Anatomical subject validation of an instrumented hammer using machine learning for the classification of osteotomy fracture in rhinoplasty.

Osteotomies during rhinoplasty are usually based on the surgeon's proprioception to determine the number and the strength of the impacts. The aim of this study is to determine whether a hammer instrumented with a force sensor can be used to classify fractures and to determine the location of the osteotome tip. Two lateral osteotomies were realized in nine anatomical subjects using an instrumented hammer recording the evolution of the impact force. Two indicators τ and λ were derived from the signal, and video analysis was used to determine whether the osteotome tip was located in nasal or frontal bone as well as the condition of the bone tissue around the osteotome tip. A machine-learning algorithm was used to predict the condition of bone tissue after each impact. The algorithm was able to predict the condition of the bone after the impacts with an accuracy of 83%, 91%, and 93% when considering a tolerance of 0, 1, and 2 impacts, respectively. Moreover, in nasal bone, the values of τ and λ were significantly lower (p < 10) and higher (p < 10) than in frontal bone, respectively. This study paves the way for the development of the instrumented hammer as a decision support system.