ArXiv最新文献

The Multimodal Direct Inversion (MMDI) algorithm is widely used in Magnetic Resonance Elastography (MRE) to estimate tissue shear stiffness. However, MMDI relies on the Helmholtz equation, which assumes wave propagation in a uniform, homogeneous, and infinite medium. Furthermore, the use of the Laplacian operator makes MMDI highly sensitive to noise, which compromises the accuracy and reliability of stiffness estimates. In this study, we propose the Deep-Learning driven Inversion Framework for Shear Modulus Estimation in MRE (DIME), aimed at enhancing the robustness of inversion. DIME is trained on the displacement fields-stiffness maps pair generated through Finite Element Modelling (FEM) simulations. To capture local wave behavior and improve robustness to global image variations, DIME is trained on small image patches. We first validated DIME using homogeneous and heterogeneous datasets simulated with FEM, where DIME produced stiffness maps with low inter-pixel variability, accurate boundary delineation, and higher correlation with ground truth (GT) compared to MMDI. Next, DIME was evaluated in a realistic anatomy-informed simulated liver dataset with known GT and compared directly to MMDI. DIME reproduced ground-truth stiffness patterns with high fidelity (r = 0.99, R² = 0.98), while MMDI showed greater underestimation. After validating DIME on synthetic data, we tested the model in in vivo liver MRE data from eight healthy and seven fibrotic subjects. DIME preserved physiologically consistent stiffness patterns and closely matched MMDI, which showed directional bias. Overall, DIME showed higher correlation with ground truth and visually similar stiffness patterns, whereas MMDI displayed a larger bias that can potentially be attributed to directional filtering. These preliminary results highlight the feasibility of DIME for clinical applications in MRE.

Purpose: To develop DarkSPARC, a retrospective, training-free, self-calibrated spectral reconstruction method that converts routine bright-blood 3D left atrial (LA) late gadolinium enhancement (LGE) MRI into a dark-blood image, and to quantify its impact on LA scar-pool CNR, SNR, effective CNR (eCNR), and scar quantification accuracy.

Methods: DarkSPARC embeds bright-blood LA LGE into a calibrator-conditioned (N+1)-dimensional spectral domain and reconstructs a dark-blood-like image using scan-specific spectral landmarks. A scan-specific 3D numerical phantom framework was built from LAScarQS post-ablation LGE by cloning remote myocardium into the LA wall and imposing controlled scar burden. Five baseline cases spanning the 5th-95th percentiles of native scar-pool CNR, each with multiple scar burdens and 10 CNR degradation levels, yielded 200 phantoms. For every phantom, LA scar-pool CNR, SNR, eCNR, and Scar% were measured on bright-blood and DarkSPARC images. In vivo performance was evaluated in 60 public post-ablation scans of atrial fibrillation patients.

Results: In scan-specific phantoms, DarkSPARC increased LA scar-pool CNR, SNR, and eCNR over bright-blood in all 200 experiments, with DarkSPARC/bright-blood ratios up to about 30-fold for CNR and about 6-fold for SNR in the lowest-CNR conditions. At 70% CNR degradation, bright-blood underestimated ground-truth LA Scar% by -37% to -54%, whereas DarkSPARC reduced bias to about -3% to -5%. In vivo, DarkSPARC similarly improved metrics: median scar-pool CNR, SNR, and eCNR increased from 20.0 to 135.9 (6.8x), 70.6 to 200.6 (2.8x), and 0.22 to 0.75 (3.4x), respectively (all p<0.001), and LA Scar% increased from 3.9% to 9.75%.

Conclusion: DarkSPARC is a self-calibrated, training-free reconstruction that yields dark-blood 3D LA LGE, boosting CNR/SNR/eCNR and stabilizing reliable scar quantification without extra scans.

Protein S (PS) is a notable anticoagulant implicated in both bleeding and thrombotic disorders, making it a promising drug target. Importantly, PS enhances the anticoagulant function of TFPI$α$, likely circulating in the bloodstream together with TFPI$α$ and a truncated form of factor V (fVshort) in the trimolecular complex, TFPI$α$-fVshort-PS, which we call protein S complex (PSC). PSC has been proposed to strongly inhibit thrombin production by enhancing the ability of TFPI$α$ to inhibit clotting factor Xa up to 100-fold and by localizing to platelet membranes, limiting fXa activity shortly after coagulation starts. Yet, exactly how PS functions with TFPI$α$ as an anticoagulant remains poorly understood. To investigate, we extend an experimentally validated mathematical model of blood coagulation to include PSC and free PS (not part of PSC) in the plasma, as well as free PS and TFPI$α$ in platelets. We find that shortly after coagulation initiation, PSC strongly inhibits thrombin production. We find that the (unknown) magnitude of the enhanced affinity of PSC binding to inhibit fXa critically regulates PSC's impact on thrombin production. We find that under flow, PSC can unexpectedly accumulate on platelets to concentrations ~50 times higher than in the plasma. We also find that PSC limits thrombin production by occupying fV-specific binding sites on platelets. Our results show that changes in PSC can dramatically impact severity of pathological bleeding disorders. For the east Texas bleeding disorder, elevated PSC concentrations eliminate thrombin bursts, leading to bleeding. With fV deficiency, reducing PSC rescues thrombin production in severe fV deficiency and returns thrombin production due to mild fV deficiency to normal. Finally, thrombin production in severe hemophilia A can be substantially improved by blocking PSC's anticoagulant function.

