Medical Image Fusion using Pulse Coupled Neural Networks with Meta Heuristic Optimization techniques

Abstract

Image fusion is a method for using image processing techniques to merge multimodal images. Specifically, the goal is to combine diverse and complementary evidence to boost the details visible in the photographs and maximize the clarity of the interpretation. The accuracy of the result can be high by merging two images like CT and MRI images. The fusion of images is closely connected to many various areas of image processing, such as images acquired from satellites, dense areas, and is highly recommended in the medical field. Initially, the research on image fusion is performed for satellite imaging, based on the results obtained, the field of research is extended to medical imaging.The aim of the fusion of images is to combine descriptions of the multiple scanned images from same scene. An image which is new, is highly suitable for machine and human interpretation or processing of image tasks like extraction of features, object detection, and segmentation is the result of image fusion. In this work, a fusion of MRI and CT images is proposed and a new model is designed to approach the fusion based on neural networks and Optimization techniques which works better and gives good results. To determine the right parameters, an adaptive PCNN ( Pulse Coupled Neural Network ) is utilized, and using Quantum Cuckoo Search Algorithm(QCSA) optimization of parameters is done. The suggested paradigm is a variation of QCSA-PCNN. The reliability and accuracy of the image is increased. The fitness function of the proposed optimization technique is defined using spatial frequency (SF), entropy (EN), mutual information(MI), and sharpness of the image for finding the optimal solution. Various parametric values are being tested to show that the suggested QCSA-PCNN is superior compared to other current techniques like QPSO-PCNN. The PSNR obtained using QPSO-PCNN is 40.82, and the proposed QCSA-PCNN the PSNR (Peak Signal to Noise Ratio) value is 43.79. The suggested newline newline