurement of BAHA systems. The ASIST uses an impact technique coupled with an analytical model to provide a measure of bone-implant interface integrity that is independent of attached components. The application to the Oticon Medical Ponto and Cochlear Baha Connect systems were presented and evaluated through laboratory testing.
There is a good fit between the measured acceleration signal and the analytical model in all cases, indicating that the analytical model is a good representation of the physical system and is of appropriate complexity to capture the mechanical behaviour at the boneimplant interface. There is minimal effect of abutment length on the ASC value, indicating that the method can isolate the interface properties from the overall system and the measurement is independent of attached components. Clinically, the abutment is only changed in rare cases, so one may argue that independence from abutment length is not important for relative longitudinal changes. However, this result gives us confidence that the measurement is representative of the interface properties and would allow for consistent comparisons across clinical situations. It also allows comparison over time for the few patients where the abutment is changed. Additionally, it may allow comparisons between patients or development of a suggested range of ASC values that indicate healthy implants. Independence from the system components simplifies the interpretation of the results because the interpretation of the ASC is the same regardless of abutment length.
18
L Westover et al.
The ASIST was able to detect differences between different implant installations suggesting that it may be sensitive to changes in interface stiffness. Ultimately this system will be used clinically to detect stability changes so it must be sensitive enough to detect meaningful changes.
The two samples of each Oticon Medical implant type installed in FRB-20 followed the same installation procedure; however, variations in the stability values between the two similar samples were noted in some cases, specifically the long/narrow implant M50128 and the short/wide implant M51061 (Figure 7). These variations can be attributed to uncontrolled differences in the installation technique such as installation torque, application of epoxy, or slight angulation in the implant. Thus each installation should be treated separately and the interface stiffness values cannot be expected to be equal for all implant samples.
The ASIST can provide significant advantages over commercially available systems such as the Periotest® or the OsstellTM. With the OsstellTM, a specific SmartPegTM must be used for each application and these cannot be used with systems such as dental implants unless the prosthetic tooth is removed. The ASIST is more versatile and can be used with any implant-abutment system. The Periotest® uses a similar impact technique to the ASIST; however, it relies on a filtered acceleration signal which does not reflect the actual acceleration response and alters the contact time calculation (Swain et al., 2008a; Swain et al., 2008b). Conversely, the ASIST uses the raw acceleration signal, more accurately reflecting the system response. The primary disadvantage of both the OsstellTM 19
L Westover et al.
and the Periotest® is that the system components are not included in the stability calculation, thus they are not able to isolate the interface properties. The ISQ is affected by various factors including measurement location (implant or abutment), implant and abutment length, and SmartPegTM type (Nelissen et al., 2015). Similar comments can be made regarding the Periotest® measure (Swain et al., 2008a, Swain et al., 2008b). Conversely, the analytical model allows the ASIST to isolate the interface properties from the rest of the system and provides a better understanding of the mechanical properties at the interface. This presents a significant advantage in terms of the clinical utility of the device.
The present study assumes that the interface stiffness is uniformly distributed over the implant, while physiologically the interface may vary along the length. Since the implant length for BAHA applications is small compared to the overall system, it is expected that a potentially non-uniform interface would not significantly affect the model response. This may be more important for applications where the implant length comprises a greater proportion of the overall system, which can be investigated in future applications.
Future work will include sensitivity analyses on the experimental measurement, analytical model, and numerical procedure including recommendations for clinical protocols. Additionally, an in vivo clinical study will be presented to evaluate longitudinal changes in interface stability for BAHA patients. Further, correlations between the ASIST and biological measures of osseointegration can be investigated using
20
L Westover et al.
experimental animal models. Additional future work will focus on developing the ASIST for a range of applications including dental implants and natural teeth.
The ASIST is a non-invasive technique for assessing implant stability that addresses the limitations of previous approaches by reliably isolating the bone-implant interface properties. This system can allow clinicians to quickly and non-invasively evaluate the status of the bone-implant interface from initial implantation over the entire life of the BAHA system.
CONFLICT OF INTERESTS STATEMENT There are no conflicts of interest to be reported for this publication.
ACKNOWLEDGMENTS The authors would like to acknowledge Fraaz Kamal and Edmond Lou from the Glenrose Rehabilitation Hospital for their work on this study. This research was supported by the Natural Sciences and Engineering Research Council of Canada (NSERC), Alberta Innovates Health Solutions (AIHS), and Western Economic Diversification Canada (WD).
21
L Westover et al.
REFERENCES Aparicio, C., Lang, N.P., Rangert, B., 2006. Validity and clinical significance of biomechanical testing of implant/bone interfaces. Clinical Oral Implants Research 17(Suppl 2), 2-7. Atsumi, M., Park, SH., Wang, HL., 2007. Methods used to assess implant stability: Current status. International Journal of Oral & Maxillofacial Implants 22(5), 743-754. Brånemark, PI., 1983. Osseointegration and its experimental background. Journal of Prosthetic Dentistry 50(3), 399-410. Brånemark, PI., Zarb, GA., Albrektsson, T., 1985. Tissue-Integrated Prosthesis: Osseointegration in Clinical Dentistry. Quintessence Publishing Co., Chicago IL, USA. Den Hartog, JP., 1985. Mechanical Vibrations. Dover Publications Inc. New York, NY, USA. Faber, H., Dun, C., Nelissen, R., Mylanus, E., Cremers, C., Hol, M., 2012. Boneanchored hearing implant loading at 3 weeks: Stability and tolerability after 6 months. Otology & Neurotology 34, 104-110. Faulkner, G., Wolfaardt, J.F., Chan, A., 1999. Measuring abutment/implant joint integrity with the Periotest instrument. International Journal of Oral & Maxillofacial Implants 14(5), 681-688. Foghsgaard, S., Caye-Thomasen, P., 2014. A new wide-diameter bone-anchored hearing implant – Prospective 1-year data on complications, implant stability, and survival. Otology & Neurotology 35, 1238-1241. Hobkirk, J.A., Wiskott, H.W.A., 2006. Biomechanical aspects of oral implants: Consensus report of working group 1. Clinical Oral Implants Research 17(Suppl 2), 5254. Kamal, F., Lou, E., Faulkner, G., Westover, L., 2015. ASIST – A Bluetooth Real-Time Implant/Bone Interface Stability Measurement Device. 5th International Congress on Bone Conduction Hearing and Related Technologies, Fairmont Chateau Lake Louise, Canada. Lin, D., Li, Q., Li, W., Duckmanton, N., Swain, M. 2010. Mandibular bone remodeling induced by dental implant. Journal of Biomechanics 43(2), 287-293. Lukas, D. and Schulte, W., 1990. Periotest – A dynamic procedure for the diagnosis of the human periodontium. Clinical Physics and Physiological Measurement 11(1), 65-75. 22
L Westover et al.
