This article reviews methods that use radiofrequency (RF) magnetic field gradients to encode magnetic resonance imaging (MRI), offering an alternative to the static magnetic field gradients conventionally used to form images. Since RF encoding was first demonstrated in the 1970s, several approaches have been developed, with potential advantages for both MRI and magnetic resonance spectroscopy (MRS), which measures the chemical composition of tissues. However, challenges remain in hardware design, integration of these techniques into imaging systems, and their use in living subjects. This review summarizes major RF encoding methods, their potential benefits and limitations, and future directions for their technical development, research use, and clinical application.

Fig. 1.��PubMed count of journal articles containing keywords relating to RF encoding (1979–2026). Year of reference to seminal work of surveyed methods (i.e. RFI��, TRASE��, Bloch–Siegert��, SENF��, FREE��) are indicated. See search terms at��.

Epidermal growth factor receptor (EGFR) signaling helps regulate the growth and development of intestinal cells, but excessive activity can promote tumor formation. The protein leucine-rich repeats and immunoglobulin-like domains 1 (LRIG1) normally limits EGFR signaling, while the role of the related protein LRIG3 is less clear. Using genetically modified mice, researchers examined how loss of LRIG3 affects intestinal cells and colorectal tumor development, particularly when LRIG1 is reduced or absent. Unlike loss of LRIG1, which increased EGFR and Wnt signaling, loss of LRIG3 reduced activity in both pathways and altered the balance between intestinal stem and secretory cells. Removing one copy of LRIG3 in mice with reduced LRIG1 also substantially decreased colon tumor formation. These findings indicate that LRIG1 and LRIG3 have opposing effects that help balance EGFR signaling, maintain normal intestinal function, and influence the development of colon tumors.

Figure 1��Lrig1��and��Lrig3��expression in mouse and human intestinal tissue.��UMAP analysis indicating cell-type assignments for clusters generated based on canonical (A) mouse and (B) human gene markers, colored by gene expression levels of��Lrig1,��Lrig3, and��Lgr5, with corresponding dot plots indicating the mean expression level of��Lrig1��and��Lrig3��in each of the cell types listed. (C) Schematic detailing generation of��Lrig3^fl/fl��mice. To generate the��Lrig3^fl��allele, the��129 Lrig3��BAC clone was obtained from the Sanger Institute, and BAC recombineering was used to construct a targeting vector that placed LoxP sites 5′ and 3′ of exons 4–12 and a PGK-Neo cassette. Homologous recombination was used to modify the native��Lrig3��locus in embryonic stem cells, which were injected into blastocyst stage 129 mouse embryos by the Transgenic Mouse/ES Cell Shared Resource at Vanderbilt University. The resulting chimeric mice were bred to��FlpE��mice to remove the PGK-Neo cassette. SI, small intestine.

]]> /valiant/2026/10/28/sccotag-prior-informed-co-optimal-transport-for-robust-single-cell-multi-omics-integration/ Wed, 28 Oct 2026 17:23:17 +0000 /valiant/?p=7715

Li, Yang; Tan, Yuting; Chen, Rui; Zhou, Xin Maizie; Chen, You; Huo, Yuankai; Guo, Xingyi; Tiwari, Anshul; Wen, Zhexing; Zhong, Xue; Malin, Bradley A.; Li, Bingshan; Yin, Zhijun. (2026).��.��Proceedings of the ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 2-B, 11363–11374.��

Predicting the remaining useful life (RUL) of industrial equipment is important for planning maintenance and preventing failures. Many data-driven methods predict RUL directly from sensor measurements without explicitly modeling how equipment degrades over time. This study proposes a framework that instead separates the process into two stages: estimating the system’s current health and predicting its future degradation. The method uses sensor data to estimate health, continuously updates uncertain degradation parameters as new data become available, and produces a range of possible RUL values that reflects this uncertainty. A neural network helps adapt the estimation process while preserving uncertainty information. Testing on the N-CMAPSS benchmark showed competitive RUL prediction accuracy and statistically grounded uncertainty estimates across operating conditions, supporting the potential benefits of separating health estimation from future prediction in data-driven prognostics.

