An Automated System for the Segmentation of Dynamic Scintigraphic Images

Abstract

This paper presents a novel system for the automatic segmentation of Dynamic Scintigraphic Images (DSI). DSI allow a functional exploration of the studied structure and therefore a good understanding of the pathological phenomenon. The proposed system aims at achieving high accuracy of the segmentation process for a precise calculation of the renal functions and the ventricular ejection fraction, thus speeding up the clinical diagnosis and decision-making. The proposed method combines the Fast Marching Method (FMM) and HOG3D spatiotemporal descriptor with the supervision of a Multi Agent System (MAS). Our MAS is composed of a set of supervisor and explorer agents able to detect the regions of interest from a video sequence of scintigraphic images captured over time. The behavior of communicating supervising and exploring  agentsis inspired from the Active contour technique. This automatic segmentation system is expected to assist physicians in both clinical diagnosis and educational training. The results of the application of our method on several dynamic images are presented and discussed.