Augmented-Reality Based Digital Twin to Control an MR Safe Robot for Breast Biopsy: A Benchmark Study

This is the authors' abstract. We don't add key points for this paper.

Magnetic resonance (MR) guided breast biopsies can be demanding and time-consuming procedures. Inaccurate needle insertion can yield false negatives, potentially leading to inaccurate diagnoses with severe consequences. Incorporating robots into these procedures has the potential to improve accuracy, alleviate patient strain, and reduce the time and cost of this procedure. A control interface is essential for surgeons to maintain control and make necessary adjustments. In this study, an augmented reality (AR)-based control method for surgeons is proposed by offering a new means of interacting with robots with enhanced intuition. A virtual surrogate of the MR safe robot in the AR scene was developed, which represents the digital embodiment of the actual robot in the physical world. Users can interact with the virtual surrogate robot to control the actual robot that is positioned in the MR bore while observing the 3D reconstruction of the clinical environment. Two AR-based interactive approaches (gesture-based control and auto-targeting) were developed and compared with joystick control. A pseudo-randomized study including ten subjects was conducted. Both gesture-based (2.24 mm) and auto-targeting (2.55 mm) control methods showed higher median accuracy than the joysticks (3.20 mm). This AR-based digital surrogate robot demonstrated the capabilities of coupling kinematics and motions of the physical MR safe robot for tasks in needle positioning and injection. It shows great potential to facilitate the clinical applications of MR safe robots in the future, beyond the MR-guided breast biopsy.