Improved particle swarm optimization algorithm applied to rigid registration in medical images

Abstract

Image registration is a space-temporary correlation process that allows comparison and/or image matching. This process has value into medical area when it comes to compare images acquired by different modalities or in different times. In this work, we present a method based on computational intelligence techniques from Particle Swarm Optimization (PSO) algorithm, to seek the best answer (particle) exploring a solutions set (swarm). The algorithm we developed includes an original idea for starting and for avoiding local minimum values, in order to achieve good rigid registration results (scale, rotation, translation). The mono and multi modal registration are performed on 2D magnetic resonance imaging (MRI) in T1-T2 sequences and single-photon emission computed tomography (SPECT) images. Experimental results show better optimal solution and decrease in convergence time compared to PSO original algorithm.