Fig. 1.��Overview of the proposed estimation–prediction framework. In the estimation stage, the system condition and degradation-related information are inferred from the available observations. In the prediction stage, performance-specific degradation models are used to obtain the predictive distributions of the times at which each health index reaches a performance level��𝑠∈[0,1]. These distributions are combined with the system threshold function and the desired confidence level to compute RUL predictions and associated confidence intervals.

Electronic health records are increasingly used to study factors associated with disease, but how researchers select comparison groups can influence the results. This study examined whether requiring imaging when selecting control patients could bias estimates of associations between metabolic factors and rotator cuff tears. Researchers compared two control groups: patients who had undergone imaging and those who had not. Controls with imaging were more similar to rotator cuff tear cases in characteristics such as arthritis and previous shoulder injury. Importantly, the estimated relationship between type 1 diabetes and rotator cuff tears reversed depending on the control group: type 1 diabetes was associated with higher odds of tears when controls without imaging were used, but lower odds when imaging was required. These findings demonstrate how conditioning control selection on imaging can introduce selection bias and potentially change conclusions drawn from electronic health record studies.

Figure 1

Directed acyclic graphs depicting collider stratification bias by conditioning on shoulder imaging. Individuals are indicated for shoulder imaging, such as MRI, by their provider once they present with pain or shoulder-related symptomatology. The reason for symptomatology could be due to either rotator cuff tear or other shoulder-related pathologies. Shoulder imaging is a descendant of a collider pain/symptomatology, and requiring shoulder imaging (conditioning on the descendant of the collider as depicted by the box around “Shoulder Imaging”) for control selection enriches the control population to have a larger proportion of other shoulder-related pathologies. Therefore, by restricting on imaging, an inverse correlation is created between rotator cuff tear and other shoulder pathologies (depicted by the orange dashed line). In such a study evaluating the relationship between an exposure and rotator cuff tears, if the shoulder-related pathologies are associated with exposure, type 1 diabetes (A) or type 2 diabetes, or obesity (B), then the association between the exposure and rotator cuff tear could be biased.

Patients with advanced non-small cell lung cancer whose disease progresses despite immune checkpoint inhibitor therapy have limited treatment options. This phase 2a study evaluated CAN-2409, a gene therapy injected directly into tumors, combined with the antiviral drug valacyclovir and ongoing immunotherapy. Among 73 patients who received treatment, median overall survival was 14.3 months. Among 46 patients who completed the protocol-defined treatment and 12-week imaging, median survival was 24.5 months, 10.9% had tumor shrinkage meeting response criteria, and 71.7% achieved disease control. Exploratory analyses also found signs of immune activation and regression of some tumors that were not directly injected. Most treatment-related side effects were mild to moderate, although 13.7% of treated patients experienced severe (grade 3) events; no grade 4 treatment-related events or treatment-related deaths were reported. These findings suggest potentially encouraging activity, but because this was a single-arm study and the results may be affected by patient selection and other biases, the treatment’s effectiveness must be confirmed in a randomized clinical trial.

Figure 1

Patient survival in evaluable (per-protocol) and safety populations. (a) Kaplan-Meier curve representing overall survival in months (from date of first injection) for evaluable patients in cohorts 1 and 2 (n=46). (b) Kaplan-Meier curve for overall survival analysis in months for all evaluable patients in cohort 2 (n=41). (c) Table of number of cohort 2 patients alive at different timepoints, illustrating the long tail of survival. (d) Kaplan-Meier curve for overall survival analysis for cohort 2 evaluable patients (n=41) by baseline PD-L1 status of <1% (negative) and >1% (positive).