Mathieu, V., Vayron, R., Richard, G., Lambert, G., Naili, S., Meningaud, J-P., Haiat, G. 2014. Biomechanical determinants of the stability of dental implants: Influence of the bone-implant interface properties. Journal of Biomechanics 47(1), 3-13. McLarnon, C.M., Johnson, I., Davison, T., Hill, J., Henderson, B., Leese, D., Marley, S., 2012. Evidence for early loading of osseointegrated implants for bone conduction at 4 weeks. Otology & Neurotology 33, 1578-1582. Meredith, N., Alleyne, D., Cawley, P., 1996. Quantitative determination of the stability of the implant-tissue interface using resonance frequency analysis. Clinical Oral Implants Research 7(3), 261-267. Meredith, N., 1997. On the clinical measurement of implant stability and osseointegration. Ph.D. thesis, Göteborg University, Sweden. Mo, A., 2009. The impact technique for monitoring intraoral implant stability. M.Sc. Thesis, University of Alberta, Edmonton, Canada. Nelissen, R., den Besten, C., Mylanus, E., Hol, M., 2016. Stability, survival, and tolerability of a 4.5 mm-wide bone-anchored hearing implant: 6-month data from a randomized controlled trial. European Archives of Otorhinolaryngology 273(1), 105-111. Nelissen, R., Wigren, S., Flynn, M., Meijer, G., Mylanus, E., Hol, M. 2015. Application and interpretation of resonance frequency analysis in auditory osseointegrated implants: A review of literature and establishment of practical recommendations. Otology & Neurotology 36(9), 1518-1524. Pérez, M.A., Moreo, P., García-Aznar, J.M., Doblaré, M. 2008. Computational simulation of dental implant osseointegration through resonance frequency analysis. Journal of Biomechanics 41(2), 316-325. Sennerby, L., Meredith, N., 2008. Implant stability measurements using resonance frequency analysis: biological and biomechanical aspects and clinical implications. Periodontology 2000 47, 51-66. Swain, R., 2006. The development and modeling of an impact test to determine the boneimplant interface properties of osseointegrated implants. Ph.D. Thesis, University of Alberta, Edmonton, Canada. Swain, R., Faulkner, G., Raboud, D., Wolfaardt, J., 2008a. A dynamic analytical model for impact evaluation of percutaneous implants. Journal of Biomechanical Engineering 130(5), 051013-1-13. Swain, R., Faulkner, G., Raboud, D., Wolfaardt, J., 2008b. An Improved Impact Technique for Monitoring Percutaneous Implant Integrity. International Journal of Oral & Maxillofacial Implants 23(2), 263-269. 23
L Westover et al.
Tjellström, A., Rosenhall, U., Lindström, J., Hallen, O., Albrektsson, T., Brånemark, PI., 1983. Five-year experience with skin-penetrating bone-anchored implants in the temporal bone. Acta Oto-Laryngologica 95(5-6), 568-575.
24
L Westover et al.
FIGURE CAPTIONS LIST Figure 1: ASIST components including the ASIST unit, the handpiece, a USB dongle, and a computer Figure 2: Schematic of the 4-DOF model of the BAHA implant/abutment system (not to scale) Figure 3: Measured accelerometer data (colored) with the curve fit model (red) Figure 4: Matched acceleration signal between the model (black) and measured data (colored) Figure 5: Laboratory test set-up showing the FRB-20 disc containing the implant with a 9 mm abutment attached. The disc is clamped to a platform for testing and the ASIST handpiece is held in a custom stand. Figure 6: Representative example for one Oticon Medical implant sample with three abutment lengths Figure 7: ASC values for all Oticon Medical samples with three abutment lengths (6 mm, 9 mm, 12 mm) Figure 8: Representative example for two different samples of one Oticon Medical implant type (4.50 mm diameter, 4 mm length) and 9 mm abutment Figure 9: One Cochlear implant sample with four different abutment lengths
25
L Westover et al.
TABLES Table 1: Mass and geometry properties for Oticon Medical implants and abutments. Mass (rt ) (u)
Diameter (vwx ) (yy)
Length (zt − z| ) (yy)
0.126
3.75
3.00
0.157
3.75
0.173
~> (yy)
t (u yyv )
4.00
2.71
0.51
4.50
3.00
2.04
0.46
0.222
4.50
4.00
2.54
0.76
Oticon Medical Abutments
Mass (r ) (u)
Shaft Diam. (mm)
Length (z ) (yy)
6 mm 9 mm 12 mm
0.51 0.72 0.97
5.00 5.00 5.00
6.00 9.00 12.00
~> (yy)
(u yyv )
Oticon Medical Implants M50106 (Short/Narrow) M50128 (Long/Narrow) M51061 (Short/Wide) M51062 (Long/Wide)
2.19
2.98 4.52 5.95
0.31
2.72 5.99 13.07
*Note that the location of the center of mass (~>)is measured from the bottom of the component and the mass moment of inertia () is about the center of mass. Table 2: Mass and geometry properties for Cochlear implants and abutments Cochlear Implant BI300 Cochlear Abutments 6 mm 8 mm 10 mm 12 mm
Mass (rt ) (u) 0.22 Mass (r ) (u) 0.50 0.70 1.03 1.40
Diameter (vwx ) (yy) 4.45
Length (zt − z| ) (yy) 4.00 Length (z ) (yy) 4.67 6.67 8.67 10.67
~> (yy) 2.66
t (u yyv )
~> (yy)
(u yyv )
2.15 3.85 4.93 6.01
0.78
2.68 4.42 8.22 14.39
*Note that the location of the center of mass (~>) is measured from the bottom of the component and the mass moment of inertia () is about the center of mass.
