| Sumario: | In many maxillofacial surgeries, the surgeon must perform a pterygomaxillary disjunction in order to separate the jaw from the skull. The disjunction is performed using an osteotome which is impacted with a hammer until its tip crosses through the pterygoid plates. To determine whether the disjunction has occurred and avoid complications, surgeons still rely on their proprioception. The aim of this study is to validate a vibroacoustic set-up constituted by an instrumented hammer equipped with a force sensor, aiming at detecting if the disjunction has occurred. To do so, fourteen osteotomies were performed in anatomical subject jaws. For each impact, the force signal was recorded and analyzed. A classification algorithm (Support Vector Machine method coupled with a cost matrix) was developed based on indicators extracted from the signal to determine whether the disjunction had occurred. Impacts were classified into two classes, one before and the other after the osteotome had crossed the pterygoid plates. To avoid undetected disjunctions a cost matrix was added to the algorithm. The coefficients from the matrix were chosen using by minimizing the Negative Likelihood Ratio. The algorithm was able to distinguish impacts before and after the disjunction with an accuracy of 96 %. Moreover, the instrumented hammer was able to detect the disjunction with a maximum delay of two impacts. These results pave the way for the development of a per-operative decision support system for the pterygomaxillary disjunction. • During maxillofacial surgeries, a disjunction must be performed between two bones. • A vibroacoustic method was used to extract information from the impact signal. • A classification algorithm was applied to information extracted from the signal. • This method can determine whether a disjunction has occurred.
|
|---|