Development of an automatic workflow for stochastic risk analysis of cementless hip stems

Total Hip Arthroplasty (THA) is a life-changing surgical intervention that consists in the substitution of the hip joint with an artificial prosthesis, as a solution for proximal femur fractures and pain or movement restriction in patients with end-stage hip diseases. Over 1 million of THA surgeries are performed each year worldwide, and this number will increase due to the population ageing and the broadening of the indications for this procedure, including also younger and more active patients. The increased number of THA procedures performed, associated with an extended life expectancy, will result in a higher number of THA failures and consequent revisions, usually due to abductor insufficiency, nerve injury, dislocations, aseptic loosening, infections, and intraoperative and postoperative periprosthetic fractures. Reducing THA failure rate is crucial in order to reduce patient’s discomfort, length of hospitalizations and costs for the healthcare system.?? As demonstrated in literature, Finite Element Analysis (FEA) is recognized as a promising tool to simulate in silico the insertion of the prosthesis components during THA in order to predict implant failure and reduce complications. The aim of this work, conducted at the Medical Technology Laboratory of the Rizzoli Orthopaedic Institute in Bologna, was to develop an automatic workflow able to generate finite-element models for the insertion of cementless femoral component during THA, useful to evaluate wear, primary and secondary stability of the implant, and the risk of periprosthetic femoral fractures. This pipeline, with respect to the manual workflow, has the advantage of being faster and less operator-dependent; moreover, it enables to model the prosthesis insertion taking account possible surgeon’s positioning or sizing errors, thus allowing the automatic modelling of the insertion in different pose.?? The algorithm at the base of the automated workflow was initially tested using an open-source model of femoral stem with a simple morphology for a preliminary analysis. The insertion of this prosthesis was modelled for 13 different femurs, for each of which 30 different stem poses were stochastically generated through Inverse Latin Hypercube Sampling, perturbing the ideal position established by the surgeon during pre-surgical planning.?? Successively, the procedure was applied to the more elaborated Avenir femoral stem by Zimmer Biomet. In both cases the obtained results are satisfying: the automated workflow is executed in much less time than the manual one and with reduced human error, and the automated procedure allows to analyse more easily how implantation outcomes vary in dependence on some variables of interest. The workflow is robust, since it is not influenced by the prosthesis model or by the femoral morphology, and it is also very adaptable to the different purposes of the simulations. Hence, the automated workflow developed in this work could be used for the design of new or custom models of prosthesis, for which meticulous test procedures are required, in particular after the approval of the Medical Device Reporting regulation. Furthermore, the workflow can be extended in order to perform a Monte Carlo analysis with the aim of evaluating the consequences of the prosthesis’ insertion by varying the variables of influence (e.g., prosthesis’ pose and size, degree of osteoporosis of the patient).??