Prediction of premorbid three-dimensional anatomy of the glenoid based on statistical shape modeling

Abstract

IntroductionRestoration of an anatomic joint line after anatomic total shoulder arthroplasty (aTSA) and of the optimal lateral offset after reverse total shoulder arthroplasty (rTSA) may be relatively straightforward when the glenoid does not present with severe erosion. However, in cases of severe glenoid bone loss, the surgeon is left with no preoperative landmark to restore these parameters. The objective of this study was to use statistical shape modeling (SSM), to predict the premorbid morphology of the glenoid. We hypothesized that this would allow us to accurately determine premorbid glenoid version and inclination, in addition to accurately quantifying bone loss and medialization. MethodsFifty-six bilateral CT scans of the shoulders of patients scheduled for shoulder arthroplasty and determined to have unilateral osteoarthritis (primary osteoarthritis or cuff tear arthropathy with a healthy contralateral side) were obtained. A statistical shape model was automatically applied on the pathologic arthritic side to predict its premorbid anatomy. Glenoid version, inclination, height, width and glenoid and scapula lateral offset were measured automatically. These measurements were obtained on the pathological arthritic cases, on the contralateral control healthy cases and on the premorbid predictions of the pathological arthritic cases and were compared pair by pair. ResultsThe mean difference between the pathological arthritic side and the contralateral healthy side was 9.1° ± 7.3° for version, 4.8° ± 4.8° for inclination, 4.9 ± 4.5 mm for height, 4.7 ± 5.3 mm for width, 2.4 ± 1.9 mm for scapula lateral offset, and the glenoid lateral offset was 1.5 ± 1.5 mm. The mean difference between the premorbid prediction of the pathological side and the contralateral healthy side was reduced to 3.3° ± 2.4° for version, 3.4° ± 2.6° for inclination, 3.0 ± 1.9 mm for height, 2.3 ± 1.4 mm for width, 2.2 ± 1.7 mm for scapula lateral offset, and the glenoid lateral offset was 0.9 ± 0.8 mm. ConclusionThis study shows that statistical shape modeling (SSM) can allow accurate prediction of the premorbid morphology of the glenoid. This could help optimize implant selection and positioning after aTSA and rTSA in order to restore optimal soft-tissue tension.