Touching the tumor boundary: a pilot study on ultrasound-based virtual fixtures for breast-conserving surgery

Laura Connolly, Tamas Ungi, Adnan Munawar, Anton Deguet, Chris Yeung, Russell H. Taylor, Parvin Mousavi, Gabor Fichtinger, Keyvan Hashtrudi-ZaadInt. J. Comput. Assist. Radiol. Surg 2025 · DOI

TL;DR

This paper introduces a cooperative robotic guidance system that uses haptic feedback to help surgeons precisely identify and resect tumor boundaries during breast-conserving surgery (BCS). By imposing a "virtual fixture" that resists tool movement when it nears the tumor, the system aims to reduce positive margins and improve surgical outcomes, showing promising results in simulated resections.

What Problem Does This Paper Address?

Delineating tumor boundaries accurately during breast-conserving surgery (BCS) is a critical yet challenging task. Tumors can be difficult to feel (non-palpable), shift during surgery, and have irregular shapes, leading to positive margins (tumor cells left behind) in 10-40% of cases. This often necessitates reoperation, increasing costs, complications, and patient distress. While existing navigation systems use ultrasound and electromagnetic tracking to visualize tumor location, they rely on the surgeon's visual interpretation, which can be mentally demanding. This paper addresses the need for a more intuitive and precise method to guide surgeons, specifically by leveraging haptic feedback to create a "virtual boundary" around the tumor.

Methods

The core of this work is a cooperative robotic guidance system that integrates an electromagnetic (EM) navigation system, an ultrasound probe, and a haptic device (specifically, a modified Phantom Omni). The system uses ultrasound to delineate tumor boundaries, creating a 3D model. A "forbidden region" virtual fixture is then imposed on this model. When the surgical tool (a modified electrocautery blade attached to the haptic device) approaches or collides with this virtual boundary, the haptic device provides resistance, guiding the surgeon to stay within the desired resection area. The system is built using open-source software like AMBF for simulation and 3D Slicer for visualization and navigation. Collision detection and haptic feedback run at a high rate (1000 Hz) for responsiveness.

Datasets

Not discussed in the paper. The study uses custom-made gel wax breast simulants with embedded tumors for evaluation.

Key Insights and Ideas

  • Haptic feedback as a "virtual boundary": The concept of using haptic resistance to create an intuitive, physical boundary around the tumor is a key innovation, moving beyond purely visual guidance.
  • Reduced cognitive load: Users reported feeling less mentally demanding, frustrated, and effortful when using the haptic guidance, suggesting it simplifies the task.
  • Trade-off between time and margin: While haptic guidance improved margin accuracy, it did increase resection time, highlighting a design consideration for clinical adoption.
  • Importance of training: The study suggests that with adequate training, the initial increase in resection time might diminish, making the system more practical.
  • Simulant variability: The use of gel wax simulants with inherent variability in tumor visibility and shape mimics real-world clinical challenges, making the evaluation more realistic.

Key Results

  • Improved Resection Margins: Virtual fixture guidance generally led to fewer positive margins compared to visual guidance alone. Specific users showed significantly cleaner resection cavities with haptic feedback.
  • Reduced Subjective Load: On average, users reported lower mental demand, temporal demand, effort, and frustration when using VF guidance.
  • Increased Confidence: Users felt more confident in their performance when using VF guidance.
  • Increased Resection Time: Resection time was generally longer (approximately 13.8 mins with VF vs. 9.5 mins without in training rounds) when using VF guidance, attributed to more careful palpation.
  • Increased Volume Removed: The percentage of total volume removed was higher with VF guidance, indicating more complete resections.

Limitations and Future Work

  • Simulant Limitations: The gel wax simulants, while useful, had some limitations in accurately representing tumor visibility and shape, and the tumor could expand after pouring. Future work will explore AI segmentation models to improve boundary delineation.
  • Workspace Limitations: The haptic device used has a limited workspace, which may not be representative of clinical robotic systems. Future work will involve larger workspace robots.
  • Freehand Baseline: The study did not include a freehand baseline comparison, which will be considered in future evaluations to fully assess the impact of the VF system.
  • Further Refinement: Future work will focus on optimizing VF design, enhancing haptic transitions, and developing comprehensive training protocols to fully integrate the system into BCS practices. The goal is to ensure surgeons can intuitively "feel" the tumor boundary.

Summary updated

BibTeX
@article{connolly2025touching,
  author = {Laura Connolly and Tamas Ungi and Adnan Munawar and Anton Deguet and Chris Yeung and Russell H. Taylor and Parvin Mousavi and Gabor Fichtinger and Keyvan Hashtrudi-Zaad},
  title = {{Touching the tumor boundary: a pilot study on ultrasound-based virtual fixtures for breast-conserving surgery}},
  journal = {Int. J. Comput. Assist. Radiol. Surg},
  year = {2025},
  doi = {10.1007/s11548-025-03342-z},
}
Figure