# Assessing the accuracy of 3D assistive technologies for surgical guidance of osteosarcoma resections: a comparative laboratory study of mixed reality, patient-specific instruments and freehand approaches

## TL;DR

This study compares the accuracy of mixed reality, patient-specific instruments, and freehand techniques for osteosarcoma resection in 3D-printed models to determine which method allows for the narrowest surgical margins.

## Contribution

The study introduces a non-clinical comparison of mixed reality and patient-specific instrumentation for osteosarcoma resection accuracy.

## Key findings

- Freehand resection had significantly higher cutting inaccuracy compared to mixed reality and patient-specific instruments.
- Mixed reality and patient-specific instruments showed similar accuracy and required smaller surgical margins than freehand techniques.
- Estimated margins for 99% tumor-free cuts were 15.9 mm (freehand), 6.2 mm (patient-specific), and 8.6 mm (mixed reality).

## Abstract

The survival rate after surgical osteosarcoma resection is low, particularly when the sarcoma is not fully removed. Therefore, wide surgical margins are used in surgery, limiting how much bone can be salvaged. Patient-specific instrumentation (PSI) enables smaller margins, but utilization is low. Mixed reality-based techniques (MR) might be easier to implement. The purpose of this study was to compare the cutting accuracy of MR, PSI, and freehand techniques in 3D-printed osteosarcoma models and determine the corresponding technique-related minimal surgical margins.

CT-scans of patients with extremity osteosarcoma were acquired, segmented, and the bones 3D-printed three times. Scans were excluded if they had low resolution or metal artifacts. Pre-surgical planning for full resection was conducted, and corresponding PSI and MR plans were created. Tumor resections were separately done via a freehand, PSI, and MR approach. Resected bone models were 3D scanned, and the cutting accuracy was determined. Differences in accuracy were determined via Bartlett’s test and corresponding post-hoc tests for a significance level of 0.05. The techniques’ surgical margins were determined for 90, 95, 97.5, and 99% successful cuts.

Ten osteosarcomas with one to three cutting planes were included, leading to a total of 19 cuts. The variance in cut accuracy was significantly greater for the freehand approach (standard deviation (STD) [95%CI]: 6.85 [5.18–10.13] mm) than the MR (STD: 3.71 [2.79–5.57] mm) and the PSI (STD: 2.68 [2.02–3.96] mm) approach. No significant difference in variance between PSI and MR was found (P = 0.44). Surgical margins varied between techniques, with the freehand approach being about twice those of the MR and PSI approaches. To achieve 99% tumor-free cuts, the estimated required margins were 15.9 mm (freehand), 6.2 mm (PSI), and 8.6 mm (MR).

This study acts as a non-clinical proof of concept that the adoption of patient-specific instrumentation or mixed reality techniques for osteosarcoma resection might enable narrower margins pending in-vivo validation, potentially enabling bone and joint preservation and restoration, while decreasing resection failure rates.

The online version contains supplementary material available at 10.1186/s42836-026-00369-8.