Title: MDLNet: fusion of magnetic resonance and nuclear medicine brain images using deep learning based techniques

Authors: Syed Munawwar; P.V. Gopi Krishna Rao

Addresses: Department of Electronics and Communication Engineering, Jawaharlal Nehru Technological University Anantapur, Ananthapuramu, Andhra Pradesh, 515002, India; Department of Electronics and Communication Engineering, Santhiram Engineering College, Nandyal, Andhra Pradesh, 518501, India ' Department of Electronics and Communication Engineering, Rajeev Gandhi Memorial College of Engineering and Technology, Nandyal, Andhra Pradesh, 518501, India

Abstract: High-quality multi-modal picture fusion is critical in medical imaging to improve diagnosis accuracy. Medical multi-modal image fusion creates more comprehensive and accurate images by combining data from many imaging modalities. An effective deep learning-based model is proposed in this research for medical image fusion. To enhance the image quality, the MDLNet model is trained and optimised using multi-scale feature extraction network and multi-scale feature fusion network models. A modified ResNet-50 model is used to extract the local features, and the DenseNet-169 model is used to extract global features. The method known as the Artificial Gorilla Troops Optimiser (AGTO) is used to choose significant features. Furthermore, a novel multi-scale feature fusion network named the modified FasterNet model is used to fuse selected local and global features. The final fused images are created by reconstructing fused features using three 3 × 3 convolutional kernels. The proposed approach produces more information and better objective metrics in the fusion results.

Keywords: image fusion; modified ResNet-50; DenseNet-169 model; modified FasterNet; convolutional kernels.

DOI: 10.1504/IJSISE.2025.150008

International Journal of Signal and Imaging Systems Engineering, 2025 Vol.14 No.1, pp.1 - 19

Received: 26 Aug 2024
Accepted: 05 May 2025
Published online: 21 Nov 2025 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article