Visual needle tip tracking in 2D US guided robotic interventions

Abstract

Percutaneous needle procedures are among the most frequently performed minimally invasive surgical procedure. For tracking the needle tip in the tissue, 2D ultrasound (US) imaging is commonly used; however, the low resolution of the images creates a challenge for tracking. This paper describes a robotic system that can perform US image guided biopsies by tracking the needle and the target simultaneously. It uses a template-based visual tracking method for small and deformable targets. During the experiments, a needle was inserted into realistic phantoms using a 5-DOF robot. The 2D US probe was held by a robotic arm that was servoed along the needle path. The 3D shape of the needle was estimated using the 2D transverse US images, which was used to align the needle axis with the 2D imaging plane. The accuracy of the visual needle tip tracking was evaluated using an optical tracking system, and a computed tomography scanner was used to determine the accuracy of the 3D needle shape estimation method. Target reaching accuracies were measured using an electromagnetic tracking system. The results of the experiments showed that the proposed system can track the needle tip in 2D US guided needle procedures in real-time with a sub-millimeter positional error.