A clinical applicable study on lower limb segmentation from CT images for total knee arthroplasty
File(s)
Published version
Author(s)
Type
Journal Article
Abstract
Background
Segmentation of lower limb bones is essential for accurate preoperative planning in total knee arthroplasty (TKA), yet fully annotated CT datasets are rarely available in clinical practice. This study evaluated whether a partially supervised deep learning framework can leverage incompletely annotated CT data to generate anatomically accurate femur and tibia segmentations suitable for TKA planning.
Method
A 3D nnU-Net model was trained using 205 healthy full-leg CT scans with mixed annotation completeness,
including 17 fully annotated cases and partially labelled femur or tibia in the remaining scans. Performance was
evaluated on an internal healthy dataset (n = 40), a cadaveric dataset (n = 15), and an osteoarthritis (OA) dataset acquired for robotic TKA planning (n = 10). Accuracy was assessed using Dice similarity coefficient (DSC), Hausdorff distance (HD), HD95, and root-mean-square surface distance (RMSE). Clinical relevance was
evaluated using landmark localisation errors and joint alignment measurements (mLDFA and mPTA).
Results
On the cadaveric dataset, mean DSC values were 96.53% (femur) and 97.41% (tibia), with RMSE < 1 mm. On
the OA dataset, mean DSC remained approximately 96.5% with HD95 < 1.7 mm across acquisition windows.
Alignment measurements derived from automatic and manual segmentations showed small differences
(mLDFA: 86.81° vs 87.02°, p = 0.11; mPTA: 86.74° vs 87.29°, p = 0.08), with no statistically significant
differences observed.
Conclusion
Partially supervised training enables accurate lower limb segmentation from incompletely annotated CT datasets
and preserves clinically relevant alignment measurements required for TKA planning.
Segmentation of lower limb bones is essential for accurate preoperative planning in total knee arthroplasty (TKA), yet fully annotated CT datasets are rarely available in clinical practice. This study evaluated whether a partially supervised deep learning framework can leverage incompletely annotated CT data to generate anatomically accurate femur and tibia segmentations suitable for TKA planning.
Method
A 3D nnU-Net model was trained using 205 healthy full-leg CT scans with mixed annotation completeness,
including 17 fully annotated cases and partially labelled femur or tibia in the remaining scans. Performance was
evaluated on an internal healthy dataset (n = 40), a cadaveric dataset (n = 15), and an osteoarthritis (OA) dataset acquired for robotic TKA planning (n = 10). Accuracy was assessed using Dice similarity coefficient (DSC), Hausdorff distance (HD), HD95, and root-mean-square surface distance (RMSE). Clinical relevance was
evaluated using landmark localisation errors and joint alignment measurements (mLDFA and mPTA).
Results
On the cadaveric dataset, mean DSC values were 96.53% (femur) and 97.41% (tibia), with RMSE < 1 mm. On
the OA dataset, mean DSC remained approximately 96.5% with HD95 < 1.7 mm across acquisition windows.
Alignment measurements derived from automatic and manual segmentations showed small differences
(mLDFA: 86.81° vs 87.02°, p = 0.11; mPTA: 86.74° vs 87.29°, p = 0.08), with no statistically significant
differences observed.
Conclusion
Partially supervised training enables accurate lower limb segmentation from incompletely annotated CT datasets
and preserves clinically relevant alignment measurements required for TKA planning.
Date Issued
2026-10-01
Date Acceptance
2026-05-24
Citation
Knee, 2026, 62
ISSN
0968-0160
Publisher
Elsevier
Journal / Book Title
Knee
Volume
62
Copyright Statement
Crown Copyright © 2026 Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
License URL
Identifier
10.1016/j.knee.2026.104526
Publication Status
Published
Article Number
ARTN 104526
Date Publish Online
2026-06-06