Papers for

medical imaging analysts

Papers whose findings have a practical use for this group, as judged from the abstract. Open a paper to read what it means in practice.

Synthetic dataset and method improve ice layer reconstruction in radar images

SyntheIce: A Synthetic Radargram Annotation Benchmark for Englacial Layer Reconstruction

Abstract: Englacial layers in radio-echo-sounding radargrams (subsurface ice images) encode the climate and flow history of ice sheets and glaciers. However, large-scale extraction is hindered by time-consuming manual annotation and semi-automated detectors that produce fragmented rather than continuous horizons. We address two challenges: reconstructing complete layers from partial detections across radargram profiles and validating layer continuity across adjacent tile boundaries. We introduce a slope-aware greedy fragment-linking algorithm that reconstructs full-width layers using row-similarity gating and an exponentially weighted moving average (EWMA) slope predictor. Each reconstruction receives a content-based hash ID for stable cross-radargram identification. For quantitative evaluation, we generate SyntheIce, a synthetic dataset of 600 radargrams spanning multiple complexity levels and realistic noise conditions, including 150 wide radargrams divided into four adjacent tiles. All samples include exact ground-truth annotations. We evaluate the dataset using XCheck, a consistency-based validation framework, extending its original two-image setting to four consecutive tiles and enforcing continuity at every boundary. This demonstrates the dataset's utility for developing and evaluating real-world annotation algorithms. Our findings extend beyond ice science: when spatial features are fragmented but retain consistent local geometry, trajectory-based linking is more reliable than appearance-based matching. Evaluation against expert-labeled real radargrams demonstrates high precision, recall, F1-score, and mean IoU, supporting the method's transfer to real-world data.

Tue 10 NovComputer Vision and Pattern Recognition
The gist
Scientists need to find and connect layers inside glaciers from radar images, but these layers are often broken up or only partly visible, making the task slow and hard. The authors created a new computer method that links partial layers into full, continuous ones by using the layers’ slopes and nearby patterns, assigning IDs to keep track across images. To test their method, they made a large set of fake radar images with exact layer markings, helping measure how well an algorithm works. Their tests show the method works well on real glacier images too, making it easier to study ice and past climate.
Open → 2611.14728v1

High-dimensional counterfactual image reasoning without inversion step

Inversion-Free High-Dimensional Counterfactual Inference

Abstract: Counterfactual inference is the task of answering hypothetical "what-if" queries about a unit, given an observational dataset, modeling assumptions, and a causal graph. This enables a modeler to reason about hypothetical scenarios that cannot otherwise be answered without assumptions or a causal graph. The simpler case involves a causal graph with scalar 1-D variables that denote different attributes in the data-generation process. While the literature is clear in the 1-D case, extending this to settings with high-dimensional nodes (e.g., images) remains challenging. Conventionally, in the literature, methods recover the latent state from the factual image (inversion), generate counterfactual attributes from the graph, and render the counterfactual image from the latent state using the updated attributes. In our work, we investigate the necessity of the inversion step in computing high-dimensional counterfactuals. We first discuss moving from the scalar 1-D case to the higher-dimensional case. Secondly, we discuss the relevance of the inversion step and evaluate an inversion-free counterpart repurposed from the image-editing literature. Finally, on image datasets such as CelebA and ADNI, we benchmark our approach using two measures: namely, effectiveness and identity preservation. While effectiveness is a measure of how faithfully the counterfactual attribute is rendered, identity preservation measures whether the same unit (image) is being rendered, both of which are crucial for high-dimensional counterfactual inference. We present the tradeoffs of using inversion-based and inversion-free approaches in our work.

Mon 9 NovMachine Learning
The gist
Counterfactual inference means asking 'what if' questions based on existing data to imagine different possible outcomes. Usually, this involves changing simple characteristics, but doing this with complex data like images is hard. The authors study whether a common step—finding the hidden cause behind an image—is necessary for imagining changes in images. They compare methods with and without this step using face and brain image datasets, measuring how well changed traits appear and if the original subject stays recognizable. They show tradeoffs between approaches, helping guide better ways to answer 'what if' questions for images.
Open → 2611.12815v1

Text guided image learning improves anomaly detection accuracy

TIRAD: Text-guided Image Residual Learning for Generalist Anomaly Detection

Abstract: Recent anomaly detection studies have shown that residual features, which model the discrepancy between normal and abnormal patterns, provide strong generalization to anomalies unseen during training. However, existing residual-based approaches mainly rely on visual discrepancies derived from image representations, while semantic discrepancies that can be explicitly expressed through language remain largely underexplored. To address this problem, we propose TIRAD, Text-guided Image Residual Learning for Generalist Anomaly Detection, a multimodal residual learning framework that defines the discrepancy between textual representations of normal and abnormal states as a Text Residual Feature and uses it to guide the Image Residual Feature. Specifically, a Text Fusion Module converts patch-level text residuals into a gate that modulates the image residual of each patch. The gate retains the visual discrepancies that the text residual identifies as defect-related and attenuates the others, leading to more discriminative anomaly representations. Extensive experiments on industrial benchmarks and industrial-to-medical transfer demonstrate that the proposed method outperforms existing image-centric residual learning methods.

Mon 9 NovComputer Vision and Pattern Recognition
The gist
Detecting unusual or defective parts in images can be tricky, especially when the system hasn't seen those exact problems before. The authors introduce a method called TIRAD that combines both what images look like and what is described in text to spot differences between normal and defective items. By using text descriptions to focus on relevant visual details, the method better identifies anomalies. This approach works well on industrial and medical images, beating older methods that only look at pictures.
Open → 2611.12023v1

Foundation models reveal hidden clinical features from multimodal patient data

Glyph: An Interpretability Engine for AI-Driven Scientific Discovery in Clinical Data

Abstract: Automated scientific discovery has the potential to advance understanding of disease mechanisms and accelerate identification of early disease markers. Manual data annotation is infeasible at population scale, motivating methods that automatically extract pre-defined phenotypes in multimodal and unstructured clinical data. Such pre-defined feature sets determine and limit which associations these approaches can discover. This limitation extends to agentic AI systems that use these tools to query population-scale data. In contrast, foundation models (FMs) trained on large volumes of clinical data can perform tasks beyond their training objectives. This suggests their latent embeddings may represent novel or unexplored features that could yield insights into disease mechanisms, yet such latent features remain difficult to interpret. To address this challenge, we introduce Glyph, an automated multistage framework that extracts features from multimodal and unstructured clinical data using FMs and sparse autoencoders, interprets them as coded clinical concepts, and evaluates their value for clinical scientific discovery. We applied Glyph to abdominal CT volumes, CT radiology reports and laboratory measurements from 41,628 patients across different institutions. We identified 397 predictive concepts supported by 4,463 Bonferroni-significant associations. Moreover, a majority of discovered associations were automatically reproducible in held-out cohorts across different sites. Glyph concepts provided complementary associations to literature-derived markers, leading to gains in predictive power. We envision that interpretability can extend beyond a mechanistic understanding of neural networks to connecting implicitly learned features with clinical knowledge and establishing their significance for patient outcomes. We release Glyph at https://github.com/StanfordMIMI/Glyph.

