Ultrasound image-guided algorithms for tracking liver motion

In this paper, an image-guided system for a liver motion tracking and tumor treatment is presented. Traditional medical image-guided systems include CT, MRI and ultrasound image (US). While considering frame rate and invasiveness, the ultrasound image-guided system is proposed for tracking the liver motion. The main objective is to use ultrasound image for continuously tracking liver motion and for tumor treatment. After locating the liver motion from ultrasound images, the corresponding relationship can immediately inform the treatment probe the location of the liver tumor existed in body. Ultrasound images have lower resolution and ill-defined edges in noisy images. To improve the accuracy and computation speed, four tracking methods are tested. These methods are: (1) K-means clustering, (2) template matching, (3) optical flow, and (4) neural network. The K-means clustering method is first used to roughly identify the liver location and the (2)-(4) methods are performed separately to precisely estimate the liver motion. Two different scenarios are experimentally tested. In the first scenario, the “subject” breathes normally. In the second scenario, the “subject” varies between taking deep and slow breathes, holding his breath, or panting rapidly. For the first scenario, all three methods could track the target motion successfully, while, for the second scenario, all methods might lose the target occasionally.