Enhanced Needle Detection in Ultrasound Images using Acoustic Excitation and Ultrasound Image Analyses

Abstract

Ultrasound imaging provides a low cost and noninvasive imaging modality to improve the safety and success rate of needle-based interventions by guiding the needle towards the target anatomy. Nevertheless, the limited needle visibility in conventional brightness mode (B-mode) ultrasound images might degrade the capability of achieving accurate localization of the needle axis and tip. In this paper, a computer-based method is introduced to enable accurate needle detection in ultrasound images. In this method, the needle is excited using a voice actuator to generate low-amplitude acoustic waves that propagate through the needle. The excited needle is scanned using ultrasound imaging to acquire a power Doppler ultrasound image and a B-mode ultrasound image. The power Doppler image is processed using thresholding and Radon transform analyses to obtain approximate estimation of the needle axis and identify a region of interest (ROI) that includes the vibrating needle. Moreover, accurate estimation of the needle axis is achieved by analyzing the ROI that includes the needle in the B-mode image using a thresholding procedure combined with a customized Radon transform. Finally, the location of the needle tip is identified by applying an iterative sliding window approach to the B-mode image to quantify the pixel intensities around the estimated needle axis. The accuracy of the proposed method is evaluated by applying the method to detect the axes and tips of eight needles inserted in ex vivo bovine muscle tissue specimens and imaged using linear and curvilinear ultrasound transducers. The results show that the proposed method was able to detect the axes and tips of the inserted needles with error values within the ranges of 0.3° to 1.0° and 0.2 mm to 1.0 mm, respectively. These results suggest the potential of applying the proposed method to enhance the localization of the needle during ultrasound-guided needle-based interventions.