Fri 6 NovArtificial Intelligence
The gist
Automatically finding helpful features in large amounts of medical data is tough because existing methods rely on known labels or features. The authors propose Glyph, a tool that extracts new clinical features from complex inputs like CT scans and lab results using advanced AI models, then links these features to known medical concepts. Tested on over 40,000 patients, Glyph found many useful features that predict disease outcomes better than traditional markers and confirmed these findings in different hospitals. This approach helps explain what the AI learns and connects it to real medical knowledge.
Open → 2611.08922v1

DART improves anomaly detection across unseen and varied domains

DART: Domain-Agnostic Residual Transfer for Generalist Anomaly Detection

Abstract: Generalist Anomaly Detection (GAD) aims to detect anomalies in unseen domains using only a few reference samples. Recent normal-abnormal guided methods improve this setting by exploiting abnormal references, but their residuals often mix transferable anomaly cues with domain-specific appearance factors such as texture, geometry, and acquisition conditions. This residual domain entanglement limits cross-domain transfer and may produce spurious activations on novel domains. We propose DART, a plug-in framework for normal-abnormal guided GAD that learns to separate residuals into anomaly-relevant and domain-specific components. DART combines a Residual Disentanglement Module, which preserves anomaly-discriminative information while suppressing domain information, with Cross-Domain Meta-Training, which forms episodes across source domains to make the separation effective. The method is architecture-agnostic and can be inserted between residual computation and the detection head of existing GAD pipelines. Experiments on MVTec AD, VisA, and BraTS show consistent improvements over multiple base methods, with the largest gains under stronger domain shifts. https://github.com/aqeeelmirza/DART

Thu 5 NovComputer Vision and Pattern RecognitionArtificial IntelligenceMachine Learning
The gist
Detecting unusual or faulty items in new settings is hard because the appearance can change a lot. The authors introduce DART, a method that separates what makes something truly abnormal from normal differences between settings like texture or lighting. This separation helps the system better spot problems even when it hasn’t seen those exact conditions before. They tested DART on several benchmarks and found it consistently improved existing methods, especially when the differences between conditions were bigger.
Open → 2611.08271v1

Calorimeter flows detect small localized particle signals better

Localized anomaly detection with a calorimeter normalizing flow

Abstract: Normalizing flows trained as fast calorimeter simulators also give the likelihood of each shower, which can serve as an anomaly score. Because this likelihood sums over every voxel, a signal confined to a few voxels can be lost in the fluctuations of the rest of the shower. For example, in a public benchmark dataset in which a long-lived neutral particle decays to two photons, the CaloFlow likelihood used in previous work selects less than 5% of the signal at a background efficiency of $10^{-3}$ when the particle has the largest simulated mass and decays at or just behind the front face of the calorimeter. In this work, we define a regional anomaly score by summing the squared latent coordinates of the same flow within small detector regions around the point where the particle enters the calorimeter, with the threshold set using background showers alone. At the same background efficiency, this regional anomaly score selects more than 95% of the signal. The per-coordinate likelihood terms, summed over the same regions, perform equally well, but the squared latent coordinates have a known distribution for background showers and clearly illustrate when a shower departs from it. Therefore, normalizing flows may be sensitive to new physics, even if the global likelihood is not.

Fri 30 OctMachine Learning
The gist
Detecting unusual particles in particle detectors can be hard when their signals get lost in noise from other parts of the detector. The paper shows that instead of looking at the entire detector signal at once, analyzing small regions where the particle enters helps find rare signals more accurately. The authors use a machine learning method called normalizing flows to do this regional analysis, dramatically improving detection of certain particle decays. Their method works better than previous approaches by focusing on smaller, relevant areas.
Open → 2611.00234v1

Tensor regression method improves robustness for complex data analysis

Partial Modal Tensor Regression

Abstract: With the rapid development of modern scientific research, tensor data have become increasingly prevalent across a wide range of fields. How to effectively characterize the relationship between tensor predictors and a response while preserving the intrinsic structure of tensor data has therefore become an important problem in high-dimensional statistical analysis. Existing tensor regression methods mainly focus on conditional means or conditional quantiles. When the response distribution is asymmetric, heavy-tailed, or contaminated by outlying observations, the conditional modal provides a robust and interpretable alternative characterization. In this paper, we propose a partial modal tensor regression (PMTR) method. The proposed method constructs the partial modal tensor covariance, which incorporates information on the conditional modal to guide tensor dimension reduction, and subsequently establishes a modal tensor regression model in the reduced tensor space. This strategy alleviates the dimensionality burden caused by direct vectorization while preserving the intrinsic structural information of tensor predictors. Under suitable regularity conditions, we establish the consistency and asymptotic properties of the proposed estimators and employ a data-driven dimension-selection procedure to determine the effective dimension along each tensor mode.

