Résumé
Background: This study proposes a method for inferring the premorbid glenoid shape and orientation of scapulae affected by glenohumeral osteoarthritis (OA) to inform restorative surgery.
Methods: A statistical shape model (SSM) built from 64 healthy scapulae was used to reconstruct the premorbid glenoid shape based on anatomic features that are considered unaffected by OA. First. the method was validated on healthy scapulae by quantifying the accuracy of the predicted shape in terms of surface distance, glenoid version, and inclination. The SSM-based reconstruction was then applied to 30 OA scapulae. Glenoid version and inclination were measured fully automatically and compared between the original OA glenoids, SSM-based glenoid reconstructions, and healthy scapulae.
Results: Validation on healthy scapulae showed a root-mean-square surface distance between original and predicted glenoids of 1.0 +/- 0.2 mm. The prediction error was 2.3 degrees +/- 1.8 degrees for glenoid version and 2.1 degrees +/- 2.0 degrees for inclination. When applied to an OA dataset. SSM-based reconstruction restored average glenoid version and inclination to values similar to the healthy situation. No differences were observed between average orientation values measured on SSM-based reconstructed and healthy scapulae (P >= 10). However, the average orientation of the reconstructed premorbid glenoid differed from the average orientation of OA glenoids for Walch classes Al (version) and 132 (version, inclination, and medialization).
Conclusion: The proposed SSM can predict the premorbid glenoid cavity of healthy scapulae with millimeter accuracy. This technique has the potential to reconstruct the premorbid glenoid cavity shape, as it was prior to OA, and thus to guide the positioning of glenoid implants in total shoulder arthmplasty. (C) 2018 Journal of Shoulder and Elbow Surgery Board of Trustees. All rights reserved.
Methods: A statistical shape model (SSM) built from 64 healthy scapulae was used to reconstruct the premorbid glenoid shape based on anatomic features that are considered unaffected by OA. First. the method was validated on healthy scapulae by quantifying the accuracy of the predicted shape in terms of surface distance, glenoid version, and inclination. The SSM-based reconstruction was then applied to 30 OA scapulae. Glenoid version and inclination were measured fully automatically and compared between the original OA glenoids, SSM-based glenoid reconstructions, and healthy scapulae.
Results: Validation on healthy scapulae showed a root-mean-square surface distance between original and predicted glenoids of 1.0 +/- 0.2 mm. The prediction error was 2.3 degrees +/- 1.8 degrees for glenoid version and 2.1 degrees +/- 2.0 degrees for inclination. When applied to an OA dataset. SSM-based reconstruction restored average glenoid version and inclination to values similar to the healthy situation. No differences were observed between average orientation values measured on SSM-based reconstructed and healthy scapulae (P >= 10). However, the average orientation of the reconstructed premorbid glenoid differed from the average orientation of OA glenoids for Walch classes Al (version) and 132 (version, inclination, and medialization).
Conclusion: The proposed SSM can predict the premorbid glenoid cavity of healthy scapulae with millimeter accuracy. This technique has the potential to reconstruct the premorbid glenoid cavity shape, as it was prior to OA, and thus to guide the positioning of glenoid implants in total shoulder arthmplasty. (C) 2018 Journal of Shoulder and Elbow Surgery Board of Trustees. All rights reserved.