26
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6
Figure 7
Figure 8
Figure 9
IMPLANT
L. Westover, G. Faulkner, W. Hodgetts, D. Raboud www.elsevier.com/locate/jbiomech
PII: DOI: Reference:
S0021-9290(16)31060-0 http://dx.doi.org/10.1016/j.jbiomech.2016.09.043 BM7907
To appear in: Journal of Biomechanics Accepted date: 30 September 2016 Cite this article as: L. Westover, G. Faulkner, W. Hodgetts and D. Raboud, ADVANCED SYSTEM FOR IMPLANT STABILITY TESTING (ASIST), Journal of Biomechanics, http://dx.doi.org/10.1016/j.jbiomech.2016.09.043 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
L Westover et al.
ADVANCED SYSTEM FOR IMPLANT STABILITY TESTING (ASIST) L. Westover1*, G. Faulkner2, W. Hodgetts3, D. Raboud1 1
2
3
Department of Mechanical Engineering University of Alberta Edmonton, AB, Canada
Rehabilitation Research and Technology Development Glenrose Rehabilitation Hospital Edmonton, AB, Canada
Communication Sciences and Disorders, Rehabilitation Medicine University of Alberta Edmonton, AB, Canada
Original Article Keywords: bone anchored implants; implant stability; bone-implant interface
Word Count: 3999 *Corresponding Author: Lindsey Westover Department of Mechanical Engineering University of Alberta 10-203 Donadeo Innovation Centre for Engineering 9211 116 Street, Edmonton AB, T6G 1H9, Canada Phone: 780-221-7344 Fax: 780-492-2244 Email: [email protected]
1
L Westover et al.
ABSTRACT (143 Words)
This study presents the Advanced System for Implant Stability Testing (ASIST) which provides a non-invasive, quantitative measure of the stability of percutaneous implants used for craniofacial rehabilitation such as bone anchored hearing aids or dental implants. The ASIST uses an impact technique coupled with an analytical model which allows the measure to be independent of the system components. This paper presents a laboratory evaluation of the ASIST for the Oticon Medical Ponto and the Cochlear Baha Connect bone anchored hearing aid (BAHA) systems. There is minimal effect of abutment length on the ASIST Stability Coefficient (ASC) value, indicating that the method is able to isolate the interface properties from the overall system and the measurement is independent of attached components. Additionally, the ASIST was able to detect differences between different implant installations suggesting that it may be sensitive to changes in interface stiffness.
1
INTRODUCTION
Percutaneous implants are used in prosthetic reconstruction of the head and neck to improve function and aesthetics in various applications. Implants can be used to reconstruct dental arches, install bone anchored hearing aids (BAHA), and reconstruct craniofacial features after trauma or cancer treatment. The procedure works by installing a screw-shaped titanium implant into the bone. An attached abutment penetrates the skin surface and the device (e.g. the hearing processor or prosthetic tooth) is attached to the 2
L Westover et al.
abutment for use by the patient. Bone anchored implants rely on a structural integration between the implant and the bone, termed osseointegration (Brånemark et al., 1983; Brånemark et al., 1985; Tjellström et al., 1983). As a result, it is important to clinically monitor the integrity of the bone-implant interface in order to prescribe loading and identify the risk of failure.
Because most conventional medical diagnostics cannot be used with implants, several mechanical testing methods have been developed, relying on the principle that the osseointegration at the interface is related to the stiffness at this interface (Atsumi et al., 2007; Mathieu et al., 2014; Meredith, 1997; Swain et al., 2008a). A well-integrated implant is believed to be stiffer than a poorly integrated one.
To measure this stiffness, resonance frequency analysis (RFA) uses the idea that the resonance frequency of the system comprised of the implant, adjoining superstructure, and supporting bone is dependent on the stiffness at the bone-implant interface. Assuming the system components remain constant, changes in the resonance frequency indicate changes at the interface (Faulkner et al., 1999; Lin et al., 2010; Meredith et al., 1996; Meredith, 1997; Pérez et al., 2008; Swain et al., 2008a; Swain et al., 2008b).
The OsstellTM (Osstell, Göteborg, Sweden) is a commercially available instrument based on RFA (Sennerby and Meredith, 2008) that provides an implant stability quotient (ISQ), with higher numbers indicating greater stability. The OsstellTM is widely used clinically (Faber et al., 2012; Foghsgaard and Caye-Thomasen, 2014; McLarnon et al., 2012; 3
L Westover et al.
Nelissen et al., 2016); however, it has several limitations. It uses SmartPegsTM which must be attached directly to either the implant or abutment and it does not consider the system components in the interpretation of the resonance frequency. As a result, the ISQ is not able to isolate the actual interface properties.
The Periotest® (Medizintechnik Gulden, Modautal, Germany) is another commercially available system originally developed for natural teeth, which uses an impact technique to assess the interface (Lukas and Schulte, 1990). This system has been applied to implants with limited success primarily because osseointegrated implants are considerably stiffer than natural teeth (Aparicio et al., 2006; Hobkirk and Wiskott, 2006). The Periotest® does not require attachment to the implant/abutment, so it can be used for any type of implant-abutment system. Similar to the OsstellTM however, the primary disadvantage of the Periotest® is that it does not take the system components into account, so again it is unclear how the measurement actually relates to the interface properties. Additionally, the Periotest® analysis uses a filtered acceleration signal which removes significant features in the system response (Swain et al., 2008a; Swain et al., 2008b).