Sun 25 OctMachine Learning
The gist
Many modern studies generate data that look like multi-dimensional arrays called tensors. Understanding how these tensors relate to an outcome is tricky, especially when data have unusual patterns or outliers. The authors propose a new method that focuses on finding the most likely outcome (the mode) rather than the average, which is often more stable in complex data. They reduce the complexity of the tensor data while keeping its natural structure, making the relationship easier to understand and analyze accurately.
Open → 2610.34008v1

Multi-domain approach identifies linked hidden factors from mixed signals

Identifying Dependent Components from Multi-domain Linear Mixtures

Abstract: We study a linear mixing model with dependent latent components, assuming multiple data domains. Most existing models assume that the components are independent or at least uncorrelated, in line with independent component analysis (ICA). Some recent work allows for dependent components, but then makes specific assumptions such as parametric forms of dependencies, multi-view settings, or interventions, or does not recover the individual components. In contrast, we consider a multi-domain setting in which domains differ through domain-specific scalings of the components, while the distribution of the underlying latent components is the same across domains. This approach can model data collected, for example, from different sensors measuring the same process, different laboratories conducting the same experiment, different experimental conditions, or different subjects that might differ in biological or physiological factors. We show that, under sufficient domain variability, latent variables and mixing functions can be identified from second-order statistics alone. We propose the Multi-Domain Covariance Matching (MuDo-CoM) algorithm that generalizes previous methods of joint diagonalization. MuDo-CoM is validated on simulated data and a real-world fMRI dataset.

Tue 20 OctMachine LearningArtificial Intelligence
The gist
When you have data from different sources measuring the same thing, the hidden factors they measure can be tangled up and related. Most techniques assume these hidden factors act independently, but the authors study a way to uncover linked hidden factors across different data sources. They do this by looking at how the variation changes across sources rather than assuming independence. Their method, MuDo-CoM, uses these differences to successfully find both the hidden factors and how they mix to form the observed data.
Open → 2610.28330v1

Radiomic feature choices influence glioblastoma survival predictions

How Robust Are Prognostic Radiomic Features? Comparing Selection Techniques on Pooled Glioblastoma Cohorts

Abstract: Radiomic survival studies in glioblastoma report heterogeneous prognostic features, but how much of this stems from the methodology itself is rarely examined. We analysed the association of 463 radiomic, anatomical and clinical features from pre-treatment MRI with overall survival in 705 IDH-wildtype glioblastoma patients from three public cohorts, using three techniques: a log-rank test after median split, univariate Cox regression (both Benjamini-Hochberg corrected) and an XGBoost regressor interpreted with SHAP. The log-rank test selected 99 features, Cox regression 72 and SHAP its top 20; five were selected by all three: age, a peritumoural T2 texture feature, the number of connected peritumoural mask components and two image-level intensity maxima. The two survival tests shared 51 features, about half of the SHAP top 20 were also selected by either test, and where selections overlapped, effect directions agreed. Many selections were PyRadiomics diagnostic descriptors that are not restricted to tumour voxels. Analytical pipeline choices, including ranking method, pre-filtering and descriptor handling, can substantially influence which features appear prognostic.

Mon 19 OctComputer Vision and Pattern RecognitionArtificial Intelligence
The gist
Glioblastoma is a serious brain cancer, and doctors often use MRI scans to find features that might predict how long patients will survive. This paper looked at hundreds of such features from MRI scans of many patients and tested three different methods to see which features seemed important. They found that different methods picked very different features, although a few key ones were consistently selected by all methods. This means that how researchers choose and analyze these MRI features can greatly affect the predictions made about patient survival.
Open → 2610.26095v1

Multimodal approach improves Alzheimer's diagnosis with missing data handling

CoPoE: Multimodal Fusion via Decomposable Disease-Coordinate Product-of-Experts for Missing-Modality Alzheimer's Diagnosis

Abstract: Multimodal Alzheimer's disease (AD) diagnosis benefits from integrating heterogeneous clinical, imaging, genomic, and biomarker evidence, but clinical cohorts frequently suffer from irregular modality missingness. Existing fusion methods often synthesize absent inputs, risking the introduction of artificial surrogates, or pool available signals into uninterpretable latent spaces. We present CoPoE (Disease-Coordinate Product-of-Experts), a disease-coordinate framework that maps multimodal evidence into a structured latent space partitioned into four distinct biological and clinical axes: genetic Risk, molecular Pathology, Neurodegeneration, and clinical Stage (R/P/N/S). Each observed modality parameterizes a diagonal Gaussian expert over the full RPNS vector, and a masked Product-of-Experts architecture fuses only the available modalities. Consequently, absent modalities add no factor to the fusion path, allowing the network to preserve a robust, decomposable posterior for any non-empty modality subset without synthetic imputation in the RPNS path. Through extensive missing-modality experiments on the ADNI dataset, CoPoE achieves the best all-modality performance and the highest mean AUROC across all 15 observed-subset evaluations among standardized missing-modality fusion baselines under a shared non-PET ADNI embedding benchmark, while substantially improving raw-probability ECE, Brier score, and NLL. Furthermore, PET-supervised probing shows evidence enrichment within the pathology (P) block under full modalities, with tau-related signal retained even when direct fluid biospecimen inputs are withheld. Our code is available at https://github.com/labhai/CoPoE.

Thu 8 OctMachine LearningComputer Vision and Pattern Recognition
The gist
Diagnosing Alzheimer's disease can use different types of medical data, like brain scans, genetics, and clinical tests. However, often some of this data is missing for patients, which makes diagnosis harder. The authors created CoPoE, a method that can combine whatever data is available without trying to guess the missing parts. This new way better predicts Alzheimer's and gives clearer insights into different biological processes involved in the disease.
Open → 2610.11394v1

One-shot method adapts vision models for varied scientific images

One-Shot Adaptive Segmentation For Scientific Images

Abstract: Scientific image segmentation methods rely on extensive annotation and task-specific training, limiting adaptation across imaging modalities and experimental conditions. We present a training-free, one-shot framework that specializes vision foundation models using a single annotated reference image. The framework combines DINOv3 representations with background-adaptive feature orthogonalization to suppress artifact-related feature directions, after which cosine similarity localizes candidate regions for SAM segmentation. We evaluate the framework on red-blood-cell microscopy, structured-illumination pool boiling, and chest radiography. Relative to the strongest baseline, the proposed method improves mean IoU by 5.91% and 78.62% on the microscopy and pool-boiling datasets, respectively, while achieving comparable performance on chest radiographs. These results demonstrate that one-shot reference conditioning can adapt general-purpose vision models to specialized scientific segmentation tasks.

