Abstract
Objectives
Implant loosening is the most common reason for revision of total or partial knee replacement, but the patient complains of pain-not a loose implant. It would be a useful diagnostic tool to interrogate the implant to ascertain whether it remains well fixed or not, thus either confirming or eliminating this mode of failure. For such technology to be adopted by manufacturers, it must be extremely low cost and simple to build into an implant. We aim to develop a sensor that meets these requirements and, when embedded in an implant, can provide information on its fixation to the underlying bone. We have previously proven that, through impedance analysis of passive piezoelectric sensors, it is possible for such sensors to determine the cured state of cement with good correlation (0.7) to a surgeon's judgement (Darton et al, 2014). In this study we now look at how the impedance trances of the sensors can be interpreted to distinguish between tibial trays that are securely cemented in sawbone blocks and those with no cement in loose fitting sawbone blocks.
Method
Small piezoelectric sensors (12 mm diameter, 0.6 mm thickness) were attached using ethyl cyanoacrylate to the top of a small metal tibial tray analogue and wired to an Impedance Analyzer (AEA Technology Inc). The sensor was swept with an alternating current between 100KHz and 400KHz. Three readings were taken using a custom-built code in MATLAB and an average impedance trace was calculated. A pre-calibrated servo-mechanical testing machine (Instron) was used to carry out a pull-out test of the tray from the sawbone block. The force required to completely disengage the tray was recorded. The same tibial tray was then cemented to the same sawbone block using PMMA. Once cured, the same impedance readings were taken before a pull out test was performed on the cemented case. This was repeated on 6 different sawbone blocks
The impedance plots were differentiated to exaggerate the jagged nature of the impedance trace, representative of multiple modes of vibration following which the mean of their differential values was calculated. The average pull out force for cemented trays was approximately 20 times greater than the un-cemented.
Results
Qualitatively, the graph in Figure 1 shows a distinct difference between mean differentiated impedance values for cemented and uncemented trays. This is quantified with paired t-tests that suggest a significant difference between the two bond situations (P«0.01).
Conclusion
Our sensors show a clear difference between trays fixed securely in place and those poorly fixed. Further work is being carried out to measure the breakdown in the mechanical interlock under simulated physiological loading. A distinct advantage of this method is that it only requires the impedance of the sensor to be determined; hence the circuitry is simply the sensor attached to a coil, the impedance characteristics of which can be interrogated through established induction methods. This technology has the potential to be extremely cheap, easy to implement and compact.