To better understand the physical properties at the interface Swain et al. (2006; 2008a; 2008b) coupled the impact technique using the Periotest® handpiece with an analytical model of the implant-abutment system and the surrounding bone for percutaneous implants. Their approach, unlike the Periotest®, uses the raw acceleration signal and, with the inclusion of the analytical model, allows the measurement to be essentially 4
L Westover et al.
independent of the details of the implant-abutment system. This approach has been evaluated with in vitro experiments for BAHA implants (with a modified abutment) and dental implants as well as in vivo experiments for BAHA patients (Mo, 2009; Swain, 2006).
While the coupled impact technique and analytical model is a promising method, the previous work has several limitations including qualitative matching between the measured accelerometer signal and the analytical model, limited scope of the analytical models, and a damping model that does not adequately represent the physiological system.
We build on the ideas of the coupled impact technique and analytical model and develop a new Advanced System for Implant Stability Testing (ASIST) which provides a noninvasive, quantitative measure of implant stability that is independent of the system components. We address the limitations of the previous work and develop the analytical models for commercially available (clinically used) BAHA implant-abutment systems. Additionally, we implement a more physiologically appropriate damping model. This study presents a laboratory evaluation of the ASIST for the Oticon Medical Ponto and Cochlear Baha Connect systems.
5
L Westover et al.
2 2.1
METHODS Hardware and Measurement
The ASIST consists of a handpiece, the ASIST unit, a USB dongle, and a computer (Figure 1). The handpiece is extracted from the Periotest® Classic (Medizintechnik Gulden, Modautal, Germany) and contains the impact rod and accelerometer. The ASIST unit contains the circuitry to initiate the impact, record the accelerometer data, and transmit the strike data via Bluetooth to the USB dongle in the computer (Kamal et al., 2015). Similar to the Periotest®, an impact event with the ASIST consists of 16 strikes in 4 seconds.
2.2
Analytical Model
The BAHA implant system (Figure 2) is modeled as a four degree of freedom (4-DOF) system composed of two rigid bodies (implant with mass and moment of inertia and abutment with mass ! and moment of inertia particle with mass " = 9.4g (Swain et al., 2006).
! ).
,
The impact rod is modeled as a
The stiffness parameters in the model are a linear spring at the impact site between the impact rod and the abutment (# ) and a linear torsional spring between the implant and the abutment (#$ ). The bone-implant interface is modeled as a distributed stiffness % per unit area surrounding the implant. The coordinates used to describe the system are the
horizontal displacement of the impact rod (&' ), the horizontal displacement of the center
of the abutment at the height of impact (&( ), the rotation of the abutment ()' ), and the 6
L Westover et al.
rotation of the implant ()( ). Thus, &̈ ' represents the acceleration of the impact rod which is experimentally measured by the accelerometer.
The equations of motion for free vibration of the 4-DOF system are given by: [+]{&̈ } + [-]{&̇ } + [#]{&} = {0}
(1)
&' &( where {&} = 1) 2 as illustrated in Figure 2. ' )(
The modeled system is only valid while the impact rod and abutment are in contact. The impact is modeled as a free vibration initial value problem during this time. The initial conditions are zero displacement for all coordinates, an initial velocity of the impact rod (&̇ ' (0) = 67 ), and zero velocity for all other coordinates.
The mass [+], stiffness [#], and damping [-] matrices can be obtained for this system
using the method of influence coefficients (Den Hartog, 1985). The matrix [+] is given by: " ⎡ 0 [+] = ⎢⎢ ⎢0 ⎣0 where:
0 + ! ! <>! − ( + ! )ℎ − (A − <> )
0 ! <>! − ( + ! )ℎ >! )( + ℎ( ! + ! (ℎ − < ℎ(A − <> )
0 ⎤ − (A − <> ) ⎥ (2) ℎ(A − <> ) ⎥⎥ > )( ⎦ + (A − <
A = length of the implant
<>! = distance from the bottom of the abutment to the center of mass
7
L Westover et al.
<> = distance from the bottom of the implant to the center of mass
ℎ = distance from the bottom of the abutment to the striking location The stiffness matrix [#] is:
# ⎡ ⎢−# [#] = ⎢ 0 ⎢ ⎣ 0
−# # + 4%E7 (A − AF ) −4%E7 (A − AF )ℎ −2%E7 (A( − A(F )
0 −4%E7 (A − AF )ℎ #$ + 4%E7 (A − AF )ℎ(
0 ⎤ −2%E7 (A( − A(F ) ⎥ −#$ + 2%E7 (A( − A(F )ℎ⎥ (3) ⎥ H −#$ + 2%E7 (A( − A(F )ℎ #$ + I %E7 (AI − AIF ) ⎦
where AF is the portion of the implant above the bone surface (i.e. the distributed stiffness
at the bone-implant interface only acts along the length JK = A − AF ) and E7 is the
implant radius.
One of the major differences in the analytical model compared to Swain et al. (2008a) is the damping assumption. We use a general viscous damping model, while Swain et al. (2008a) used a proportional damping model. In the BAHA implant-abutment system, proportional damping is not representative of the physiological conditions where considerable damping is expected to occur at the bone-implant interface with relatively minimal damping in the rest of the system. Our model assumes that damping is distributed along the bone-implant interface (similar to the interface stiffness) and is negligible elsewhere. In particular (unlike proportional damping) there is no damping at the spring locations # and #$ .
The damping matrix [-] is given by: 8
L Westover et al.
