# Interdisciplinary Collaborative Virtual Reality Planning for Chest Wall Resection and Reconstruction for Sarcoma and Other Large Chest Wall Malignancies Enhanced by Automated AI Segmentation: A Retrospective Comparative Analysis

## TL;DR

This study explores how virtual reality with AI helps surgeons plan complex chest wall surgeries, improving collaboration and understanding but showing some limitations in accuracy.

## Contribution

A streamlined workflow using VR and AI segmentation for chest wall surgery planning is introduced and evaluated.

## Key findings

- VR planning exceeded actual resection in 50% of cases, with one case showing a large overestimation.
- User experience scores indicated high hedonic quality and positive usability in VR planning.
- VR planning aids interdisciplinary collaboration but should complement rather than replace traditional methods.

## Abstract

Oncologic chest wall resection and reconstruction present significant surgical challenges due to the complex interplay of anatomical and physiological factors. Ensuring adequate oncologic margins while preserving structural integrity and function is essential for optimal oncological and physiological patient outcomes. Advanced visualization technologies such as virtual reality (VR) are being increasingly investigated for surgical use cases because of their ability to provide a comprehensive and immersive representation of anatomical structures, thereby enhancing preoperative planning and, potentially, intraoperative guidance. The goal of this study is to establish a streamlined workflow using state-of-the-art technology to optimize surgical planning and potentially improve patient outcomes in the complex field of chest wall reconstruction.

Eight cases of complex chest wall resection were retrospectively analyzed using the “Medical Imaging XR” VR platform with AI-driven anatomical auto segmentation. An interdisciplinary team of surgeons collaboratively planned the surgical procedures in VR, and predicted parameters such as resection extent, defect dimensions, and reconstruction strategies. These were then quantitatively compared with actual intraoperative findings. User experience was assessed with the User Experience Questionnaire (UEQ), workspace perception ratings, and Simulator Sickness Questionnaire (SSQ).

In 3 cases (37.5%), the actual resection exceeded the VR-predicted extent due to underestimated tumor infiltration. Planning exceeded resection in 50% of cases by up to 24% and one case (12.5%) showed a large overestimation in VR. UEQ scores showed high hedonic quality (Stimulation = 2.19, Novelty = 2.69) and positive pragmatic usability (Efficiency = 1.13, Dependability = 1.63). Workspace perception was favorable (mean 4.9/6), and cybersickness remained low.

AI-enhanced VR planning enables interdisciplinary collaboration and can improve spatial understanding in complex chest wall surgery. Although it facilitates structured preoperative planning and communication, it should be viewed as a complementary tool to select the surgical strategy rather than as a definitive predictor of the extent of resection. Limitations in imaging resolution and segmentation accuracy can lead to under- or overestimation of tumor boundaries. Further development and clinical validation are necessary to determine its full impact on surgical planning quality and outcomes.