Statistical shape model-based prediction of tibiofemoral cartilage

Abstract

Computed tomography is used more routinely to design patient-specific instrumentation for knee replacement surgery. Its moderate imaging cost and simplified segmentation reduce design costs compared with magnetic resonance (MR) imaging, but it cannot provide the necessary cartilage information. Our method based on statistical shape modelling proved to be successful in predicting tibiofemoral cartilage in leave-one-out experiments. The obtained accuracy of 0.54 mm for femur and 0.49 mm for tibia outperforms the average cartilage thickness distribution and reported inter-observer MR segmentation variability. These results suggest that shape modelling is able to predict tibiofemoral cartilage with sufficient accuracy to design patient-specific instrumentation.

Keywords:

• Statistical shape models
• cartilage
• bone
• tibiofemoral joint

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This work was supported by VLAIO – Agentschap Innoveren & Ondernemen, Koning Albert II-laan 35, bus 16, 1030 Brussel, Belgium [130837].