0 ⎡0 ⎢ [-] = ⎢0 ⎢ ⎣0
0 4LE7 (A − AF ) −4LE7 (A − AF )ℎ −2LE7 (A( − A(F )
0 −4LE7 (A − AF )ℎ 4LE7 (A − AF )ℎ( 2LE7 (A( − A(F )ℎ
0 −2LE7 (A( − A(F )⎤ ⎥ 2LE7 (A( − A(F )ℎ ⎥ ⎥ H LE7 (AI − AIF ) ⎦ I
(4)
where L is the viscous damping coefficient per unit area surrounding the implant. The damping at the interface includes the effects of damping within the bone and the connection between the bone and the implant at their interface.
2.3
Estimating the Interface Properties
A custom Mathematica program (Wolfram Mathematica 10, Champaign, IL, USA) was developed for the ASIST analysis.
2.3.1 Measured Acceleration Signals – Curve Fit Approximation To correlate the measured signal with the acceleration response of the analytical model, it is necessary to extract relevant information from the measured signal. This is done by fitting a curve to the measured data using least squares minimization. This intermediate analysis procedure allows the damping in the measured signal to be estimated and provides a measure of the amplitude of the signal which can be used to normalize the acceleration response. Normalization allows proper matching between the measured data and the modeled signal because the model prediction has units of acceleration (m/s ( ) while the measured data has units of voltage (Swain et al., 2008a).
9
L Westover et al.
Since the measured signals show two prominent modes of vibration, the curve fit model is based on a simplified 2-DOF vibration response which provides an adequate representation of the data. The model parameters are determined by minimizing the sum of the squared error between the curve fit model and the measured data. An example of the measured accelerometer data and the resulting curve fit are shown in Figure 3.
2.3.2 Matching the Analytical Model to the Measured Acceleration
The analytical model consists of three stiffness parameters: #$ , # , and %. #$ and # are
determined as described in Section 2.6.2. The interface stiffness is, in general, unknown and is determined from the experimental/numerical procedure by matching the model predicted acceleration with the measured acceleration during an impact.
A numerical procedure was implemented to match the curve fit model from the measured data to the analytical model of the system. Matching is done by minimizing the Euclidean norm between the model predicted acceleration and the curve fit approximation of the measured signal over the time of the strike. Thus the interface stiffness per unit area is estimated by finding the value of % that minimizes: N = O∑VKW'QR(SK ) − RT (SK )U
(
(5)
where:
R(SK ) = the model predicted acceleration of the impact rod at time SK
10
L Westover et al.
RT (SK ) = the curve fit approximation of the measured acceleration at time SK
N = the Euclidean norm (measure of the distance between the two functions over the strike)
Figure 4 shows an example of the measured signal and the matched acceleration response from the analytical model.
2.3.3 Outcome Measure The primary outcome measure is the ASIST Stability Coefficient (ASC) which is a nondimensional value directly related to the effective interface stiffness (over the implant area) and normalized to a nominal stiffness at the impact site (# ). The interface stiffness per unit area (%) is determined from the numerical procedure described above. Subsequently, the effective stiffness in the direction of the impact (#XYY ) over the interface is calculated as: #XYY = 4%E7 JK
(6)
where JK = A − AT is the length of the threaded portion of the implant. #XYY represents
the total effective stiffness of the bone-implant connection (N/m). The ASC relates this effective interface stiffness to a nominal stiffness at the impact site, providing a nondimensional stiffness ratio. The ASC is calculated from:
11
L Westover et al.
Z\- =
^_`` ^a
=
Hbcd ef ^a
with # = 4 × 10i N/m
(7)
The normalization procedure is done simply for ease of interpretation in a clinical setting.
2.4
Application to Oticon Medical Ponto Implant-Abutment System
Four sizes of Oticon Medical Ponto BAHA implants were obtained from the manufacturer (Oticon Medical, Göteborg, Sweden) (Table 1). Three lengths of Ponto abutments were obtained: 6mm, 9mm, and 12mm. Three samples of each implant and one sample of each abutment were used.
The center of mass and mass moment of inertia for the implants and abutments were calculated based on approximate, simplified geometry (Table 1). Figure 2 (not to scale) shows a representative example of the simplified geometry. The implant, abutment, and connecting screw are all made from titanium.
2.5
Application to Cochlear Baha Connect Implant-Abutment System
One type of Cochlear Baha Connect implant (BI300) was obtained from the manufacturer (Cochlear Bone Anchored Solutions, Göteborg, Sweden). Four different lengths of the BA400 abutment were obtained (nominal lengths: 6mm, 8mm, 10mm, 12mm). One sample of the implant and one sample of each abutment were used. To use the analytical model for the Cochlear system, the mass and geometry properties must be modified (Table 2). The implant, abutment, and connecting screw are all made from titanium with 12
L Westover et al.
portions of the implant coated with TiOblastTM and portions of the abutment coated with hydroxyapatite (DermaLockTM Technology) (www.cochlear.com).
2.6
Development and Experimental Evaluation
2.6.1 Experimental Testing Implants were installed into discs of either photoelastic FRB-20 (Measurements Group Inc., Raleigh, NC) or 3D printed PLA (Cubify, 3D Systems Inc.) to simulate a wellintegrated implant in bone. PLA (polylactic acid) is a thermoplastic commonly used in 3D printing. Two samples of each Oticon Medical implant type were installed in FRB-20 and one sample was installed in PLA. An adhesive (5-minute epoxy for the FRB-20 samples and super glue for the PLA samples) was used to ensure as uniform an interface as possible between the implant and the material (Swain et al., 2008a; Swain et al., 2008b). The different materials and adhesives were used to simulate a range of interface stiffness values (i.e. the FRB-20 samples were expected to be stiffer than the PLA samples). Each of the Oticon Medical implant samples were tested with the three corresponding abutments (6mm, 9mm, 12mm).
