Frequency and Phase Correction of GABA-Edited Magnetic Resonance Spectroscopy using Complex-Valued Convolutional Neural Networks

Abstract

PurposeTo determine the significance of complex-valued inputs and complex-valued convolutions compared to real-valued inputs and real-valued convolutions in Convolutional Neural Networks (CNNs) for frequency and phase correction (FPC) of GABA-edited Magnetic Resonance Spectroscopy (MRS) data. MethodsAn ablation study was performed to determine the most effective input (real or complex) and convolution type (real or complex) to predict frequency and phase shifts in GABA-edited MEGA-PRESS data using CNNs. The best CNN model was subsequently compared to two recently proposed deep learning (DL) methods for FPC of GABA-edited MRS. All methods were trained using the same experimental setup and evaluated using GABAs signal-to-noise ratio and linewidth, Choline artifact, and by analyzing the reconstructed final difference spectrum. Statistical significance and effect size were assessed using the Wilcoxon signed rank test and Cohens d respectively. ResultsThe ablation study showed that using complex values for the input represented by real and imaginary channels in our model input tensor, with real (conventional) convolutions was most effective for FPC. For the comparative study, the simulated data test set showed that our CNN model that received complex-valued inputs with real convolutions outperformed models it was compared against with a lower mean absolute error (p<0.05). For the in vivo data test set, our model performed similarly to other DL FPC models. ConclusionOur results indicate that the optimal CNN configuration for GABA-edited MRS FPC uses a complex-valued input and real-valued convolutions. This model outperformed existing DL models on simulated data and performed similarly on in vivo data.