Transformer-based Parameter Fitting of Models derived from Bloch-McConnell Equations for CEST MRI Analysis
This addresses a domain-specific problem for CEST MRI analysis by providing a more accurate method for metabolite detection, though it appears incremental as it applies a known neural network architecture to a specific bottleneck.
The paper tackled the challenge of quantifying CEST MRI data by introducing a transformer-based neural network to fit parameters like metabolite concentrations and exchange rates from Bloch-McConnell equations, showing it clearly outperforms classical gradient-based solvers.
Chemical exchange saturation transfer (CEST) MRI is a non-invasive imaging modality for detecting metabolites. It offers higher resolution and sensitivity compared to conventional magnetic resonance spectroscopy (MRS). However, quantification of CEST data is challenging because the measured signal results from a complex interplay of many physiological variables. Here, we introduce a transformer-based neural network to fit parameters such as metabolite concentrations, exchange and relaxation rates of a physical model derived from Bloch-McConnell equations to in-vitro CEST spectra. We show that our self-supervised trained neural network clearly outperforms the solution of classical gradient-based solver.