The foot models
The experiments were performed on a foot dummy consisting of a polyamide forefoot skeleton with exchangable first metatarsal bone (MT I) with a soft tissue cover, made of medical silicone (Figure 1).
To build the 3 dimensional forefoot skeleton, a CT scan (Philips Brilliance 40 CT, Philips Healthcare, The Netherlands) from a cadaveric foot with hallux valgus deformity was conducted under axial load of 350 N in a custom made clamping system. The data of the 3 dimensional model of the foot were segmented and implemented in the preoperative planning software CASPA (version 4.18, Balgrist CARD AG, Zurich, Switzerland). The foot showed a moderate hallux valgus deformity with an intermetatarsal angle (IMA) of 15°. In order to obtain three different hallux valgus manifestations (slight, moderate and severe), the MT I was rotated parallel to the plantar plate to an IMA of 18° (severe hallux valgus) and 13° (slight hallux valgus) through the center of the sphere defined by the proximal articular surface of the MT I. A pluggable connector was incorporated in the model between the MT I and the medial cuneiform bone to permit exchange of the MT I after completion of an osteotomy. The polyamide skeleton was printed with 60 replaceable MT I (with 20 slight, 20 moderate and 20 severe hallux valgus manifestations) using selective laser sintering[12].
The soft tissue cover of the foot dummy model was generated by using a plaster cast mold of the forefoot and midfoot of the cadaver. The cast was lined with a dividing layer (Body Double, SmoothOn, Macungie PA, USA) and filled with medical silicon (Elastosil, Wacker, Riemerling, Germany). The polyamide skeleton model was subsequently immerged in the silicone before the silicon had cured.
A dorsomedial approach through the silicon cover to the MT I was performed (AFV and SMZ). Finally, the model was fixed to a wooden frame with 3 dimensional markers attached to it for facilitated alignment of the hologram.
For the augmented reality (AR) experiment, a cutting plane perpendicular to the axis of the second metatarsal bone and perpendicular to the plantar plate was integrated in the model to serve as cutting guide in the experiments (Figure 2). The model for slight, moderate and severe hallux valgus including the cutting plane and the model of the frame and its markers were then transferred to the head mounted device (HMD) (Hololens (Microsoft Corporation, Redmond, WA 98052-6399, USA)).
Experiments
The distal osteotomy of the MT I was performed using an oscillating saw blade (Sodem Systems, Geneva, Switzerland). The cuts were alternately carried out freehand or with an overlaid hologram wearing the HMD (Figure 3). They were performed by a consultant of the foot and ankle department (SMZ), and a resident (LJ) who had no prior experience in hallux valgus surgery at the time of the experiments. The experiments were randomized concerning freehand or AR cuts and the manifestation of the hallux valgus deformity. A total of 30 osteotomies (10 for each degree of deformity) were performed by the two surgeons. In an attempt to avoid adaptation to the setting, a maximum of 6 experiments was set for each surgeon per day.
Analysis
A second skeletal model (Figure 4) was printed for analysis without the first metatarsal but a plane perpendicular to the plantar plane and parallel to the axis of the second metatarsal. The osteotomized first metatarsals of the experiments could be connected to the analysis skeleton model through a plug. A picture of each MT I osteotomy was taken perpendicular to the plantar plate and centered over the distal first metatarsal with a camera (Nikon, Tokyo, Japan). The angle between the axis of the second metatarsal and the osteotomy was measured using the measurement tool in Synedra (Synedra Schweiz AG, Dübendorf, Switzerland). The measurements were carried out by two independent observers (LJ and AFV). The statistical analysis was performed using SPSS (IBM, Zurich, Switzerland). The student’s T test for independent samples was used to determine whether the differences were significant (p<0.05), assuming normal distribution of data. Cronbach's Alpha was used to measure internal consistency (inter observer correlation). A post hoc power analysis was performed with G*Power (Heinrich-Heine-University, Düsseldorf, Germany)
Source of Funding: ResOrtho Foundation, University Hospital Balgrist. Funding source did not play a role in this investigation.