Automatic needle segmentation in 3D ultrasound data using a hough transform approach

Abstract

Segmentation in ultrasound data is a very challenging field of research in medical image processing. This article presents a method for automatic segmentation of biopsy needles and straight objects in noisy 3D image data. It uses a Hough-based segmentation approach, which has been exemplary adapted for the application on prostate biopsy data. An evaluation was performed on in-vivo 3D US data and shows promising results. Angular segmentation accuracy was evaluated with a mean of 2.1 degrees, which is comparable to human observers.