Portable nuclear magnetic resonance (NMR) spectrometers are susceptible to external electromagnetic interference (EMI), leading to low signal-to-noise ratio and reduced analytical accuracy. This paper proposes a method based on multiple reference coils and deep learning: multi-channel coils are used to acquire ambient electromagnetic noise in parallel, which is fed into a multi-channel Transformer reconstruction (MCTR) network. The network captures long-range dependencies of ambient electromagnetic signals, predicts and removes the environmental noise in the main receiving coil in real time. Experiments show that this method effectively suppresses noise in both simulated and actual NMR data, improves detection performance, outperforms traditional methods, maintains high signal quality, and has strong robustness. It provides effective support for the application of portable NMR in complex electromagnetic environments, and is expected to promote the development of on-site detection.

One-dimensional proton nuclear magnetic resonance (NMR) spectroscopy is a high-resolution, non-invasive technique widely used for structure elucidation and composition analysis. However, when applied to complex systems, its effectiveness is often hampered by overlapping peaks from similar chemical shifts and J-coupling, along with concentration variations obscuring weak signals from low-abundance compounds. To enhance the detection sensitivity for weak signals in crowded spectral regions, this work proposes Hadamard-DQF-LMO, integrating longitudinal multi-spin orders (LMOs), Hadamard encoding, and double quantum filtering (DQF). The approach utilizes DQF to selectively detect LMO signals while suppressing strong interference. The incorporation of polychromatic transition pulses and Hadamard-encoded 180^° pulses enables parallel acquisition of multiple frequencies, which significantly improves both sensitivity and detection efficiency. Experiments on orange juice and mixed amino acid samples demonstrate peak separation with enhanced selectivity and signal-to-noise ratio (SNR), offering a novel and effective strategy for the NMR analysis of complex systems.

Ammonium fluoride (NH₄F) etching is an effective method for constructing hierarchical zeolites, yet its precise impact on the detailed structure and acidity remains unclear. This study employs probe molecule-assisted solid-state NMR technique to systematically investigate the evolution of acid site structures and properties during NH₄F etching of ZSM-5 zeolites. The results reveal that the degree of NH₄F etching significantly manipulates the amount, strength, type, and spatial accessibility of acid sites, and it is also directly associated with the development of hierarchical pore structures. This work elucidates the influence mechanism of NH₄F etching on zeolite structure and acidic property at the atomic level, providing a theoretical foundation for the rational design and optimization of efficient hierarchical zeolites.

The tautomeric distribution of monosaccharides could influence reaction pathways and product selectivity during their conversion and utilization. Using DMSO-d₆ as cosolvent, we investigated the tautomeric distribution of monosaccharides in imidazolyl ionic liquids at 25 ℃ by quantitative ¹H NMR. In acidic ionic liquids, the results indicate that the proportion of β-pyranose after equilibrium decreases in the following order: D-glucose > D-glucosamine hydrochloride > N-acetyl-D-glucosamine > D-mannose. D-fructose is almost completely converted to 5-hydroxymethylfurfural (formed from the furanose) in [HSO₃-BMIM]HSO₄. In other ionic liquids, the proportion of furanose even exceeds that of pyranose after equilibrium. The introduction of amphoteric metal chloride and fluorine may lead to a higher proportion of furanose. The addition of [BMIM]BF₄ and other reagents may inhibit the tautomeric conversion of monosaccharides. The fundamental data on monosaccharide tautomer distribution can guide the selection and design of ionic liquids for biomass conversion.

The geometric configuration of cis-trans isomers critically influences their biological activities, making the separation and configurational identification essential for efficacy evaluation. In this study, cis/trans-1,2,4-oxadiazole derivatives were synthesized and comprehensively characterized by nuclear magnetic resonance (NMR), Fourier transform infrared spectroscopy (FT-IR), high-resolution mass spectrometry (HRMS), and theoretical calculations. This approach enabled unambiguous identification and distinction of a pair of isomers, including cis-4-[2-(3-((1H-indol-3-yl)methyl)-1,2,4-oxadiazol-5-yl)vinyl]-N,N-dimethylaniline and its trans-configuration analogue. Anti-Toxoplasma gondii activity assessment revealed that the cis-isomer (compound 2, selectivity index SI = 1.50) exhibited higher activity than the trans-isomer (compound 1, SI = 0.92) and the positive control spiramycin (SI = 0.98), highlighting the value of configurational separation and identification in pharmaceutical research. This study deepens the understanding of configuration-activity relationships and provides new insights for developing highly selective anti-Toxoplasma agents.

