We proposed a spectrum-to-spectrum/spectrum-to-phase phase correction method based on a neural network for magnetic resonance spectra. The former network obtains phase-corrected spectra by the end-to-end training the mapping between the manually corrected spectra and uncorrected spectra. And the latter can achieve more accurate phase correction by predicting the zero- and first-order phases for correction. The result shows that the proposed network can effectively obtain high-quality phase correction spectra even under noisy and baseline distortion conditions.
This abstract and the presentation materials are available to members only; a login is required.