Wed 7 OctComputer Vision and Pattern Recognition
The gist
Segmenting scientific images usually needs lots of labeled examples and training for each kind of image. The authors created a method that only needs one labeled image to adjust general vision models to new scientific images without retraining. Their technique uses smart features to ignore unwanted parts and then finds regions to segment. It worked well on different types of scientific images, improving accuracy especially where previous methods struggled.
Open → 2610.10306v1

Geometry shapes memory limits in compressed associative storage

A Geometry-Based Capacity Theory for Finite-Feature Associative Memory

Abstract: We develop a geometry-based capacity theory for exact-key retrieval in compressed finite-feature Hebbian associative memory. For random or approximately isotropic values, retrieval interference separates into finite-feature noise, which decreases with feature dimension, and structural interference, which is determined by squared kernel overlap among stored keys and persists in the infinite-feature limit. This yields a fit-free prediction of retrieval quality, reveals a geometry-dependent capacity ceiling, and predicts the feature budget required for a target retrieval quality. When stored values are correlated, we show that retrieval depends jointly on the key kernel and value Gram matrix, and derive finite-feature approximations that account for this interaction. We validate the theory on synthetic, visual, and medical-image representations. Overall, the framework links representation geometry directly to memory capacity and distinguishes when performance can be improved by increasing the feature budget and when the representation itself must be changed. Across these settings, the predicted retrieval curves closely match empirical behavior and correctly identify changes in the preferred memory design.

Tue 6 OctMachine Learning
The gist
This paper explains how the shape and structure of data affect the performance of memory systems that store and retrieve information using compressed features. The authors show that interference during memory retrieval comes from two sources: noise that decreases as features grow, and overlap in data that persists. Their theory predicts exactly how much memory can store before errors rise, and when changing the data representation is better than adding features. They test their ideas on various data types like images and find the predictions match real results closely.
Open → 2610.09056v1

Corpus callosum shape changes with aging differ in Alzheimer's disease

Riemannian Shape Analysis of the Corpus Callosum in Kendall Space: Aging and Alzheimer's Disease

Abstract: The corpus callosum (CC) is a major white-matter structure and a well-established marker of brain aging, but most studies quantify it using scalar summaries that discard its boundary geometry. We present a Riemannian shape-space framework for analyzing age-related morphological change in the midsagittal CC, applied to the OASIS-1 cohort. Each contour is represented by $128$ landmarks and embedded into Kendall shape space, where translation, rotation, and scale are removed. We derive a multivariate geodesic regression with exact Riemannian gradients and use the fitted age-velocity field to localize age-related deformation to five anatomical sub-regions. In the cognitively normal cohort ($n = 252$), geodesic regression outperforms the Euclidean linear benchmark ($R^2 = 0.1355$ vs.\ $0.1216$). Regional energy is posterior-dominant: the Splenium carries $41.4\%$ and the Isthmus $23.8\%$ of total age-related shape change, together accounting for $\sim 65\%$ despite comprising only $\sim 35\%$ of landmarks. Signed projections confirm the ordering (Splenium $r = 0.570$; Isthmus $r = 0.473$). In contrast, age explains less than $0.5\%$ of shape variance in Alzheimer's disease ($n = 88$), indicating that the disease disrupts the healthy aging trajectory. A tangent-space classifier achieves an age-group AUC of $0.791$ from the 2D contour alone, exceeding a recent volumetric benchmark ($0.67$).

Sun 4 OctComputer Vision and Pattern Recognition
The gist
The corpus callosum is a brain structure that changes shape as people age. This paper studies detailed shape changes using mathematical methods that focus on the edges of this structure, not just its size. The researchers found that certain parts of the corpus callosum show the most change with healthy aging, but these patterns are disrupted in people with Alzheimer's disease. They also showed that these shape details can help distinguish different age groups better than measuring volume alone.
Open → 2610.05326v1

Concept aware saliency maps improve visual model explanations

ConEx: Human-Interpretable Saliency Maps via Concept-Aware Attribution

Abstract: Many visual explanation methods in computer vision highlight pixel importance but struggle to link these low-level cues to semantically meaningful concepts, limiting their interpretability and trustworthiness. We introduce Concept-based Explanations (ConEx), a novel framework that bridges saliency visualization with concept-based reasoning to provide both faithfulness and interpretability. ConEx automatically discovers class-specific concepts and represents them through concept activation vectors (CAVs), learned without manual supervision using an architecture-specific masking mechanism that reduces noise introduced by the segmentation masks to enhance concept purity. ConEx generates faithful saliency maps that reveal where each concept appears in the image and how it contributes to the prediction. To evaluate the reliability of these learned concepts, we propose two complementary metrics, Vector-Concept Match (VCM) and Concept-Class Match (CCM), that quantify concept alignment and enable direct comparison with existing methods. Extensive experiments across diverse settings demonstrate that ConEx achieves state-of-the-art performance on faithfulness, segmentation, and concept-quality benchmarks. Overall, ConEx advances the field toward truly interpretable and concept-grounded explanations in vision models.

Sat 3 OctComputer Vision and Pattern RecognitionArtificial IntelligenceMachine Learning
The gist
Many tools show which pixels in images influence a computer vision model’s decisions, but these highlights often lack clear meaning. The authors present ConEx, a new method that automatically finds meaningful visual concepts specific to each class and links these to areas in the image. ConEx creates clearer, concept-focused maps that help people understand how different parts of an image contribute to a model's prediction. They also introduce new metrics to measure how well these concepts align with actual image features and classes. Their tests show that ConEx provides more faithful and interpretable explanations than prior methods.
Open → 2610.04605v1

Label agreement misses key authorization and completeness issues

Label Agreement Does Not Measure Authorization

Abstract: Many groups now delegate label ontology and metadata harmonization to agentic LLM pipelines. We built one and audited it. Our aggregate scores looked healthy, but the pipeline kept failing in ways they did not show, so we set out to find what they hid. Label agreement asks whether a proposed label matches a reference. It does not ask whether the agent was entitled to propose it, whether the output was complete enough to act on, or whether the label moved when the evidence moved. We measured those three separately on COBRE and FBIRN, two schizophrenia and control neuroimaging cohorts from different consortia, and they come apart, from label agreement and from each other. Showing the agent an upstream proposal barely moves label agreement, 0.857 to 0.870, while agreement on the chosen action doubles, 0.409 to 0.830. Output that parses as JSON still drops a required field on 10% of one model's cases and 33% of the other's. And an agent that replays its first answer scores perfectly on original cases and zero once we change the evidence that decides them. Downstream, a row-order error that none of these metrics reports erases most of the diagnostic signal. So we measure these properties apart, pair each with a control, and gate commitment on the result, which makes failures visible and easy to route to a person. None of this prevents failure. Our reference labels are rule-derived, so agreement with them means consistency, not correctness. Code is available at https://github.com/amir-sbg/Label-Agreement-Does-Not-Measure-Authorization.

