3D CT-based pre-operative planning and intraoperative navigation for glenoid component fixation in reverse shoulder arthroplasty: an in vivo impact study

Reverse shoulder arthroplasty (RSA) is an established surgical option for the treatment of glenohumeral osteoarthritis associated with an irreparable rotator cuff tear. However, procedural success hinges largely on the appropriate management of glenoid bone loss, a key determinant for optimizing implant stability and clinical and functional outcomes. Within this framework, computer-assisted technologies—including 3D CT-based preoperative planning, intraoperative navigation, and, more recently, artificial intelligence (AI) systems—have assumed an increasingly prominent role. This doctoral thesis integrates four complementary studies to assess the role of these emerging technologies—3D CT planning, intraoperative navigation, and AI—across the decision-making and therapeutic pathway of RSA. The first study (Chapter 6) investigates the interobserver reliability of preoperative planning in 81 cases assessed by nine experienced Italian surgeons. The findings demonstrate substantial individual variability (single-measure ICC 0.13–0.33; Fleiss’ κ −0.093 to 0.325), balanced by reliable collective consensus (average-measure ICC 0.58–0.82) and rational, defect-specific decision patterns (Kruskal–Wallis H = 48.079, p < 0.001). Shared planning strategies emerge, including the preferential use of posterior augments in cases of retroversion (55%) and superior augments for inclination defects (68%). The second study (Chapter 7) evaluates the validity of the AI-based Predict® software in a cohort of 10 patients. Aggregate predictions show good accuracy (MAE VAS 1.4 < MCID), but limited patient-specific performance (ρ = 0.15–0.33), suggesting current constraints in achieving true personalization. The third study (Chapter 8) quantifies the clinical impact of preoperative planning and intraoperative navigation in 158 consecutive cases performed by a single surgeon, documenting an 83% increase in the use of augmented baseplates, the use of longer screws (+25%; 36.5 vs 29.1 mm), and a reduction in the mean number of screws (−13%; 2.1 vs 2.4), thereby supporting the standardization of surgical execution. Finally, the fourth study (Chapter 9) compares mid-term clinical outcomes (mean follow-up 42 months) between 80 patients treated with navigated versus conventional techniques, showing comparable functional results (Constant 67±16; DASH 20±19; p > 0.05) and confirming the non-inferiority of navigation relative to the traditional approach. Taken together, these studies indicate that advanced technologies in RSA do not eliminate decision-making variability, but rather render its underlying mechanisms quantifiable, facilitating coherent and reproducible corrective strategies. Three-dimensional planning helps identify shared biomechanical principles in the management of glenoid bone loss, while intraoperative navigation contributes to standardizing surgical execution without compromising clinical outcomes. AI-based systems appear promising as an aggregate decision-support tool, although further development is required to achieve robust patient-specific accuracy. Overall, these technologies should be regarded as complementary tools that—when appropriately integrated—may improve the quality of the decision-making process and provide a solid foundation for the future evolution of RSA toward more predictive, standardized, and personalized care.