Mixed-reality Holographic-Assisted Placement of Glenoid Guidewire in Shoulder Arthroplasty: Preliminary Comparison to Patient-Specific Instrumentation in B2 Glenoid Model

Three-dimensionally (3D) printed patient-specific instrumentation (PSI) guides and computer navigation are 2 forms of commercially available options for improved accuracy of glenoid guide pin placement during shoulder arthroplasty. Mixed reality (MR) and virtual reality devices have been used in medical education thus far, but there has been limited assessment of their intraoperative viability in specific settings such as shoulder arthroplasty. The purpose of this study was to compare the accuracy of MR holographic model–assisted glenoid guidewire placement to freehand (FH) and PSI options. Sixty 3D printed glenoid polyurethane sawbones models with a B2 glenoid defect were created. The 3D model of the B2 glenoid with a guide pin in place was programmed into a Unity-based application installed on the Microsoft HoloLens2 MR device. In randomized fashion, 5 surgeons placed a guide pin into the “sawbones” models using FH, PSI, or MR with the attempt to replicate the desired preoperative plan. Using fine-cut computed tomography, average version, average inclination, and starting point were analyzed among all models and surgeons. The average starting point from the plan for the FH group was 2.21 ± 0.95 mm, 2.27 ± 0.80 mm for the PSI group, and 1.745 ± 0.84 mm for the MR group, P = .12. The average inclination was 10.56 ± 7.37 degrees, 3.02 ± 3.36 degrees, and 8.16 ± 5.69 degrees for the FH, PSI, and MRI groups, respectively, P = .0004. The average version was 13.52 ± 9.21 degrees, 7.26 ± 4.64 degrees, and 12.36 ± 9.19 degrees for the FH, PSI, and MRI groups, respectively, P = .04. Subgroup analysis of FH to PSI and PSI to MR demonstrated significantly less inclination from plan in the PSI group, P = .0003 and P = .02, respectively. There were no differences in version in the subgroup analysis between the FH and MR and PSI and MR groups. Degrees of deviation of inclination from plan were significantly less when comparing all models among more experienced surgeons, P = .026, whereas no statistical differences in deviation of start point or version were noted across all models in the more experienced group. When further analyzing the inclination among the FH, PSI, and MR groups among higher level surgeons, the PSI model demonstrated less deviation from plan compared to the FH model, P = .02, but there were no differences in inclination when comparing the PSI to MR and MR to FH models. MR permits accurate glenoid guide pin placement comparable to FH placement in a polyurethane B2 glenoid bone substitute model.