Sat 3 OctArtificial IntelligenceMachine Learning
The gist
Label agreement checks if two labels match, but it doesn’t show if the label was allowed, complete, or sensitive to changes in evidence. The authors built a pipeline using large language models to label neuroimaging data and discovered that while label agreement scores looked good, other important measures revealed many failures. For example, sometimes required data fields were missing or the label did not update when the input changed, and these problems were hidden by usual agreement metrics. The authors suggest measuring these aspects separately and involving humans when needed to catch errors.
Open → 2610.04544v1

Accuracy measures improved for small samples in machine learning

When Is Accuracy Evidence? A Unified Theory of Generalisation, Validation, and Information Fusion

Abstract: K-fold cross-validation (CV) is widely used as evidence of out-of-sample performance, although folds are neither independent experiments nor equally informative under heterogeneous data. Cross Upper-Bound Validation (CUBV) replaces point-wise CV accuracy by conservative upper bounds on true risk. Here we generalise CUBV through a single exponential framework in which the moment-generating function of the generalisation gap is controlled by a cumulant envelope gamma(lambda). This yields a family of risk bounds covering Hoeffding-, Bernstein-, dependency-aware, PAC-Bayesian, and heterogeneous source-fusion settings. For K-fold CV, dependence between fold-wise gaps is modelled through a joint sub-Gaussian proxy matrix. Under equicorrelation, this gives an effective number of folds, Keff = K/[1+(K-1)rho], showing that increasing K does not necessarily increase statistical evidence when folds are strongly dependent. The framework is also extended to posterior distributions over predictors and weighted multi-source fusion, where weights are selected by minimising an upper bound on future risk rather than empirical error alone. Experiments with trained linear classifiers on heterogeneous multimodal Gaussian mixtures compare K-fold CV with full-sample resubstitution plus risk correction. Bounds are evaluated by coverage and tightness. In low-dimensional small-sample settings, K-fold partitioning can increase uncertainty because individual folds under-represent minority modes, while corrected resubstitution can remain valid and tighter; this effect disappears as sample size increases. Overall, gamma-CUBV separates observed performance, uncertainty, dependence, model complexity, and confidence into explicit terms, providing a unified route from CV scores to risk statements and a principled validation criterion for heterogeneous small-sample applications such as neuroimaging.

Fri 2 OctMachine Learning
The gist
Evaluating how well a machine learning model performs is tricky, especially when using limited or varied data. The authors developed a method that better estimates the true error of models by accounting for dependencies and differences in data groups. This approach prevents overconfidence in results from common methods like cross-validation, especially with small or mixed data sets. Their work helps users understand when accuracy scores genuinely reflect how well a model will do on new data.
Open → 2610.03465v1

Reeb transform captures shape changes in 3D but not higher dimensions

The Reeb Transform

Abstract: Reeb Transforms offer a compact representation of how shapes in $\mathbb{R}^n$ change when sliced along varying directions. We define the Reeb Transform as the family of Reeb graphs induced by height functions along all directions in the unit sphere. In this paper, we develop a rigorous treatment of Reeb Transforms for o-minimal definable sets, with particular emphasis on one-dimensional stratified spaces embedded in $\mathbb{R}^d$ and on surfaces in $\mathbb{R}^3$. We establish injectivity of the Reeb Transform in multiple settings, including compact surfaces in $\mathbb{R}^3$, capturing the essential topological features needed to uniquely reconstruct such surfaces. However, in dimensions above three, the Reeb Transform ceases to be injective, indicating the limitations of this descriptor in higher-dimensional settings.

Fri 2 OctComputational Geometry
The gist
Understanding how shapes change when sliced is useful in many areas like graphics and geometry. This paper defines a way to describe all slices of a shape, called the Reeb Transform. The authors show that for surfaces in 3D space, this description is detailed enough to reconstruct the shape uniquely. However, in spaces with more than three dimensions, this approach no longer works perfectly, missing some differences between shapes.
Open → 2610.03126v1

Simulation method improves trust in brain tissue MRI estimates

Trusting the Inverse: Reliability-Aware Mapping for Simulation-Based Microstructure Estimation in Diffusion MRI

Abstract: Diffusion-weighted MRI can probe tissue microstructure non-invasively, however interpreting microstructural parameter estimates remains challenging due to the intrinsic ambiguity of the inverse problem. While simulation-based approaches can incorporate increasingly realistic tissue models, they still lack voxel-wise characterisation of the reliability and degeneracy of the inferred parameters. Here, we introduce a reliability framework for simulation-based microstructure estimation based on three complementary scores that identify distinct sources of unreliability in the estimation process: out-of-distribution signals, local signal mismatch, and parameter degeneracy. The framework was implemented using a Monte Carlo dictionary of 1,050 synthetic voxels generated from geometrically realistic substrates, with parameter ranges grounded in electron microscopy measurements of rat corpus callosum. The dictionary spans biologically plausible axon radii (0.25-0.85 $μ$m), microscopic angular spread (0-10$^\circ$), packing densities (60-92%), and intrinsic diffusivities (1.75-3.0 $μ$m$^2$/ms). On synthetic data, the resulting Reliability Index correlated with actual estimation error (Spearman $ρ= -0.742$) and distinguished between extrapolation, poor local interpolation, and parameter degeneracy. Applied to in vivo corpus callosum DW-MRI in rat (256 voxels, four animals) and human (MGH-USC HCP, 18,765 voxels, nine subjects), 91% and 73% of voxels respectively exceeded $R > 0.5$. These results demonstrate how reliability-aware analysis can support the interpretation and future development of simulation-based diffusion MRI microstructure models.

