Out-Bore MRI-Guided Breast Biopsy Employing Machine Vision and Augmented Reality Techniques: Preliminary Study

Abstract

Magnetic resonance imaging (MRI) of the breast demonstrates superior sensitivity and specificity in distinguishing cancerous tissue. Despite the availability of in-bore MRI-guided breast biopsy, its application is restricted to a limited subset of individuals. We present a novel methodology developed for out-bore MRI-guided breast biopsy. Our approach integrates three primary functionalities: real-time pose tracking of the biopsy needle and a head-mounted display (HMD), real-time shape sensing of the breast with deformable registration, and augmented reality (AR)-assisted visualization of the MRI target and navigational information. Comprehensive experiments were performed to investigate the core performance metrics of the method, including targeting accuracy, update rate of the deformable registration, and data refresh rate and latency of the AR visualization. The mean targeting accuracy was 4.55 ± 1.11mm. The average update rate for deformable registration was 42.76±13.96ms. Furthermore, the data refresh rate and latency associated with AR visualization were recorded as 59 ± 23fps and 192 ± 73ms, respectively. The experimental findings underscore the system’s promising potential in facilitating successful outcomes for out-bore MRI-guided breast biopsy.