Artificial intelligence (AI) is reshaping computational and network biology by enabling new approaches to decode cellular communication networks. We introduce Hierarchical Molecular Language Models (HMLMs), a novel framework that models cellular signaling as a specialized molecular language, where signaling molecules function as tokens, protein interactions define syntax, and functional consequences constitute semantics. HMLMs employ a transformer-based architecture adapted to accommodate graph-structured signaling networks through information transducers, mathematical entities that capture how molecules receive, process, and transmit signals. The architecture integrates multi-modal data sources across molecular, pathway, and cellular scales through hierarchical attention mechanisms and scale-bridging operators that enable information flow across biological hierarchies. Applied to a complex network of cardiac fibroblast signaling, HMLMs outperformed traditional approaches in temporal dynamics prediction, particularly under sparse sampling conditions. Attention-based analysis revealed biologically meaningful crosstalk patterns, including previously uncharacterized interactions between signaling pathways. By bridging molecular mechanisms with cellular phenotypes through AI-driven molecular language representation, HMLMs establish a foundation for biology-oriented large language models (LLMs) that could be pre-trained on comprehensive pathway datasets and applied across diverse signaling systems and tissues, advancing precision medicine and therapeutic discovery.

Accurate simulations of electric fields (E-fields) in brain stimulation depend on tissue conductivity representations that link macroscopic assumptions with underlying microscopic tissue structure. Mesoscale conductivity variations can produce meaningful changes in E-fields and neural activation thresholds but remain largely absent from standard macroscopic models. Recent microscopic models have suggested substantial local E-field perturbations and could, in principle, inform mesoscale conductivity. However, the quantitative validity of microscopic models is limited by fixation-related tissue distortion and incomplete extracellular-space reconstruction. We outline approaches that bridge macro- and microscales to derive consistent mesoscale conductivity distributions, providing a foundation for accurate multiscale models of E-fields and neural activation in brain stimulation.

Purpose: Magnetization transfer (MT) has been identified as the principal source of $T_1$ variability in the MRI literature. This study assesses the sensitivity of established $T_1$ mapping techniques to variations in the underlying MT parameters.

Methods: For each $T_1$ -mapping method, the observed $T_1$ was simulated as a function of the underlying MT parameters $p_i^{\text{MT}}$ , corresponding to different brain regions of interest (ROIs) at 3T. As measures of sensitivity, the derivatives $\partial T_1^{\text{observed}}/\partial p_i^{\text{MT}}$ were computed and analyzed with a linear mixed-effects model as a function of $p_i^{\text{MT}}$ , ROI, pulse sequence type (e.g., inversion recovery, variable flip angle), and the individual sequences.

Results: The analyzed $T_1$ -mapping sequences have a considerable sensitivity to changes in the semi-solid spin pool size $m_0^{\text{s}}$ , $T_1^{\text{f}}$ of the free, $T_1^{\text{s}}$ of the semi-solid spin pool, and the (inverse) exchange rate $T_{\text{x}}$ . All derivatives vary considerably with the underlying MT parameters and between pulse sequences. In general, the derivatives cannot be determined by the sequence type, but rather depend on the implementation details of the sequence. One notable exception is that variable-flip-angle methods are, in general, more sensitive to the exchange rate than inversion-recovery methods.

Conclusion: Variations in the observed $T_1$ can be caused by several underlying MT parameters, and the sensitivity to each parameter depends on both the underlying MT parameters and the sequence.

Single-cell transcriptomics and proteomics have become a great source for data-driven insights into biology, enabling the use of advanced deep learning methods to understand cellular heterogeneity and gene expression at the single-cell level. With the advent of spatial-omics data, we have the promise of characterizing cells within their tissue context as it provides both spatial coordinates and intra-cellular transcriptional or protein counts. Beyond transcriptomics, proteomics offers a complementary view by directly measuring proteins, which are the primary effectors of cellular function and key therapeutic targets. However, existing models either ignore the spatial information or the complex genetic and proteomic programs within cells. Thus they cannot infer how cell internal regulation adapts to microenvironmental cues. Furthermore, these models often utilize fixed gene vocabularies, hindering their generalizability to datasets with different genes than pretraining. In this paper, we introduce HEIST, a hierarchical graph transformer foundation model for spatial transcriptomics and proteomics. HEIST models tissues as hierarchical graphs. The higher level graph is a spatial cell graph, and each cell in turn, is represented by its lower level gene co-expression network graph. Rather than using a fixed gene vocabulary, HEIST computes gene embeddings from its co-expression network and cellular context. HEIST achieves this by performing both intra-level and cross-level message passing to utilize the hierarchy in its embeddings and can thus generalize to novel datatypes including spatial proteomics without retraining. HEIST is pretrained on 22.3M cells from 124 tissues across 15 organs using spatially-aware contrastive and masked autoencoding objectives. Unsupervised analysis of HEIST embeddings reveals spatially informed subpopulations missed by prior models. Downstream evaluations demonstrate generalizability to proteomics data and state-of-the-art performance in clinical outcome prediction, cell type annotation, and gene imputation across multiple technologies.

