Joint time-frequency WaveNet architecture for signal processing: application to vibration artifact suppression in high resolution NMR relaxometry experiments

Read full text Summarise with VIRAT
Authors: Rajka Pejanovic, Lucas Siemons, Alessandro Ruda, Vineeth Francis Thalakottoor Jose Chacko, Guillaume Bouvignies, Philippe Pelupessy, Jorge A. Villanueva‐Garibay, D. Flemming Hansen, Fabien Ferrage
Cited by: 0
References: 51

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Fourier-domain methods are central to both signal processing in magnetic resonance and image processing in imaging techniques. Modern analysis tools, including deeplearning methods, rarely exploit both the time-domain and the frequency-domain representations simultaneously. Although the time-and frequency domains are generally unitarily equivalent, a combined analysis of the two representations of the data allows for a highly improved convergence of the analysis. Here, we introduce the Joint Time-Frequency WaveNet (JTF-WaveNet) deep-learning architecture that processes NMR signal in both domains, producing the desired output along with point-wise uncertainty estimates. We use this architecture to suppress artefacts in spectra obtained by high-resolution relaxometry (HRR). In HRR, low-field relaxation decays are recorded on a high-field magnet by transferring the sample to a position in the stray field of the magnet and rapid transfers are needed to prevent polarization loss. However, these transfers lead to vibrations, which induce low-frequency modulations of the free-induction decay, thus generating anti-phase sidebands that compromise quantitative interpretation. To address this, we trained JTF-WaveNet, with physics-informed simulations, to suppress these artifacts. Critically, when characterising the interaction of small molecules with human serum albumin, applying this deep learning approach improves the precision and accuracy of site-specific low-field relaxation rates by ≈ 60%.

Fields of Study & Subjects

Advanced NMR Techniques and Applications NMR spectroscopy and applications Advanced MRI Techniques and Applications

Publication details

Journal / source
ChemRxiv
Publication date
20 March 2026
Language
EN
Document type
Preprint