Geometric Modeling of Detected Objects in Ultrasound Imaging

Abstract

Ultrasound imaging is widely used in medical diagnostics due to its safety, real-time capability, and cost-effectiveness. Nonetheless, images produced by ultrasound are prone to speckle noise, low contrast and sharp delimiting boundary thus making it hard to interpret objects accurately. In this study, it is proposed that geometric modeling approach will be used to model objects found in ultrasound images as a result of contour detection and segmentation in ultrasound. The study combines edge based contour extraction and region based segmentation in order to localize anatomical structures, and pathological regions. The identified contours are also used to form three-dimensional and two-dimensional geometric representations of objects in the two- and three-dimensional space respectively. The framework proposed allows more detailed description of the structure, and quantitative analysis of medical images. It is proved by experimental analysis that the contour detection together with the segmentation is beneficial in enhancing both the localization boundaries and the geometric consistency of modeled objects. The methodology has found application in diagnosis, treatment planning and medical image analysis systems which are computer aided. The results highlight the potential of geometric modeling as a tool for enhancing the interpretability of ultrasound data.