Related Experiment Video
Updated: Jun 15, 2025
Four-Dimensional CT Analysis Using Sequential 3D-3D Registration
Published on: November 23, 2019
Automatic generation of knee kinematic models from medical imaging
Beichen Shi1, Martina Barzan1, Azadeh Nasseri1
1Griffith Centre of Biomedical and Rehabilitation Engineering, Menzies Health Institute Queensland, Gold Coast campus Griffith University QLD 4222, Australia; School of Health Sciences and Social Work, Gold Coast campus Griffith University, Parklands Dr Southport QLD 4222, Australia.
An automated pipeline accurately predicts knee joint kinematics, comparable to manual methods. This approach simplifies the creation of tibiofemoral (TFJ) and patellofemoral (PFJ) models from MRI, reducing processing time.
Area of Science:
- Biomechanics
- Medical Imaging Analysis
- Computational Modeling
Background:
- Three-dimensional spatial mechanisms enable accurate prediction of passive knee kinematics.
- Multibody kinematic models offer anatomical consistency but require extensive medical image processing, limiting their adoption.
- Automating model generation is crucial for wider application.
Purpose of the Study:
- To automate the generation of kinematic models for the tibiofemoral (TFJ) and patellofemoral (PFJ) joints.
- To compare an automated pipeline against a manual pipeline for creating these knee joint models from segmented MRI data.
Main Methods:
- Geometric parameters (articular surfaces, ligament attachments) were extracted from segmented MRI of eight participants using both automatic and manual pipelines.
- TFJ and PFJ kinematic models were assembled using these parameters to predict passive kinematics.
- A Multiple Objective Particle Swarm Optimization was employed to refine geometric parameters for physiological predictions.
Main Results:
- Strong agreement was observed between automatic and manual pipelines for geometric parameters (median error: 2.8 mm landmarks, 1.5 mm ligament lengths).
- Predicted TFJ and PFJ kinematics showed no significant statistical differences between pipelines, with minor exceptions in tibial translation.
- The automatic pipeline demonstrated mean errors of 3.6°/12.4° for TFJ rotation and <6-9° for PFJ rotation, and <7 mm/6 mm for translational errors.
Conclusions:
- The developed automatic pipeline effectively predicts passive knee kinematics comparable to manual methods.
- This automated approach significantly reduces laborious manual processing.
- It offers a systematic and efficient method for creating anatomically consistent knee kinematic models.