Wed 30 SeptComputational Engineering, Finance, and Science
The gist
Diffusion MRI helps us see tiny structures in the brain without surgery, but it's hard to trust the numbers it gives because different tissue types can look similar. The authors developed a way to measure how reliable each estimate is by using computer simulations that mimic real brain tissue. Their method gives scores showing when an MRI scan might be confusing or unclear, helping doctors and researchers know which measurements they can trust. They tested this approach on brain images from rats and humans, showing it can tell apart trustworthy from unreliable results.
Open → 2609.38922v1

Spectral filtering improves brain data models for personalized prediction

Flattening the Connectome Spectrum: A Spectral Filter for FC Induces a Pretraining Target for fMRI Encoders

Abstract: Self-supervised pretraining reshaped prediction in language and vision, and brain foundation models (BFMs) inherited its promise. Representations learned from large unlabelled corpora should capture individual functional dynamics and generalise across cohorts. However, kernel ridge regression (KRR) fitted on functional connectivity (FC) matrices still predicts individual phenotypes more accurately than any BFM we tested. In this paper, we show that KRR is weighted by the eigenvalues of the FC which are miscalibrated for phenotype prediction. We apply an efficient spectral filter to recalibrate the eigenvalues of each subject's FC matrix, enabling the model to exploit more inter-individual variance. Across the 5 datasets, 11 parcellations and 6 prediction targets we tested, we match or exceed the KRR baseline. Based on this finding, we then pretrain a small encoder model on about 4,000 hours of fMRI from 162 open datasets, whereby we align the pairwise similarities between the embeddings of recording snippets with those between the recalibrated connectomes. Our model performs on par with the best of the 6 published BFMs we tested while having an order of magnitude fewer parameters. Our encoder performs better than FC on short scans and in smaller cohorts, especially in fingerprinting. We release the pretrained model weights, the code and the pretraining data, preprocessed and parcellated.

Tue 29 SeptArtificial IntelligenceMachine Learning
The gist
Functional MRI scans show how different parts of the brain connect, but predicting individual traits from this data can be tricky. The authors found that adjusting the mathematical properties of brain connectivity data helps models detect personal differences better. They used this adjustment to train a smaller brain data model that works as well as bigger models and does better with shorter scans and smaller groups. They also shared their model and code so others can use this improved approach.
Open → 2609.37642v1

Volcanite enables fast detailed 3D segmentation viewing on regular computers

Volcanite: Commodity-Hardware Segmentation Volume Visualization for Connectomics and Beyond

Abstract: Modern imaging produces terabyte-scale segmentation volumes, assigning each voxel an object label. These categorical, boundary-sensitive and label-rich data underpin connectomics and other imaging-driven fields, yet their scale often forces interpretation through slices, approximate meshes or distributed workflows that obscure spatial context and voxel-level defects. Here we show that such volumes can be explored directly on commodity hardware with Volcanite, an open-source framework for dense-segmentation rendering. Combining compression-aware data handling, a Vulkan GPU backend and segmentation-specific rendering, Volcanite enables low-latency exploration without meshing or distributed infrastructure, including for unpublished, sensitive or proprietary data. Across multi-domain datasets, it preserves voxel labels, adds shadows and global illumination, and renders up to half a trillion voxels at 100 frames per second. By replacing lengthy preprocessing with direct inspection, Volcanite decouples hypothesis generation from data preparation and turns teravoxel label fields into interactive evidence for discovery, validation and cross-domain spatial analysis

Tue 29 SeptGraphics
The gist
Very detailed 3D images from scientific imaging can be huge, and viewing them usually requires complex software or hardware. The authors created Volcanite, a program that lets you explore these massive 3D segmented images directly on ordinary computers without complicated processing. It keeps all the small details and labels intact, showing shadows and lighting to help with understanding. This makes it easier and faster for people to study these images and find important features without waiting for long setups.
Open → 2609.36898v1

Multimodal model splits task signals from noise for better prediction

Structured Latent Modeling for Supervised Multimodal Information Decomposition

Abstract: Multimodal prediction relies on diverse forms of evidence: information repeated across modalities, cues specific to a single source, and complex cross-modal dependencies that emerge only when inputs are considered together. While recent methods promote richer interactions, they lack a principled way to isolate these target-relative contributions within learned continuous representations. We introduce a framework that applies contrastive or masked objectives at intermediate layers, coupled with source-wise invertible normalizing flows and a supervised, low-rank latent variable model. This architecture explicitly factorizes the joint distribution into shared task-relevant variation, modality-specific predictive variation, and task-irrelevant dependence. Drawing connections to prior multimodal learning assumptions, our approach evaluates how modalities independently and jointly contribute to the target. Ultimately, this framework unites intermediate representation learning with structured likelihood-based guidance, offering a practical latent-variable lens for characterizing continuous multimodal interactions. Empirically, we demonstrate the effectiveness of our approach across diverse multimodal benchmarks, showing robust improvements in predictive performance.

Mon 28 SeptMachine Learning
The gist
When computers look at multiple types of input, like images and text together, they need to figure out which parts of these inputs really help solve a problem and which parts are just noise or unrelated. The authors created a new method that breaks down the information into parts that are shared across all inputs, parts unique to each input type, and irrelevant parts, all in a structured way within the computer’s internal reasoning. This helps the computer understand how each kind of input adds to solving the problem, leading to better predictions. They showed this works well on various tests with different types of data.
Open → 2609.35502v1

Model improves visual answer accuracy by learning from decision boundaries

BIRD: Distilling Decision Boundaries into Rationales for MLLM Adaptation

Abstract: Adapting general-purpose multimodal large language models (MLLMs) to specialized domains requires learning domain-specific decision criteria, which often hinge on subtle visual distinctions between otherwise plausible answers. Rationale augmentation aims to expose such evidence through additional observations or inter-sample comparisons, yet a visually valid cue is not necessarily decision-relevant: it may describe how samples differ without changing the model's relative preference between competing answers. We therefore introduce BIRD, a self-improving Boundary-Informed Rationale Distillation framework that uses model-specific confusions to locate unresolved local decision boundaries and distills the evidence that resolves these confusions into rationales. For each sample, BIRD retrieves candidate neighbors from the target MLLM's own representation space and selects the most confusable one according to its answer preferences. It then generates answer-blind candidate evidence from their visual differences and functionally verifies which evidence most effectively strengthens the model's preference for the correct answer while avoiding inappropriate transfer across the pair. The verified evidence is then distilled into a single-sample rationale for standard supervised fine-tuning. Experiments on medical and chart VQA show that BIRD outperforms competing rationale-augmentation methods across two target MLLMs, while further analyses demonstrate clearer separation of confusable answers and stronger gains from model-matched supervision.

