Tensor denoising of high-dimensional MRI data

The signal to noise ratio (SNR) fundamentally limits the information accessible by magnetic resonance imaging (MRI). This limitation has been addressed by a host of denoising techniques, recently including so-called MPPCA: Principal component analysis (PCA) of the signal followed by automated rank estimation, exploiting the Marchenko-Pastur (MP) distribution of noise singular values. Operating on matrices comprised by data-patches, this popular approach objectively identifies noise components and, ideally, allows noise to be removed without introducing artifacts such as image blurring or non-local averaging. The MPPCA rank estimation, however, relies on a large number of noise singular values relative to the number of signal components to avoid such ill effects. This condition is unlikely to be met when data-patches and therefore matrices are small, for example due to spatially varying noise. Here, we introduce tensor MPPCA (tMPPCA) for the purpose of denoising multidimensional data, for example from multi-contrast acquisitions. Rather than combining dimensions in matrices, tMPPCA utilizes each dimension of the multidimensional data's inherent tensor-structure to better characterize noise, and to recursively estimate signal components. Relative to matrix-based MPPCA, tMPPCA requires no additional assumptions, and comparing the two in a numerical phantom and a multi-TE diffusion MRI dataset, tMPPCA dramatically improves denoising performance. This is particularly true for small data-patches, which we believe will improve denoising in cases of spatially varying noise.