Publication Date

8-2022

Date of Final Oral Examination (Defense)

4-25-2022

Type of Culminating Activity

Thesis

Degree Title

Master of Science in Mechanical Engineering

Department

Mechanical and Biomechanical Engineering

Supervisory Committee Chair

Clare Fitzpatrick, Ph.D.

Supervisory Committee Member

Mahmood Mamivand, Ph.D.

Supervisory Committee Member

Trevor Lujan, Ph.D.

Supervisory Committee Member

Nathan L. Grimm, M.D.

Abstract

Introduction

Lateral dislocation of the patella is a common injury in active adolescents and young adults. Patients who are ultimately managed surgically have a significantly lower risk of recurrent dislocation. However, determining the optimal surgical treatment remains a challenge, with patients sometimes undergoing multiple surgeries prior to successful stabilization. The aim of this study is to computationally evaluate patients that have undergone multiple surgeries to correct for recurrent lateral patellar dislocation and predict their clinical outcome.

Methods

Our patient cohort consisted of 16 patients with patella dislocation. Patient-specific imaging were used to create three-dimensional (3D) finite element (FE) models of the knee joint and evaluate patellofemoral (PF) stability at multiple time points pre- and post-surgery for each patient. We applied these models to predict the clinical success or failure of each surgery. Specifically, the FE model simulated a knee extension activity while a tibia external torsion, a recognized cause of patellofemoral pain and instability, was applied to assess PF stability. A healthy control group of 12 participants was also included to assess the ability of the model to identify successful outcomes. In addition, five anatomic factors of risk were measured, and statistical analysis was performed to establish if significant differences exist among pre-surgery, post-surgery and healthy control groups. Lastly, a logistic regression model was implemented, trained with anatomic values, and used to classify subjects into likelihood of dislocation categories in order to differentiate between successful and unsuccessful surgical outcomes. Feature scaling and feature combination (namely, principal component analysis (PCA)) was applied to improve the predictive performance of the regression model.

Results

Of 12 control participants, 12 pre-surgery subjects (8 patients after an initial unsuccessful MRPLR and 4 without any), and 9 post-surgery subjects (5 after a successful trochleoplasty and 4 patients after MPFLR), the FE model correctly predicted 29 out of 33 surgery outcomes (87.9% accuracy). Post-surgery simulations predicted patellofemoral stability metrics similar to the healthy control group. Particularly, post-trochleoplasty subjects were associated with an increased ability to provide constraint force on the patella lateral facet, and a lower involvement of the medial patellofemoral ligament, particularly close to full extension. A one-way ANOVA showed that four out of five anatomic factors were significantly different between the pre-surgery and the control group, and three of them also between the pre- and post- surgery group, suggesting that the surgery was able to restore a physiological condition. Lastly, logistic regression classification performance demonstrated 72.2% and 78.9% accuracy before and after PCA, respectively.

Conclusion

The overall aim of this study is to provide surgeons with a useful and validated computational tool that can predict the likelihood of patellar dislocation and differentiate, prior to clinical intervention, between a successful versus unsuccessful surgery, to determine the optimal treatment pathways for individual patients. Preliminary results are promising, but an improvement of the model and a larger clinical dataset are necessary to improve accuracy and comprehensively validate model performance.

DOI

https://doi.org/10.18122/td.1966.boisestate

Download

Share

COinS