Batch effects pose a significant challenge in the analysis of single-cell omics data, introducing technical artifacts that confound biological signals. While various computational methods have achieved empirical success in correcting these effects, they lack the formal theoretical guarantees required to assess their reliability and generalization. To bridge this gap, we introduce Mixture-Model-based Data Harmonization (MoDaH), a principled batch correction algorithm grounded in a rigorous statistical framework. Under a new Gaussian-mixture-model with explicit parametrization of batch effects, we establish the minimax optimal error rates for batch correction and prove that MoDaH achieves this rate by leveraging the recent theoretical advances in clustering data from anisotropic Gaussian mixtures. This constitutes, to the best of our knowledge, the first theoretical guarantee for batch correction. Extensive experiments on diverse single-cell RNA-seq and spatial proteomics datasets demonstrate that MoDaH not only attains theoretical optimality but also achieves empirical performance comparable to or even surpassing those of state-of-the-art heuristics (e.g., Harmony, Seurat-V5, and LIGER), effectively balancing the removal of technical noise with the conservation of biological signal.

Background: Adolescence is a critical period of brain maturation and heightened vulnerability to cognitive and mental health disorders. Sleep plays a vital role in neurodevelopment, yet the mechanisms linking insufficient sleep to adverse brain and behavioral outcomes remain unclear. The glymphatic system (GS), a brain-wide clearance pathway, may provide a key mechanistic link.

Methods: Leveraging baseline data from the Adolescent Brain Cognitive Development (ABCD) Study, we examined whether GS function mediates the effects of sleep on brain structure, cognition, and mental health. GS function was indexed by perivascular space (PVS) burden derived from structural MRI. Participants (n ≈ 6,800; age ≈ 11 years) were categorized into sleep-sufficient (≥9 h/night) and sleep-insufficient (<9 h/night) groups. Linear models tested associations among sleep, PVS burden, brain volumes, and behavioral outcomes. Mediation analyses evaluated whether PVS burden explained sleep-related effects.

Results: Adolescents with insufficient sleep exhibited significantly greater PVS burden (Cohen's d ≈ 0.15), reduced cortical, subcortical, and white matter volumes, poorer cognitive performance across multiple domains (largest effect in crystallized intelligence, d ≈ 0.20), and elevated psychopathology (largest effect in general problems, d ≈ -0.34). Sleep duration and quality were strongly associated with PVS burden (p < 10^-8). Mediation analyses revealed that PVS burden partially mediated sleep effects on cognition (e.g., crystallized intelligence, episodic memory) and mental health (e.g., psychosis severity), with indirect proportions up to 10.9%. Sequential models suggested a pathway from sleep → PVS → brain volume → behavior as the most plausible route.

Conclusions: Insufficient sleep during adolescence is linked to glymphatic dysfunction, reflected by increased PVS burden, which partially accounts for adverse effects on brain structure, cognition, and mental health. These findings highlight the glymphatic system as a potential mechanistic pathway and imaging biomarker, underscoring the importance of promoting adequate sleep to support neurodevelopment and mental health.

Simulation-guided transcranial focused ultrasound therapies rely on estimating skull acoustic properties from pretreatment imaging. Typical clinical resolution (0.5 mm isotropic) cannot resolve bone microstructure, making the acoustic properties underdetermined and sensitive to modeling assumptions. Here, we examine how viscous and viscoelastic models predict changes in attenuation and phase velocity due to microstructure. Using viscous and viscoelastic k-Wave implementations, we simulated transmission of a broadband 625 kHz tone burst (250 kHz-1 MHz) through skull-mimicking digital phantoms. The phantoms contained spherical pores (0.1-1.0 mm diameter) randomly embedded within cortical bone (2.5%-90% porosity). Virtual sensors measured attenuation and phase velocity using a time-distance matrix approach. Both models predict increased attenuation with increasing pore size at a fixed porosity, but differ in the strength and porosity dependence of this relationship. The viscoelastic model generally predicts attenuation peaks at higher porosities than the viscous model. For 1.0 mm pores, the viscous peak (1.98 Np/cm) occurs at 20% porosity, while the viscoelastic peak (2.98 Np/cm) occurs at 70%. Phase velocity decreases with pore size for both models, though the viscoelastic predictions are less sensitive to pore size. These results demonstrate that viscous and viscoelastic models exhibit distinct attenuation and phase-velocity behavior for idealized bone microstructures. While both indicate that microstructure has a strong impact on attenuation, it has a lesser effect on phase velocity for the viscoelastic model compared to the viscous model. This work highlights the importance of acoustic model choice when estimating skull acoustic properties from computed tomography images. Future work will identify which acoustic model best represents ultrasound propagation through skull microstructure.