Abstract
Introduction
Cementless femoral hip stems crucially depend on the initial stability to ensure a long survival of the prosthesis. There is only a small margin between obtaining the optimal press fit and a femoral fracture. The incidence of an intraoperative fracture is reported to be as high as 30% for revision surgery. The aim of this study is to assess what information is contained in the acoustic sound produced by the insertion hammer blows and explore whether this information can be used to assess optimal seating and warn for impeding fractures.
Materials and Methods
Acoustic measurements of the stem insertion hammer blows were taken intra-operatively during 7 cementless primary (Wright Profemur Primary) and 2 cementless revision surgeries (Wright Profemur R Revision). All surgeries were carried out by the same experienced surgeon. The sound was recorded using 6 microphones (PCB 130E2), mounted at a distance of approximately 1 meter from the surgical theater. The 7 primary implants were inserted without complication, 1 revision stem induced a fracture distally during the insertion process. Two surgeons were asked to listen independently to the acoustic sounds post-surgery and to label the hits in the signal they would associate with either a fully fixated implant or with a fracture sound. For 3 out of 7 primary measurements the data was labeled the same by the two surgeons, 4 were labeled differently or undecided and both indicated several hits that would be associated with fracture for the fractured revision case. The acquired time signals were processed using a number of time and frequency domain processing techniques.
Results
Figure 1 shows the convergence of a set of time and frequency features (selected temporal moments, decay and 99% energy time [1]) during a primary cementless insertion for which both surgeons labeled hit 12 as the final insertion hit. However, such convergence of the feature set was not as clear for the other 6 cases. Figure 2 shows the result of a feature that tracks the relative weight of low frequency content in the signal relative to the peak power present in the total frequency range for the two revision surgeries. This feature shows several spikes above 0.4 during the case with fractures, whereas none are present for the non-fractured revision case. The spikes concurred with the hits indicated by the surgeon panel post-surgery to have a sound associated with fractures.
Conclusions
Assessment of this initial stability is a challenging task for the surgeon, who mainly has to rely on auditory and sensatory feedback. Although these findings look promising for an early detection and warning for (micro-) fractures, endpoint detection based on acoustic information is more challenging. The difficulty to determine the endpoint based solely on acoustic information was also reflected by the challenge of the surgeon panel to label the acoustic signals post-surgery. Data gathering is currently in progress to extend both the primary and revision set to 15 intra-operative measurements for further validation of these preliminary results.
