Particle Swarm Optimization — Functional Link Multilayer Perceptron for Rician noise suppression from MRI images
Ashruti Upadhyaya, P. Reshma Sagar, Manish Kumar, Sudhanshu Mishra · 2016
In this paper, a swarm intelligence based neural network named as Particle Swarm Optimization - Functional Link Multi Layer Perceptron (PSO-FLMLP) is proposed for suppression of Rician noise from Magnetic Resonance Imaging (MRI) images. The performance of PSO-FLMLP is also compared with three other competitive networks such as Multi Layer Perceptron (MLP), Least Mean Square based Functional Link Artificial Neural Network (LMS-FLANN), Particle Swarm Optimization based Functional Link Artificial Neural Network (PSO-FLANN). The comparison of performance includes parameters like SSIM, PSNR, NRDB and computational time. Non parametric statistical analysis using Friedman test is also performed to demonstrate the superiority of our proposed network. It is observed that the introduced network gives improved performance with less computational time. The proposed technique can also be applied to any other medical image with noise of different probability density function.