International Conference on Computer and Knowledge Engineering

Home / 15th International Conference on Computer and Knowledge Engineering

MC-BioCLIPSR: A Mamba-CNN Hybrid Network with BioMedCLIP-Guided Loss for High-Resolution Brain MRI Reconstruction

Authors :

Amin Kazempour¹ Jafar Tanha² SeyedEhsan Roshan³ Mahdi Zarrin⁴ Haniyeh Nikkhah⁵

1- University of Tabriz Electrical & Computer Engineering department 2- University of Tabriz Electrical & Computer Engineering department 3- University of Tabriz Electrical & Computer Engineering department 4- University of Tabriz Electrical & Computer Engineering department 5- University of Tabriz Electrical & Computer Engineering department

Keywords :

Medical Image Super-Resolution،Mamba،Brain MRI Reconstruction،BioMedCLIP،Semantic Consistency Loss

Abstract :

High-resolution images are essential in medical imaging, as they provide critical details necessary for accurate analysis. However, acquiring high-resolution images is often time-consuming and expensive. A practical alternative is to reconstruct high-resolution images from their low-resolution counterparts. In this paper, we propose a novel encoder-decoder architecture based on deep learning, which integrates convolutional neural networks (CNNs) and Mamba blocks to perform high-resolution image reconstruction. Additionally, we introduce a new loss function inspired by the BioMedCLIP approach to enhance training effectiveness. To demonstrate the superiority of our method, we compare it with several state-of-the-art techniques on the IXI dataset.