Automated in-plane visualization of steep needles from 3D ultrasound data volumes

Abstract

During ultrasound-guided needle interventions, low signal-to-noise ratio and poor needle visibility limit the performance of automated detection systems. This becomes even more challenging when the needle is inserted at higher angles with respect to the ultrasound probe. For very large insertion angles, the needle becomes virtually invisible in the ultrasound data and medical specialists need to find the needle indirectly either from out-of-plane or in-plane views. In this paper, we propose a novel method to automatically detect steep needles in 3D ultrasound data and visualize its 2D in-plane view to the medical specialist. Our method exploits indirect information regarding the presence of a needle in the volume by examining the shadow traces of structures. The proposed algorithm successfully detects the needle plane with high accuracy for all the ten measured datasets. Furthermore, the full-length needle and its tip are always visible in the extracted scan planes. The proposed method is efficient and robust to noise and artifacts, thereby strongly supporting the clinical intervention and eliminating the need for external tracking devices.