Hybrid Swarm Intelligence for FPGA-Based Noise Mitigation in Gradient-Sensitive MRI Systems

This research presents an advanced Magnetic Resonance Imaging (MRI) signal processing framework to address time-varying noise and environmental disturbances. A novel adaptive filtering mechanism is proposed to enhance noise mitigation while dynamically adjusting coefficients based on gradient field variations and magnetic coil responses. The core contribution lies in the development of a Modified Opposition-Based Artificial Ant Colony Optimization (MOACO) algorithm, marking the first application of an optimization algorithm tailored for MRI systems. The proposed method integrates swarm intelligence by combining opposition-based ant bee colony optimization, leveraging the cooperative behaviors of ants and bees to optimize the adaptive filter's weight estimation. The optimization process dynamically adjusts the step size, ensuring faster convergence and improved noise suppression. A Parallel Architecture Opposition-Based ABC algorithm is introduced for efficient hardware implementation. The effectiveness of the proposed solution is evaluated against conventional filtering methods, including Two-Dimensional LMS (TDLMS), Recursive Least Squares (RLS), and Adaptive Filtering Least Mean Error (AFLME). The algorithms are implemented on a Field-Programmable Gate Array (FPGA) platform, specifically the EP-FPGA-256C6 Cyclone, utilizing 65 nm and 90 nm CMOS technology nodes to ensure optimal power efficiency, reduced Look-Up Table (LUT) utilization, and minimal delay. Experimental results demonstrate that the MOACO-enhanced adaptive filtering framework significantly improves power efficiency, circuit performance, and noise reduction, leading to enhanced MRI image reconstruction and minimized motion artifacts. This study establishes a novel direction for integrating AI-driven optimization techniques with hardware-based MRI signal processing, paving the way for more efficient medical imaging systems.