Image Fusion Using A Convolutional Neural Network

Authors

  • Nibras amer mohamed, Mohammed Sabbih Hamoud Al- Tamimi

Abstract

The image fusion process is characterized as collecting all of the important information from multiple images, as well as its inclusion in fewer, usually one, images. In this paper, there is indeed the solution to Problems with the various images, such as multi-focus images and medical images through a simulation process using images of brain magnetic resonance(MR) to the fuse 's work based on previously abused fusion techniques such as convolutional neural networks (CNN) and In the experimentation, The (CNN) algorithm is being developed with the introduction of the Euclidean distance algorithm as part of operations to make implementation faster and with higher efficiency than (CNN) Standard. Two objective fusion measures widely used in multimodal fusion of medical images are applied to perform quantitative and qualitative assessments. Peak signal to noise ratio ( PSNR) . Image fusion system (IFS) was tested to use standard datasets This dataset includes brain MR images along with manual FLAIR segmentation masks for anomalies. The photos were taken from The Center for Cancer Imaging (TCIA). They a collection of 3740 different brain images samples correspond to (110) patients each patient contains ( 20 – 70) Segmentation Low-grade glioma collection that have at least fluid-attenuated inversion recovery (FLAIR) Required sequence and genomic cluster data brain tumor in magnetic resonance imaging included in The Cancer Genome Atlas ( TCGA). The results of the experiments carried out show that the use of convolutional neural networks (CNN) with the Euclidean distance algorithm is used in training and testing as a classifier of the medical images provides approximately (98.18%) accuracy. Compared with the findings of other published works these rates are considered high.

Published

2020-12-01

Issue

Section

Articles