Hardware acceleration for ultra-fast Neural Network training on FPGA for MRF map reconstruction

TL;DR

This paper presents an FPGA-based neural network that accelerates MRF map reconstruction, enabling real-time brain imaging analysis with significantly reduced training time, suitable for mobile and clinical applications.

Contribution

The authors introduce a novel FPGA implementation of neural network training for MRF reconstruction, achieving rapid training times and real-time processing capabilities.

Findings

Training time reduced to 200 seconds on FPGA

Significantly faster than CPU-based training (up to 250x)

Potential for real-time brain analysis in clinical settings

Abstract

Magnetic Resonance Fingerprinting (MRF) is a fast quantitative MR Imaging technique that provides multi-parametric maps with a single acquisition. Neural Networks (NNs) accelerate reconstruction but require significant resources for training. We propose an FPGA-based NN for real-time brain parameter reconstruction from MRF data. Training the NN takes an estimated 200 seconds, significantly faster than standard CPU-based training, which can be up to 250 times slower. This method could enable real-time brain analysis on mobile devices, revolutionizing clinical decision-making and telemedicine.