Abstract
In needle steering, estimating the needle pose is a critical problem. In 3D ultrasound volumes, fine needle localization is difficult and requires a combination of estimation and image processing to be successful. Indeed, 3D ultrasound imaging suffers from noise, artifacts and works at a low frequency. We propose a needle tip pose estimation method in the context of 3D robotic needle steering under 3D ultrasound feedback, based on multi-rate, multi-sensor fusion [1].