Statistical shape models (SSM) have become a common tool to create reference models for design input and verification of total joint implants. In a recent discussion paper around Artificial Intelligence and Machine Learning, the FDA emphasizes the importance of independent test data [1]. A leave-one-out test is a standard way to evaluate the generalization ability of an SSM [2]; however, this test does not fulfill the independence requirement of the FDA. In this study, we constructed an SSM of the knee (femur and tibia). Next to the standard leave-one-out validation, we used an independent test set of patients from a different geographical region than the patients used to build the SSM. We assessed the ability of the SSM to predict the shapes of knees in this independent test set. A dataset of 82 computed tomography (CT) scans of Caucasian patients (42 male, 40 female) from 11 different geographic locations in France, Germany, Austria, Italy and Australia were used as training set to make an SSM of the femur and tibia. A leave-one-out test was performed to assess the ability of the SSM to predict shapes within the training set. A test dataset of 4 CT scans of Caucasian patients from Russia were used for the validation. The SSM was fitted onto each of the femur and tibia shapes and the root mean square error (RMSE) was measured.INTRODUCTION
METHODS
A good anatomic fit of a Total Knee Arthroplasty is crucial to a good clinical outcome. The big variability of anatomies in the Asian and Caucasian populations makes it very challenging to define a design that optimally fits both populations. Statistical Shape Models (SSMs) are a valuable tool to represent the morphology of a population. The question is how to use this tool in practice to evaluate the morphologic fit of modern knee designs. The goal of our study was to define a set of bone geometries based on SSMs that well represent both the Caucasian and the Asian populations. A Statistical Shape Model (SSM) was built and validated for each population: the Caucasian Model is based on 120 CT scans from Russian, French, German and Australian patients. The Asian Model is based on 80 CT scans from Japanese and Chinese patients. We defined 7 Caucasian and 5 Asian bone models by using mode 1 of the SSM. We measured the antero-posterior (AP) and medio-lateral (ML) dimensions of the distal femur on all anatomies (input models and generated models) to check that those bone models well represent the studied population. In order to cover the whole population, 10 additional bone models were generated by using an optimization algorithm. First, a combined Asian-Caucasian SSM was generated of 92 patients, equally balanced between male and female, Caucasian and Asian. 10 AP/ML dimensions were defined to obtain a good coverage of the population. For a given AP/ML dimension, Markov chain Monte Carlo sampler was used to find the most average shape with AP/ML dimensions as close as possible to the target dimensions. The difference of the AP/ML dimensions of the generated models to the target dimensions was computed. A chi-squared distribution was used to assess how average the resulting shapes were compared to typical patient shapes.Introduction
Methods