One implant sample was prepared for the Cochlear Baha Connect System. The BI300 implant was installed in a disc of FRB-20, secured with super glue, and tested with each of the four BA400 abutments (6mm, 8mm, 10mm 12mm).
13
L Westover et al.
The disc containing an implant was clamped to a platform and an abutment was attached with 25Ncm of torque as per manufacturer instructions (Figure 5). The ASIST handpiece was held in a custom built adjustable stand (Swain et al., 2008a) to minimize variations in handpiece position. Five ASIST measurements were taken with each implant-abutment combination.
2.6.2 Determining model parameters
In order to objectively determine the interface stiffness %, all other parameters in the
system should be known. Thus, we must determine appropriate values for the stiffness parameters #$ and # .
Similar to Swain et al. (2008a), the maximum torsional stiffness for each system (#$ jkl ) is determined by assuming the abutment acts as a cantilever beam fixed at the implantabutment connection. It can be shown that #$ jkl is dependent on the abutment length. The actual torsional stiffness is assumed to be less than the maximum value: #$ = n#$ jkl
(7)
The torsional stiffness coefficient (α) is assumed to be constant within each BAHA system.
14
L Westover et al.
The impact stiffness # represents the local deformation of the impact rod and the
abutment at the location of the strike. # is also assumed to be constant across abutments from the same system.
The developed model allows the matching (minimization of Equation (5)) to be done
using multivariable minimization. For example, the combination of # and % in the analytical model resulting in the best match with the given measured acceleration can be determined. This multivariable minimization was used to determine appropriate average values for the model parameters # and n that are applicable across the range of cases
tested for each BAHA system. Once these appropriate values were determined, they were fixed in the model, leaving % as the only unknown stiffness parameter.
Appropriate values for # and n were determined for the Oticon Medical system using the numerical model (# = 2.90 × 10i N/m, n = 0.26). Similarly, the appropriate values for the Cochlear system were found to be # = 4.74 × 10i N/m and n = 0.50.
For any patient with a given implant-abutment system, all geometric and inertia
parameters (Tables 1-2) as well as the stiffness parameters # and #$ are known. Thus
the only unknown in the analytical model is the interface stiffness %, which is the outcome variable of interest (ASC is directly calculated from %) and is determined from
the ASIST analysis procedure (Section 2.3).
15
L Westover et al.
2.6.3 ASIST Evaluation The measured data was analyzed with the model parameters chosen as described above. For each measurement, the interface stiffness and ASC value were determined.
3
RESULTS
The results from a representative Oticon Medical implant sample (4.50mm diameter, 4mm length) with three abutment lengths are shown in Figure 6. The differences in the acceleration signal with each abutment are clearly visible in the figure. These differences are expected and illustrate that changing the abutment changes the system and thus changes the measured acceleration. The shorter abutments result in a higher natural frequency than the longer abutments. There is a good fit between the measured
acceleration signal and the analytical model in all cases with average E ( values from 0.97-0.98
(averaged
over
repeated
measurements
for
each
implant-abutment
combination). Additionally, despite the significant change in the measured response, the calculated ASC values are within 4.0 points (range: 37.9-41.9) for this example.
The average ASC values for each Oticon Medical implant-abutment combination are shown in Figure 7. For this system, average ASC values ranged from 7.6-78.1 (average over repeated measurements). Considering each group of bars in Figure 7 (each implant sample with the three abutments), the effect of abutment length on the ASC value was found to be small as the average difference in ASC due to changes in abutment length is 2.9 (8.0% of ASC). In general, samples with higher ASC values have a slightly larger variation due to changing abutment length (Figure 7). At higher values of the interface 16
L Westover et al.
stiffness and higher natural frequencies, small changes in the frequency response can result in larger changes in the ASC. The sensitivity of the ASC at higher frequencies should not be clinically important because implants with higher ASC values would be considered stable and would not be of clinical concern.
Comparing the different groups of bars in each subfigure in Figure 7, there are clear differences in the ASC values determined for each implant sample, suggesting that the measure is sensitive to changes in the support conditions. Figure 8 shows a representative example for two different samples of the same Oticon Medical implant type (4.50mm diameter, 4mm length). The first sample is installed in FRB-20 and the second sample is installed in PLA. Both samples in have a 9mm abutment attached. There are clear differences in the measured acceleration response with the PLA sample having a lower fundamental frequency than the FRB-20 sample. In this case, this results in a 78% reduction in the ASC value (ASC=37.9 for FRB-20, ASC=8.3 for PLA). In both cases there is a good fit between the measured acceleration and the analytical model with E ( values of 0.98 and 0.95 for the stiffer and softer samples respectively.
Figure 9 shows the results for the Cochlear implant with four different abutments. There is a good fit between the measured acceleration signal and the analytical model in all cases (E ( between 0.94-0.98). The difference in ASC due to changing abutment length is 2.8 (9.5% of ASC). The range of ASC values for the four different abutments is 26.631.8, noting that only one implant sample was tested for the Cochlear system.
17
L Westover et al.
4
DISCUSSION
This study presents the Advanced System for Implant Stability Testing (ASIST) for stability measurement of BAHA systems. The ASIST uses an impact technique coupled with an analytical model to provide a measure of bone-implant interface integrity that is independent of attached components. The application to the Oticon Medical Ponto and Cochlear Baha Connect systems were presented and evaluated through laboratory testing.
There is a good fit between the measured acceleration signal and the analytical model in all cases, indicating that the analytical model is a good representation of the physical system and is of appropriate complexity to capture the mechanical behaviour at the boneimplant interface. There is minimal effect of abutment length on the ASC value, indicating that the method can isolate the interface properties from the overall system and the measurement is independent of attached components. Clinically, the abutment is only changed in rare cases, so one may argue that independence from abutment length is not important for relative longitudinal changes. However, this result gives us confidence that the measurement is representative of the interface properties and would allow for consistent comparisons across clinical situations. It also allows comparison over time for the few patients where the abutment is changed. Additionally, it may allow comparisons between patients or development of a suggested range of ASC values that indicate healthy implants. Independence from the system components simplifies the interpretation of the results because the interpretation of the ASC is the same regardless of abutment length.