Sun 27 SeptArtificial Intelligence
The gist
It can be hard for AI that looks at pictures and answers questions to tell closely related answers apart, especially in special fields like medicine. The researchers designed a method called BIRD that finds tricky cases where the AI is unsure and learns exactly which visual clues help it choose the right answer. This approach teaches the AI with clearer explanations based on those clues, helping it make better decisions. Testing showed this method works better than similar ways to improve AI understanding in areas like medical images and charts.
Open → 2609.33713v1

Brain signal analysis improves by modeling complex noise patterns

Beyond Gaussian Assumptions: Distribution-Aware Channel Capacity for Effective Connectivity

Abstract: Effective-connectivity estimation from brain signals often relies on Gaussian residual modeling, which enables tractable estimation but can discard informative distributional structure and distort inferred directed interactions when empirical residuals are non-Gaussian. We show across multiple modalities, species, and experimental conditions that both brain signals and fitted channel residuals frequently deviate from Gaussianity. We therefore introduce a distribution-aware, information-theoretic measure of effective connectivity based on channel capacity under general residual distributions. To estimate the resulting capacity from empirical, potentially non-Gaussian residuals, we develop a dual-flow min-max estimator based on normalizing flows, in which a generator searches over admissible input distributions under a power constraint while an observer estimates output entropy. We provide a theoretical characterization of the estimator, showing that the observer objective recovers differential entropy up to a KL approximation term, that the formulation reduces to classical Gaussian capacity as a special case, and that residual entropy can alter achievable information rates beyond variance; game-gap and error analyses further characterize optimization and approximation sources. In brain-like simulations with known directed connectivity, Dual-flow achieves the highest AUROC and AUPRC across ten conditions spanning diverse network topologies, hidden drivers, feedback, and heterogeneous hemodynamics, compared with Gaussian capacity, Granger causality, VAR-LiNGAM, and GIMME. Applied to multimodal brain signals, the method reveals time- and condition-resolved directed interactions consistent with known neurobiological circuitry. Together, these results establish a principled distribution-aware framework for effective-connectivity estimation beyond Gaussian residual modeling.

Sat 26 SeptMachine Learning
The gist
Estimating how different parts of the brain interact often assumes a simple type of noise called Gaussian noise, which can miss important details. The authors found that brain signals and their noise usually don’t follow this simple pattern. They created a new way to measure brain connectivity that considers these more complex noise patterns, resulting in more accurate insight into brain interactions. Their method uses advanced mathematical tools to better capture information flow in brain signals across different species and conditions.
Open → 2609.32774v1

Brain-age prediction improved by modeling stable multifractal patterns

Beyond Feature Reliability: Repeat-Informed Multifractal Curve Regression for Brain-Age Prediction

Abstract: Brain-age prediction from resting-state fMRI provides a quantitative framework for characterizing age-related changes in spontaneous brain dynamics and for identifying functional signatures. Existing studies have linked fractal and multifractal scaling to age and examined the reliability of individual features. However, prediction repeatability depends on how features fluctuate jointly and how a predictor combines them, which feature-wise reliability assessments do not capture. To address this problem, we propose Repeat-informed Multifractal Curve Regression (RMCR), a structured framework for learning stable age-predictive patterns from multifractal curves. By jointly modeling curve structure and repeat-scan variability, RMCR learns predictive combinations of fluctuation orders that target both accuracy and within-subject consistency. Relative to a matched run-level ridge baseline, RMCR reduces single-run MAE by 6.1% on HCP-A and 7.9% on an external Cam-CAN cohort, and within-visit repeat absolute difference by 18.5% on HCP-A, using a single scan at inference.

Thu 24 SeptMachine Learning
The gist
Predicting a person’s brain age using resting brain scans can help understand how brain activity changes with age. The authors introduce a new method called RMCR that looks at complex patterns in brain signals and how these patterns are consistent across repeated scans. This approach not only makes age predictions more accurate but also ensures the predictions are more reliable when scans are repeated. They tested RMCR on two separate brain scan datasets and found it improved prediction accuracy and repeatability compared to existing methods.
Open → 2609.29307v1

Active spot selection does not clearly beat random sampling in spatial transcriptomics

Benchmarking Active Spot Selection for Cost-Efficient Spatial Transcriptomics

Abstract: Spatial transcriptomics (ST) measures gene expression in tissue context, but dense capture grids can be costly and may repeatedly sample morphologically similar regions. Most active learning strategies were developed for categorical labels and independent samples. We conduct a retrospective pool-based benchmark of active learning versus uniform Random sampling for ST, where expression vectors are high-dimensional and continuous and candidates are spatially correlated. Using two fully profiled public ST cohorts, we mask candidate expression vectors and simulate multi-round selection with uncertainty-based Monte Carlo dropout (MC-dropout) and temporal output discrepancy (TOD), and diversity-based CoreSet and TypiClust-inspired selection. We compare 160 completed configurations at 5%, 10%, 30%, and 50% of the fold-wide training spot pool under patient-level cross-validation, with a separate full-label reference. Within each budget, strategies share the selection schedule, morphology-to-expression predictor, and optimization protocol. We assess mean per-gene within-slide Pearson correlation coefficient (PCC), expression-cluster agreement, and Moran's I fidelity. On HER2-positive breast cancer, pooled mean PCC differences from Random across the four active strategies were -0.0176, -0.0117, +0.0056, and +0.0057 at 5%, 10%, 30%, and 50%, respectively. On cutaneous squamous cell carcinoma (cSCC), three strategies were below Random at 5%, and all four were below Random at 10%. On HER2-positive breast cancer, CoreSet and MC-dropout had lower PCC but higher expression-cluster agreement than Random at the two smallest budgets; this pattern did not reproduce on cSCC. Under the reported fixed training horizons, the evaluated active strategies do not consistently improve on Random at small budgets, and rankings depend on the evaluation measure.