This study aims to explore the structural changes in white matter and the physiological mechanisms in type 2 diabetes mellitus (T2DM) rats by magnetic resonance imaging (MRI) technology. A T2DM model was established by 8 weeks of high-fat diet combined with intraperitoneal injection of a dose of 30 mg/kg streptozotocin (STZ). 4 weeks after modeling, visual analysis of white matter in rats was performed by MRI, and the physiological mechanism of MRI index changes was further explained by immunohistology. The results showed that the striatal white matter volume decreased, the fractional anisotropy (FA), mean diffusivity (MD) and axial diffusivity (AD) decreased, and radial diffusivity (RD) increased in T2DM rats. Immunostain of phosphorylated neurofilament (SMI-31) and myelin basic protein (MBP) indicated axonal damage and demyelination of striatum in T2DM rats. In conclusion, the striatal injury in T2DM rats was observed by diffusion tensor imaging (DTI), and abnormal DTI index may be a manifestation of striatal axonal damage and demyelination, which can potentially be used as surrogates for evaluating diabetic brain injuries.

Magnetic resonance imaging (MRI) and positron emission tomography (PET) are commonly used imaging techniques for the early diagnosis of Alzheimer's disease (AD). The combination of these two modalities enables a more comprehensive assessment of brain status by utilizing both anatomical and metabolic information. However, traditional multimodal fusion, which relies primarily on simple channel splicing, fails to fully exploit the complementary information across modalities and limits the model's effectiveness. To address this, this paper proposes a multi-task classification model for AD based on adversarial learning and cross-attention. The model reduces inter-modal feature discrepancies through adversarial learning, followed by feature fusion via cross-attention, and introduces a brain age prediction task as an auxiliary task to improve classification performance. Experimental results demonstrate that the proposed method achieves an accuracy of 91.10% and an F1 score of 91.01% in classifying AD, mild cognitive impairment (MCI), and normal controls (NC). This not only enhances the accuracy of early diagnosis but also strengthens the ability to monitor disease progression, thereby providing strong support for clinical interventions in AD.

This study designed and constructed a pH-responsive intelligent theranostic nanoprobe. The nanoprobe utilizes perfluorocarbon nanoparticles with ¹⁹F magnetic resonance imaging (MRI) contrast capability as the core, with bovine serum albumin (BSA) adsorbed on the surface to form a “protein corona”. Furthermore, tannic acid-iron (TA-Fe^III) is complexed to create the core-shell structured PP@BSA-TAFe^III nanoparticles (NPs) probe. Owing to the unsaturated coordination state of Fe³⁺ in TA-Fe^III, the nanoprobe binds to transferrin upon entering the bloodstream, forming a “hybrid” protein corona and thereby achieving active targeting of transferrin receptors (TfR) highly expressed on the tumor cell membrane. The nanoprobe efficiently accumulates in tumor tissue and generates a strong ¹⁹F-MRI signal, enabling highly sensitive tumor imaging. Upon entry into the acidic tumor microenvironment, Fe³⁺ is released to induce ferroptosis, while TA-Fe^III exhibits excellent photothermal effects, thereby achieving synergistic therapy. This in situ regulation strategy of protein corona offers a new approach for precise tumor diagnosis and therapy.

Mild traumatic brain injury (mTBI) is a common neurological disorder in clinical practice, yet its diagnosis and management remain challenging due to the hidden nature of symptoms and the underlying pathological complexity. While imaging techniques such as computed tomography (CT) and conventional magnetic resonance imaging (MRI) are the mainstay for acute-phase assessment, they have limitations in detecting subtle injuries and evaluating long-term prognosis. In recent years, multimodal MRI technology has been developed in the research of mTBI, offering novel approaches for revealing its potential pathological mechanisms and exploring the key objective imaging indicators. Specifically, susceptibility weighted imaging (SWI) is sensitive for detecting microbleeds and iron deposition in the brain; amide proton transfer (APT) imaging reflects changes in molecular and metabolic levels; diffusion and functional imaging techniques help depict abnormalities in white matter microstructure and brain networks. The integration of multimodal MRI and the construction of imaging databases will be important directions for advancing early diagnosis, precise assessment, and AI-assisted intervention. This article systematically reviews research progress of related MRI techniques, analyzes their advantages and limitations, and discusses their prospects in clinical translation.

Diffusion magnetic resonance imaging (dMRI) is extensively employed to investigate the microstructure and fiber tract orientation of white matter in the brain. However, high angular and multi-shell sampling with high spatial resolution usually requires a prolonged scan time. In recent years, deep learning techniques have been widely adopted for dMRI super-resolution reconstruction, which aims to reconstruct high-resolution imaging signals from rapidly acquired images under sparse sampling conditions, thereby enabling more accurate fitting of brain microstructure imaging parameters. This paper surveys and analyzes the latest research progress in deep learning-based reconstruction of brain dMRI. According to different reconstruction targets, the methods are classified into three categories: reconstruction of basic diffusion metrics, reconstruction of high-order microstructure metrics, and reconstruction of the fiber orientation distribution function (fODF). The implementation techniques, evaluation metrics, and commonly used public datasets for each category are discussed in detail. Finally, the main challenges and research trends in dMRI super-resolution reconstruction are summarized.