18
L Westover et al.
The ASIST was able to detect differences between different implant installations suggesting that it may be sensitive to changes in interface stiffness. Ultimately this system will be used clinically to detect stability changes so it must be sensitive enough to detect meaningful changes.
The two samples of each Oticon Medical implant type installed in FRB-20 followed the same installation procedure; however, variations in the stability values between the two similar samples were noted in some cases, specifically the long/narrow implant M50128 and the short/wide implant M51061 (Figure 7). These variations can be attributed to uncontrolled differences in the installation technique such as installation torque, application of epoxy, or slight angulation in the implant. Thus each installation should be treated separately and the interface stiffness values cannot be expected to be equal for all implant samples.
The ASIST can provide significant advantages over commercially available systems such as the Periotest® or the OsstellTM. With the OsstellTM, a specific SmartPegTM must be used for each application and these cannot be used with systems such as dental implants unless the prosthetic tooth is removed. The ASIST is more versatile and can be used with any implant-abutment system. The Periotest® uses a similar impact technique to the ASIST; however, it relies on a filtered acceleration signal which does not reflect the actual acceleration response and alters the contact time calculation (Swain et al., 2008a; Swain et al., 2008b). Conversely, the ASIST uses the raw acceleration signal, more accurately reflecting the system response. The primary disadvantage of both the OsstellTM 19
L Westover et al.
and the Periotest® is that the system components are not included in the stability calculation, thus they are not able to isolate the interface properties. The ISQ is affected by various factors including measurement location (implant or abutment), implant and abutment length, and SmartPegTM type (Nelissen et al., 2015). Similar comments can be made regarding the Periotest® measure (Swain et al., 2008a, Swain et al., 2008b). Conversely, the analytical model allows the ASIST to isolate the interface properties from the rest of the system and provides a better understanding of the mechanical properties at the interface. This presents a significant advantage in terms of the clinical utility of the device.
The present study assumes that the interface stiffness is uniformly distributed over the implant, while physiologically the interface may vary along the length. Since the implant length for BAHA applications is small compared to the overall system, it is expected that a potentially non-uniform interface would not significantly affect the model response. This may be more important for applications where the implant length comprises a greater proportion of the overall system, which can be investigated in future applications.
Future work will include sensitivity analyses on the experimental measurement, analytical model, and numerical procedure including recommendations for clinical protocols. Additionally, an in vivo clinical study will be presented to evaluate longitudinal changes in interface stability for BAHA patients. Further, correlations between the ASIST and biological measures of osseointegration can be investigated using
20
L Westover et al.
experimental animal models. Additional future work will focus on developing the ASIST for a range of applications including dental implants and natural teeth.
The ASIST is a non-invasive technique for assessing implant stability that addresses the limitations of previous approaches by reliably isolating the bone-implant interface properties. This system can allow clinicians to quickly and non-invasively evaluate the status of the bone-implant interface from initial implantation over the entire life of the BAHA system.
CONFLICT OF INTERESTS STATEMENT There are no conflicts of interest to be reported for this publication.
ACKNOWLEDGMENTS The authors would like to acknowledge Fraaz Kamal and Edmond Lou from the Glenrose Rehabilitation Hospital for their work on this study. This research was supported by the Natural Sciences and Engineering Research Council of Canada (NSERC), Alberta Innovates Health Solutions (AIHS), and Western Economic Diversification Canada (WD).
21
L Westover et al.
REFERENCES Aparicio, C., Lang, N.P., Rangert, B., 2006. Validity and clinical significance of biomechanical testing of implant/bone interfaces. Clinical Oral Implants Research 17(Suppl 2), 2-7. Atsumi, M., Park, SH., Wang, HL., 2007. Methods used to assess implant stability: Current status. International Journal of Oral & Maxillofacial Implants 22(5), 743-754. Brånemark, PI., 1983. Osseointegration and its experimental background. Journal of Prosthetic Dentistry 50(3), 399-410. Brånemark, PI., Zarb, GA., Albrektsson, T., 1985. Tissue-Integrated Prosthesis: Osseointegration in Clinical Dentistry. Quintessence Publishing Co., Chicago IL, USA. Den Hartog, JP., 1985. Mechanical Vibrations. Dover Publications Inc. New York, NY, USA. Faber, H., Dun, C., Nelissen, R., Mylanus, E., Cremers, C., Hol, M., 2012. Boneanchored hearing implant loading at 3 weeks: Stability and tolerability after 6 months. Otology & Neurotology 34, 104-110. Faulkner, G., Wolfaardt, J.F., Chan, A., 1999. Measuring abutment/implant joint integrity with the Periotest instrument. International Journal of Oral & Maxillofacial Implants 14(5), 681-688. Foghsgaard, S., Caye-Thomasen, P., 2014. A new wide-diameter bone-anchored hearing implant – Prospective 1-year data on complications, implant stability, and survival. Otology & Neurotology 35, 1238-1241. Hobkirk, J.A., Wiskott, H.W.A., 2006. Biomechanical aspects of oral implants: Consensus report of working group 1. Clinical Oral Implants Research 17(Suppl 2), 5254. Kamal, F., Lou, E., Faulkner, G., Westover, L., 2015. ASIST – A Bluetooth Real-Time Implant/Bone Interface Stability Measurement Device. 5th International Congress on Bone Conduction Hearing and Related Technologies, Fairmont Chateau Lake Louise, Canada. Lin, D., Li, Q., Li, W., Duckmanton, N., Swain, M. 2010. Mandibular bone remodeling induced by dental implant. Journal of Biomechanics 43(2), 287-293. Lukas, D. and Schulte, W., 1990. Periotest – A dynamic procedure for the diagnosis of the human periodontium. Clinical Physics and Physiological Measurement 11(1), 65-75. 22
L Westover et al.