Wed 23 SeptComputer Vision and Pattern RecognitionMachine Learning
The gist
Spatial transcriptomics measures gene activity across different regions in tissue, but it can be expensive to collect data everywhere. The authors tested smarter ways to pick spots to measure so they could save costs. Using two public datasets, they found that these smarter methods did not consistently give better results than just picking spots randomly. The effectiveness depended on how many spots were sampled and the measurement used to judge performance.
Open → 2609.27208v1

A new method for detecting anomalies by modeling their causes

A Principled Approach to Unsupervised Anomaly Detection

Abstract: Traditional unsupervised anomaly detection (UAD) methods are designed to flag or localise deviations from a normative distribution, ignoring the underlying generative mechanisms of the anomalies. Yet the nature of an anomaly is often as important as its presence. We reformulate UAD as a Bayesian inverse problem, in which the objective is to infer the most probable corruption responsible for each observation. Our framework yields a probabilistic anomaly score as the energy of the inferred corruption parameters, and serves as a principled recipe for developing new UAD algorithms. We derive several existing methods as instances of the general framework, each corresponding to the same energy score under different modelling choices. Experimentally, we study the framework's components in a controlled setting, and improve object-class AUROC on the MVTec AD dataset by 2.3% by adapting the underlying corruption model. Finally, we validate the framework on a brain MRI benchmark, achieving strong detection performance while producing estimates of pathology intensity, bias, and geometry. Code is available at https://github.com/jgmyles/inverse-uad.

Fri 18 SeptComputer Vision and Pattern Recognition
The gist
Detecting anomalies usually means spotting anything unusual, but often it’s important to understand what caused those unusual things. The authors propose a new way to detect anomalies by guessing the hidden changes that made the data strange, using probabilities. Their method can explain how bad changes happened, not just if something is wrong. This approach improved detection accuracy in industrial images and brain MRIs, while also estimating how the abnormalities appeared.
Open → 2609.21800v1

Scientific image quality assessed by multimodal retrieval system

Scientific Image Quality Assessment via Multi-modal Retrieval-Augmented Generation

Abstract: This paper proposes a Retrieval-Augmented Generation (RAG) framework for scientific image quality assessment, designed to simultaneously address both the understanding track (SIQA-U) and the scoring track (SIQA-S) of the SIQA challenge. We construct a multimodal index that integrates textual semantics with fine-grained visual features, and develop a multi-route retrieval and fusion mechanism to provide large language models with highly relevant reference cases, thereby enhancing their capability to evaluate complex scientific images. Experimental results demonstrate that the proposed framework effectively aligns with the judgment criteria of human experts. Ultimately, our method achieves 1st place in the SIQA-U track of the SIQA challenge at the ICME 2026 Grand Challenges.

Thu 17 SeptComputer Vision and Pattern RecognitionComputation and Language
The gist
Judging the quality of scientific images can be tricky because it requires understanding both the pictures and their scientific context. The authors created a system that helps a language model by giving it relevant visual and text examples to compare with the images being evaluated. This approach helps the system better think like human experts when assessing image quality. Their method won first place in a scientific image quality assessment challenge.
Open → 2609.19634v1

Time-warping estimation improves signal analysis with faster computing

Time-warping estimation via stationarity-based learning of the de-warped signal

Abstract: Time-warping estimation is a fundamental problem in signal processing with applications in bioacoustics, radar, and biomedical analysis. This paper introduces a Time-Warping Estimation Trainable (TWET) model for estimating timewarping functions from a single observation. The proposed approach formulates time-warping estimation as a stationarization problem in the wavelet domain and leverages a hierarchical dilated convolutional architecture to estimate the time-warping functions. A differentiable stationarity criterion is introduced for end-to-end optimization. TWET is compared with existing approaches. Experimental results show improved deformation reconstruction accuracy together with significantly reduced computation time, making the framework compatible with low-latency applications.

Tue 15 SeptMachine Learning
The gist
Time-warping is like adjusting the speed of a recording to line up important events. The authors created a new method called TWET that uses a smart wavelet approach to guess how a signal has been stretched or squished in time from just one example. This method teaches a computer model to find these changes quickly and accurately by looking for stability patterns in the signal. Compared to older methods, TWET is faster and better at fixing these time changes, making it useful for things like medical or animal sound analysis.
Open → 2609.16796v1

PSMP-CLIP improves zero-shot anomaly detection with better image masks

PSMP-CLIP: Patch-Prompt SAM and Multi-Semantic Prompting for CLIP-Based Zero-Shot Anomaly Detection

Abstract: Zero-shot anomaly detection aims to localize anomalies without target-domain samples. Existing CLIP-based methods suffer from coarse anomaly maps and limited semantic prompts. We propose PSMP-CLIP, integrating patch-prompt SAM2 segmentation (PPSS) and multi-semantic guided prompt regularization (MSGPR). PPSS samples prompts directly from intermediate patch features, avoiding threshold drift and guiding SAM2 to produce precise masks. MSGPR uses multiple learnable prompts constrained by semantic anchors to preserve generalization. Experiments on 14 datasets show highly competitive performance, achieving the best pixel-level AUROC on MVTec AD, BTAD, DTD-Synthetic, CVC-ClinicDB, TN3K, Endo, and Kvasir.

Tue 15 SeptComputer Vision and Pattern Recognition
The gist
Detecting unusual parts in images without using examples from that image type is hard. Existing methods using a model called CLIP struggle to pinpoint odd spots precisely. The authors present PSMP-CLIP, which helps by creating better image masks from smaller image patches and by using multiple guided text prompts for improved detection. They tested this approach on many datasets and found it works better than similar methods in identifying anomalies at the pixel level.
Open → 2609.16785v1

AstroSpecLM links astronomical spectra with language model answers

AstroSpecLM: A Spectrum-Language Model for Evidence-Grounded Astronomical Spectral Analysis

Abstract: Astronomical spectra encode rich physical information, but drawing scientific conclusions from spectral features typically requires expert interpretation. This paper presents AstroSpecLM, a spectrum-language model that connects one-dimensional DESI spectra with Qwen3-4B to answer questions and provide explanations grounded in spectral evidence. Instead of generating question-answer pairs directly from templates or raw catalog fields, we first distill each spectrum into a compact set of catalog- and spectrum-derived facts, then use these facts as references to generate instruction-following conversations. The resulting model is competitive with specialist supervised baselines on classification and redshift estimation, while additionally producing natural-language explanations that reference specific spectral features. Our results indicate that grounding a language model in one-dimensional scientific spectra is feasible, and that fact-mediated instruction data yields a model capable of both prediction and explanation.

Mon 7 SeptArtificial Intelligence
The gist
Astronomical spectra contain detailed clues about celestial objects, but understanding them usually needs expert knowledge. The authors created AstroSpecLM, a model that reads these spectra and answers questions in natural language, explaining its reasoning based on specific features it sees. They trained it to first extract important facts from the spectral data before generating answers and explanations. This approach performs well compared to specialized methods and makes the analysis more accessible.
Open → 2609.07102v1