Statistical shape modeling predicts patellar bone geometry to enable stereo-radiographic kinematic tracking.

Area of Science:

• Orthopedic biomechanics
• Medical imaging analysis
• Computational anatomy

Background:

• Patellofemoral (PF) complications like subluxation and dislocation are common after total knee replacement (TKR), often necessitating revision surgery.
• Accurate kinematic assessment of the PF joint is crucial for evaluating implant designs and complications, but is challenging due to radiotransparent patellar implants.

Purpose of the Study:

• To evaluate the efficacy of statistical shape models (SSMs) in predicting subject-specific 3D patellar geometries from 2D radiographic images.
• To establish an optimal data collection methodology for PF kinematics using stereo-radiographic perspectives.

Main Methods:

• Developed a statistical shape model (SSM) for patellar geometry based on 50 subjects.
• Utilized a virtual stereo-radiographic system to simulate 2D image profiles and predict 3D patellar geometries.
• Employed a leave-one-out cross-validation approach to assess prediction accuracy.
• Investigated the impact of different stereo-radiographic imaging perspectives on accuracy.

Main Results:

• The SSM accurately predicted subject-specific 3D implanted patellar geometries with an average error of 0.45±0.07mm, comparable to traditional segmentation methods.
• Imaging the patella in five unique stereo-radiographic perspectives provided the most accurate geometric representation.

Conclusions:

• Statistical shape models (SSMs) offer a viable alternative to CT scans for reconstructing patellar geometry in vivo.
• This approach can significantly reduce radiation exposure and streamline kinematic evaluations of the patellofemoral joint after TKR.
• The findings support the use of SSMs and optimized imaging protocols for improved PF joint kinematic analysis.