MR spectroscopy frequency and phase correction using convolutional neural networks

Abstract

A CNN-based approach provides a solution to the automated preprocessing of MRS data, and the experimental results demonstrate the quantitatively improved spectra quality compared to the state-of-the-art approach.