Memory reduced non-Cartesian MRI encoding using the mixed-radix tensor product on CPU and GPU

TL;DR

This paper introduces a tensor-based reformulation of multi-dimensional non-Cartesian MRI encoding that significantly reduces memory usage on CPU and GPU, enabling more efficient MRI data processing with minimal accuracy loss.

Contribution

The authors propose a novel mixed-radix tensor method for NUFFT that decreases memory requirements and improves computational efficiency in MRI encoding.

Findings

Achieved up to 88.1% memory savings in 3D MRI

Achieved up to 62.4% memory savings in 2D MRI

Negligible accuracy loss compared to CPU version

Abstract

Multi-dimensional non-Cartesian MRI encoding using the precomputed interpolator can encounter the curse of dimensionality, in which the interpolator size exceeds the available memory on the parallel accelerators. Here we reformulate the multi-dimensional non-uniform fast Fourier transform (NUFFT) to a tensor form. The exponentially growing size of the fully precomputed interpolator can be reduced by tensor analysis. We propose a tree-like, mixed-radix tensor method which flexibly reduces the storage of the NUFFT. A parallel tensor product algorithm is proposed and tested with in vivo cardiac MRI data. Cross-architecture comparisons show that up to 88.1% and 62.4% memory savings are seen in 3D and 2D CINE MRI, respectively, subject only to a negligible loss of accuracy compared to the double-precision CPU version.