Linear Algebraic Approaches to Neuroimaging Data Compression: A Comparative Analysis of Matrix and Tensor Decomposition Methods for High-Dimensional Medical Images

TL;DR

This paper compares matrix and tensor decomposition methods, specifically SVD and Tucker, for neuroimaging data compression, highlighting their strengths and trade-offs in fidelity and structural preservation.

Contribution

It provides a comparative analysis of Tucker and SVD methods, demonstrating Tucker's superiority in preserving data structure and fidelity in neuroimaging compression.

Findings

Tucker decomposition achieves higher reconstruction fidelity.

SVD provides better extreme compression.

Tucker preserves multi-dimensional relationships.

Abstract

This paper evaluates Tucker decomposition and Singular Value Decomposition (SVD) for compressing neuroimaging data. Tucker decomposition preserves multi-dimensional relationships, achieving superior reconstruction fidelity and perceptual similarity. SVD excels in extreme compression but sacrifices fidelity. The results highlight Tucker decomposition's suitability for applications requiring the preservation of structural and temporal relationships.