This study investigates physics-informed neural networks (PINNs) for reconstructing the position of a magnetic-rail nScrypt printer extruder from accelerometer measurements. The researchers first developed a two-body physics model of the extruder and then incorporated its equations of motion into neural networks in two ways: directly within the network’s activation layers or as constraints during model training. The activation-based approach produced a more detailed fit to the position signal, while the constraint-based approach better captured its average trend and variability. The two approaches were also combined and evaluated using root-mean-square error and standard deviation. Although the combined model showed low variability during training, errors varied substantially on test samples. Overall, the study explores how incorporating physical equations into neural networks can support sensor-based signal reconstruction and digital twinning of a 3D printer extruder.

Fig 1

The printer head of mass��M��and nozzle mass��m��diagram with position and deflection coordinates.

Low-dose chest CT scans collected across multiple medical centers can differ in image resolution, reconstruction methods, and the amount of the body included in each scan, making quantitative measurements difficult to compare. We developed a standardization pipeline and applied it to more than 105,000 scans from 23,669 participants in the National Lung Screening Trial (NLST). The pipeline standardized image resolution, harmonized reconstruction differences across scanner manufacturers, and extended the field of view. Testing showed improved consistency in measurements of emphysema, skeletal muscle, and body fat across scans, including measurements taken over time. The approach reduced differences related to reconstruction methods and improved agreement with body measurements after field-of-view extension. This work provides a more standardized NLST imaging resource that could support reliable quantitative biomarker measurements and broader use of the dataset in lung cancer screening research.

Fig 1.��Acquisition heterogeneity in scans obtained from different vendors in the NLST.

The National Lung Screening Trial houses heterogeneous low-dose chest CT scans acquired from multiple institutions and vendors. Image variability arises from differences in through-plane spatial resolution, reconstruction kernels and semantic field-of-view. Reconstruction kernels influence the pixel noise and image texture in the underlying anatomy, creating differences in quantitative biomarker measures. In addition, scanner specific field-of-view configurations may truncate anatomy, affecting body composition assessment. These sources of heterogeneity motivate the need for a standardized dataset that enables consistent and reproducible quantitative image measures.

Qi, Yunzhe; Zhou, Qinghai; Liu, Boyang; Cui, Can; Zheng, Hua; Suo, Qiuling; Liang, Mingfu; Lov, Kenny; Liu, Xi; Jiang, Shali; Chen, Laming; Chen, Wen-Yen; Jin, Rong; Chang, Shuo; He, Jingrui; Li, Huayu. (2026).��.��Proceedings of the ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 2-B, 4000–4011.��

Brain organoids are laboratory-grown models of the brain, and reliably assessing their quality is important for research. Currently, this assessment often relies on researchers visually inspecting microscope images. We investigated whether measurable image features could help automate this process. Using 692 images from six culture batches, we compared several image-analysis approaches for predicting quality ratings assigned by an expert. A model based on interpretable, handcrafted image features performed well overall, although accuracy varied considerably between culture batches. These findings provide an initial foundation for automated quality assessment of brain organoids using standard microscope images. However, further studies are needed to confirm the consistency of expert quality ratings and determine how well the approach performs across different laboratories and culture conditions.

Figure 1.��Motivation and study framework for interpretable brightfield organoid quality assessment. (A) Routine manual QA is based on qualitative observations of compactness, boundary clarity, and internal appearance. (B) Organoid appearance changes across developmental stages and culture batches, and images obtained on different culture days represent independent sampling events rather than longitudinal tracking of the same individual organoid. (C) The study asks whether these visual observations can be translated into interpretable optical descriptors across developmental stages and independently within maturation. (D) Handcrafted feature families are compared with frozen ImageNet-pretrained ResNet-50 and self-supervised DINOv2 representations using logistic-regression and random-forest classifiers. (E) Evaluation includes repeated QA-stratified cross-validation, leakage-safe preprocessing, maturation-only analysis, and leave-one-batch-out testing.