Needle Tracking and Deflection Prediction for Robot-Assisted Needle Insertion Using 2D Ultrasound Images

In many types of percutaneous needle insertion surgeries, tissue deformation and needle deflection can create significant difficulties for accurate needle placement. In this paper, we present a method for automatic needle tracking in 2D ultrasound (US) images, which is used in a needle–tissue interaction model to estimate current and future needle tip deflection. This is demonstrated using a semi-automatic needle steering system. The US probe can be controlled to follow the needle tip or it can be stopped at an appropriate position to avoid tissue deformation of the target area. US images are used to fully parameterize the needle-tissue model. Once the needle deflection reaches a pre-determined threshold, the robot rotates the needle to correct the tip’s trajectory. Experimental results show that the final needle tip deflection can be estimated with average accuracies between 0.7mm and 1.0mm for insertions with and without rotation. The proposed method provides surgeons with improved US feedback of the needle tip deflection and minimizes the motion of the US probe to reduce tissue deformation of the target area.