Multi-Agent Segmentation using Region Growing and Contour Detection: Syntetic evaluation in MR Images with 3D CAD Reconstruction

Abstract

Computer assisted surgery navigation takes full advantage of progress in engineering disciplines. Developed models increase the accuracy of replacement technique, especially in hip surgery to reduce the risk of component malpositioning. This paper presents a 3D model reconstruction from contour extracted through a proposed multi-agent segmentation (MAS) approach. We first describe parallel agents’ behaviors for extracting the object of interest from MR Images. The proposed algorithm is formulated by combining region growing and contour detection ensuring an overall segmentation. The 3D CAD model is generated using MATLAB code implemented as per the MAS method and gives us good result of reconstruction in most of the cases. The comparison of the proposed method with the traditional approach is made in terms of run times segmentation and edge detection accuracy.