MP-PCA denoising of fMRI time-series data can lead to artificial activation "spreading"

TL;DR

This study investigates how MP-PCA denoising improves SNR in rodent fMRI but can cause artificial activation spreading, affecting the accuracy of functional maps, with implications for optimizing denoising parameters.

Contribution

The paper develops a method to denoise vendor data and demonstrates how MP-PCA denoising can lead to false activation spreading, highlighting the importance of denoising window size.

Findings

MP-PCA denoising increases SNR and spectral amplitude in fMRI maps.

Larger denoising windows cause more activation spreading and false positives.

Optimal denoising window depends on data SNR and CNR for best specificity.

Abstract

MP-PCA denoising has become the method of choice for denoising in MRI since it provides an objective threshold to separate the desired signal from unwanted thermal noise components. In rodents, thermal noise in the coils is an important source of noise that can reduce the accuracy of activation mapping in fMRI. Further confounding this problem, vendor data often contains zero-filling and other effects that may violate MP-PCA assumptions. Here, we develop an approach to denoise vendor data and assess activation "spreading" caused by MP-PCA denoising in rodent task-based fMRI data. Data was obtained from N = 3 mice using conventional multislice and ultrafast acquisitions (1 s and 50 ms temporal resolution, respectively), during visual stimulation. MP-PCA denoising produced SNR gains of 64% and 39% and Fourier spectral amplitude (FSA) increases in BOLD maps of 9% and 7% for multislice and ultrafast data, respectively, when using a small [2 2] denoising window. Larger windows provided higher SNR and FSA gains with increased spatial extent of activation that may or may not represent real activation. Simulations showed that MP-PCA denoising causes activation "spreading" with an increase in false positive rate and smoother functional maps due to local "bleeding" of principal components, and that the optimal denoising window for improved specificity of functional mapping, based on Dice score calculations, depends on the data's tSNR and functional CNR. This "spreading" effect applies also to another recently proposed low-rank denoising method (NORDIC). Our results bode well for dramatically enhancing spatial and/or temporal resolution in future fMRI work, while taking into account the sensitivity/specificity trade-offs of low-rank denoising methods.