Nuclear Magnetic Resonance (NMR) spectroscopy is a powerful analytical technique for obtaining atomic-level structural and dynamical information about chemical and biological systems. Due to the complexity of NMR signals, extensive signal processing and analysis are required before meaningful interpretation is possible. While the physical principles underlying NMR are rooted in quantum mechanics, this project focuses on advancing computational methods for NMR data processing. This project builds directly on previous work demonstrating the power of Artificial Intelligence (AI)–based approaches for NMR spectral analysis in the frequency domain. In particular, earlier studies established that AI-driven methods, including the Peak Probability Presentation (P³) framework, can generate artifact-free, high-resolution spectral representations and significantly improve peak detection and reliability in one-dimensional and relatively low-complexity spectra. These results provide strong proof of concept for the applicability of AI and probabilistic spectral representations to NMR data analysis. The present project aims to extend these successful AI-based methodologies to multidimensional NMR experiments, with a primary focus on NOESY spectra, which are characterized by substantially higher spectral complexity, dense peak overlap, and increased sensitivity to noise and artifacts. Addressing these challenges requires new computational strategies capable of handling large data volumes, complex correlation patterns, and ambiguous peak structures. To this end, the project introduces Magnetic Resonance processing with Artificial Intelligence (MR-AI), a deep-learning–based framework designed to enhance NMR signal processing and analysis beyond the capabilities of traditional methods. The MR-AI approach leverages advanced neural network architectures to model spectral features and noise characteristics, enabling robust analysis of highly complex multidimensional spectra. The P³ concept is further developed and adapted for NOESY data, providing probabilistic representations that improve confidence in cross-peak identification and interpretation. By extending AI- and P³-based spectral analysis from frequency-domain proof-of-concept studies to complex NOESY experiments, this project establishes a coherent and scalable research trajectory toward next-generation NMR data analysis. The expected outcomes include improved robustness, higher effective resolution, reduced manual intervention, and more reliable extraction of structural information from challenging NMR datasets. The project is conducted under the supervision of Prof. Vladislav Orekhov, Department of Chemistry and Molecular Biology, Swedish NMR Centre, University of Gothenburg, Gothenburg, 40530, Sweden. Relevant Publications The feasibility and scientific foundation of the proposed AI-based NMR methodology have been demonstrated in the following publications: 1. NMR spectrum reconstruction as a pattern recognition problem. Journal of Magnetic Resonance, 346, 107342. https://doi.org/10.1016/j.jmr.2022.107342 2. Beyond traditional magnetic resonance processing with artificial intelligence. Communications Chemistry, 7, 244. https://doi.org/10.1038/s42004-024-01325-w 3. Towards Ultimate NMR Resolution with Deep Learning. arXiv:2502.20793 (under revision in Science Advances). https://doi.org/10.48550/arXiv.2502.20793