A Novel System for Validating 3D Mapping for Patients with Rotator Cuff Arthropathy
Ashley Isenberg, Aman Chopra, Eliana S. LeMaster, Anand M. Murthi, Melissa A. Wright, Tyler HwangIntroduction
Accurate assessment of glenoid morphology is critical for preoperative planning in reverse total shoulder arthroplasty (rTSA) for rotator cuff arthropathy (RCA). While three-dimensional (3D) CT-based software improves precision, the accuracy of physical 3D-printed models compared with digital reconstructions remains under-investigated. This study compared glenoid version and inclination measurements from 3D-printed models against CT-based commercial planning software and evaluated regional variation across glenoid quadrants.
Methods
Thirty-nine patients with RCA who underwent primary rTSA between 2019 and 2022 were screened. CT scans were processed using commercial software (BluePrint, Stryker) to measure version and inclination. Corresponding 3D scapular models were generated using open-source reconstruction software (OsiriX, Blender) and printed on a consumer-grade 3D printer. Fifteen models met quality thresholds for inclusion. Anatomic landmarks were digitized using a MicroScribe system, and version and inclination were calculated across the entire glenoid face and four quadrants (anterosuperior, anteroinferior, posterosuperior, posteroinferior) using MATLAB. Quadrants were defined by bisecting the glenoid along its superoinferior and anteroposterior axes through the geometric center. Pearson correlation coefficients and Bland-Altman analysis were used to assess agreement between 3D-printed model and BluePrint measurements.
Results
Of the original 39 patients screened, 15 (38%) met quality thresholds for inclusion after CT-to-model conversion. Mean 3D-printed model retroversion was 16.22° ± 11.26° and inclination was 8.91° ± 7.26°. Mean BluePrint retroversion was 11.80° ± 11.40° and inclination was 7.53° ± 8.77°. Strong correlations were found for version (r = 0.95, p 0.001) and inclination (r = 0.89, p 0.001). Bland-Altman analysis demonstrated a mean bias of 4.41° for version and 1.43° for inclination. Quadrant analysis revealed consistent retroversion across all quadrants; however, superior quadrants demonstrated inferior inclination while inferior quadrants demonstrated superior inclination (p 0.001).
Discussion
Glenoid measurements on 3D-printed models strongly correlate with CT-based planning software. The mean version bias of 4.41° falls within the range of inter-method variability reported in the literature, though its clinical relevance for baseplate positioning warrants consideration. Regional variation in inclination across quadrants suggests that global measurements may not fully capture localized deformity patterns in RCA, which may have implications for baseplate positioning.