Mathieu, V., Vayron, R., Richard, G., Lambert, G., Naili, S., Meningaud, J-P., Haiat, G. 2014. Biomechanical determinants of the stability of dental implants: Influence of the bone-implant interface properties. Journal of Biomechanics 47(1), 3-13. McLarnon, C.M., Johnson, I., Davison, T., Hill, J., Henderson, B., Leese, D., Marley, S., 2012. Evidence for early loading of osseointegrated implants for bone conduction at 4 weeks. Otology & Neurotology 33, 1578-1582. Meredith, N., Alleyne, D., Cawley, P., 1996. Quantitative determination of the stability of the implant-tissue interface using resonance frequency analysis. Clinical Oral Implants Research 7(3), 261-267. Meredith, N., 1997. On the clinical measurement of implant stability and osseointegration. Ph.D. thesis, Göteborg University, Sweden. Mo, A., 2009. The impact technique for monitoring intraoral implant stability. M.Sc. Thesis, University of Alberta, Edmonton, Canada. Nelissen, R., den Besten, C., Mylanus, E., Hol, M., 2016. Stability, survival, and tolerability of a 4.5 mm-wide bone-anchored hearing implant: 6-month data from a randomized controlled trial. European Archives of Otorhinolaryngology 273(1), 105-111. Nelissen, R., Wigren, S., Flynn, M., Meijer, G., Mylanus, E., Hol, M. 2015. Application and interpretation of resonance frequency analysis in auditory osseointegrated implants: A review of literature and establishment of practical recommendations. Otology & Neurotology 36(9), 1518-1524. Pérez, M.A., Moreo, P., García-Aznar, J.M., Doblaré, M. 2008. Computational simulation of dental implant osseointegration through resonance frequency analysis. Journal of Biomechanics 41(2), 316-325. Sennerby, L., Meredith, N., 2008. Implant stability measurements using resonance frequency analysis: biological and biomechanical aspects and clinical implications. Periodontology 2000 47, 51-66. Swain, R., 2006. The development and modeling of an impact test to determine the boneimplant interface properties of osseointegrated implants. Ph.D. Thesis, University of Alberta, Edmonton, Canada. Swain, R., Faulkner, G., Raboud, D., Wolfaardt, J., 2008a. A dynamic analytical model for impact evaluation of percutaneous implants. Journal of Biomechanical Engineering 130(5), 051013-1-13. Swain, R., Faulkner, G., Raboud, D., Wolfaardt, J., 2008b. An Improved Impact Technique for Monitoring Percutaneous Implant Integrity. International Journal of Oral & Maxillofacial Implants 23(2), 263-269. 23
L Westover et al.
Tjellström, A., Rosenhall, U., Lindström, J., Hallen, O., Albrektsson, T., Brånemark, PI., 1983. Five-year experience with skin-penetrating bone-anchored implants in the temporal bone. Acta Oto-Laryngologica 95(5-6), 568-575.
24
L Westover et al.
FIGURE CAPTIONS LIST Figure 1: ASIST components including the ASIST unit, the handpiece, a USB dongle, and a computer Figure 2: Schematic of the 4-DOF model of the BAHA implant/abutment system (not to scale) Figure 3: Measured accelerometer data (colored) with the curve fit model (red) Figure 4: Matched acceleration signal between the model (black) and measured data (colored) Figure 5: Laboratory test set-up showing the FRB-20 disc containing the implant with a 9 mm abutment attached. The disc is clamped to a platform for testing and the ASIST handpiece is held in a custom stand. Figure 6: Representative example for one Oticon Medical implant sample with three abutment lengths Figure 7: ASC values for all Oticon Medical samples with three abutment lengths (6 mm, 9 mm, 12 mm) Figure 8: Representative example for two different samples of one Oticon Medical implant type (4.50 mm diameter, 4 mm length) and 9 mm abutment Figure 9: One Cochlear implant sample with four different abutment lengths
25
L Westover et al.
TABLES Table 1: Mass and geometry properties for Oticon Medical implants and abutments. Mass (rt ) (u)
Diameter (vwx ) (yy)
Length (zt − z| ) (yy)
0.126
3.75
3.00
0.157
3.75
0.173
~> (yy)
t (u yyv )
4.00
2.71
0.51
4.50
3.00
2.04
0.46
0.222
4.50
4.00
2.54
0.76
Oticon Medical Abutments
Mass (r ) (u)
Shaft Diam. (mm)
Length (z ) (yy)
6 mm 9 mm 12 mm
0.51 0.72 0.97
5.00 5.00 5.00
6.00 9.00 12.00
~> (yy)
(u yyv )
Oticon Medical Implants M50106 (Short/Narrow) M50128 (Long/Narrow) M51061 (Short/Wide) M51062 (Long/Wide)
2.19
2.98 4.52 5.95
0.31
2.72 5.99 13.07
*Note that the location of the center of mass (~>)is measured from the bottom of the component and the mass moment of inertia () is about the center of mass. Table 2: Mass and geometry properties for Cochlear implants and abutments Cochlear Implant BI300 Cochlear Abutments 6 mm 8 mm 10 mm 12 mm
Mass (rt ) (u) 0.22 Mass (r ) (u) 0.50 0.70 1.03 1.40
Diameter (vwx ) (yy) 4.45
Length (zt − z| ) (yy) 4.00 Length (z ) (yy) 4.67 6.67 8.67 10.67
~> (yy) 2.66
t (u yyv )
~> (yy)
(u yyv )
2.15 3.85 4.93 6.01
0.78
2.68 4.42 8.22 14.39
*Note that the location of the center of mass (~>) is measured from the bottom of the component and the mass moment of inertia () is about the center of mass.
26
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6
Figure 7
Figure 8
Figure 9