AI in cancer detection

Homologous recombination deficiency prediction from whole slide images using label refinement and foundation-model benchmarking in ovarian cancer

This paper evaluates if H&E WSIs from TCGA-OV can predict HRD status using label refinement and foundation models, finding modest predictive performance but limited clinical interpretability.

Automatic Bevacizumab Response Prediction in Ovarian Cancer from Digital Pathology Images via Novel AI-based Computational Pipeline

The paper presents an AI-based pipeline for predicting bevacizumab response in ovarian cancer using digital pathology images, achieving improved AUC performance compared to existing models.

Vessel Spatial Analysis (VeSpA): a tool for whole slide image segmentation, morphometry, and QuPath extension.

VeSpA is an open-source tool for automated vessel segmentation and morphometric analysis in histological images, integrating with QuPath for visualization and annotation.

Dissecting and directing pathology foundation models

This paper introduces PICASSO, a framework that makes pathology foundation models interpretable and controllable by decomposing embeddings into human-interpretable visual concepts, enabling new biological insights and downstream applications.

A multimodal foundation model linking histopathology and DNA methylation

HistoMethyl links histopathology with DNA methylation to predict molecular features from H&E slides alone, improving mutation prediction and survival outcomes across multiple cancer cohorts.

TopoMIL: Topology Improves Multiple Instance Learning in Diagnostic Microscopic Images

TopoMIL improves multiple instance learning for diagnostic microscopic images by integrating topological structures, achieving classification gains across various pooling methods.

Intra-slide calibration technology improves immunohistochemical harmonization within and between anatomic pathology laboratories

This study shows intra-slide calibration technology improves inter-laboratory consistency in immunohistochemical analysis of glioblastoma samples through computational methods.

HOPE: Interpretable Histology Analysis with Spatial Omics-Derived Signatures for Precision Oncology

HOPE learns tumor microenvironment signatures from paired H&E and spatial omics data to improve prognostic stratification using H&E images alone.

Compositional and interpretable representation of histology using AI foundation models and sparse autoencoders

This paper introduces a human-in-the-loop framework using AI foundation models and sparse autoencoders to interpret H&E histology images and integrate them with molecular spatial profiling.

SortIT - A Tool For Assessing Observer Variability And Creating Ground Truth Image Classification Datasets

DigitAb: Domain-Adaptive Cell Type Prediction Method from Light Microscopy Images

DigitAb predicts cell types from H&E-stained slides using deep learning, achieving high accuracy without specialized assays or immunostains.

Multi-Scale Tri-Modal Histology Dataset Integrating Tumor Morphology, Immune Patterns, and Clinical Outcomes

Prostate-TriMod integrates visual morphology, spatial tissue maps, and clinical outcomes for prostate cancer to enable multimodal AI research.

WSInsight: a cloud-native, agent-callable platform for single-cell whole-slide pathology

WSInsight analyzes H&E whole-slide images at scale, integrating single-cell segmentation and classification with retrainable cell-type models, and applies it to TCGA cohorts to recover immune and molecular associations.

CLEAR-HPV: Interpretable Concept Discovery for HPV-Associated Morphology in Whole-Slide Histology

CLEAR-HPV discovers interpretable morphologic concepts in HPV-associated histology slides using attention-based methods, reducing high-dimensional features to 10 concepts while preserving predictive information.

Molecular Translators as a Computational Primitive for Biomarker Discovery: Learnability Gains Under Conserved Information Ceilings

The paper develops a framework to analyze performance plateaus in H&E-to-molecular biomarker translation, distinguishing methodological limitations from intrinsic information ceilings.

Explainable Prototype Booster: Enhancing Latent Representations of Foundation Models for Gene Expression Prediction

The paper introduces EP-Booster, a method that improves gene expression prediction from H&E images by refining foundation model embeddings with biological knowledge, achieving better performance and interpretability.

Interpreting and Validating a Deep Learning Model Predictive of Spatial Morphologic-Molecular Patterns in Lung Adenocarcinoma, Using Ground Truth Immunohistochemistry

XpressO-Lung predicts spatial gene expression patterns in lung adenocarcinoma using histopathology images, validated against immunohistochemistry, showing strong concordance with known morphologic features.

Features fusion or not: harnessing multiple pathological foundation models using Meta-Encoder for downstream tasks fine-tuning

The paper introduces Meta-Encoder, a framework that combines multiple pathological foundation models to improve downstream task performance in computational pathology.

ADAPTIVE MULTI-SCALE GRAPH TRANSFORMER FRAMEWORK FORHISTOPATHOLOGICAL IMAGES

The paper introduces an Adaptive Multi-Scale Graph Transformer (AMGT) for histopathological image classification, achieving better performance and efficiency than existing models.

Exploring Multi-Scale Local and Global Features in Whole Slide Images Using State Space Models

This paper introduces MambaWSI, a state space model for whole slide image classification that extracts multi-scale local and global features, achieving better performance than existing methods.

Adaptive Integration of Heterogeneous Foundation Models to Find Histologically Predictable Genes in Breast Cancer

This paper integrates multiple foundation models to improve gene prediction in histopathology images, achieving better accuracy for breast cancer subtypes and drug-target genes.

Artificial Intelligence-enabled Histological Analysis in Preclinical Respiratory Disease Models: A Scoping Review

This paper reviews AI applications in histological analysis of preclinical respiratory disease models, finding high performance but gaps in validation, transparency, and standardization.

Translating Histopathology Foundation Model Embeddings into Cellular and Molecular Features for Clinical Studies

STpath translates histopathology image embeddings into cell-type composition and gene expression estimates for clinical applications.

ViFIT-assisted Histopathology: From H&E Style Standardization to Virtual Fiber Image Transformation

ViFIT-assisted histopathology converts diverse H&E-stained images into standardized H&E images and generates virtual fiber images, improving diagnostic workflows across cancer types.

LazySlide: accessible and interoperable whole slide image analysis

LazySlide provides an open-source Python package for whole-slide image analysis and multimodal integration, enabling efficient histopathology workflows with minimal setup.

GenBio-PathFM: A State-of-the-Art Foundation Model for Histopathology

GenBio-PathFM is a foundation model for histopathology that achieves state-of-the-art performance with less training data by using a diverse data curation pipeline and a dual-stage learning strategy.

Cross-Modal Training Using Xenium Spatial Transcriptomics Enables DINO-DETR Based Detection of Vascular Niches in H&E Whole-Slide Images

This paper develops a DINO-DETR model trained on spatial transcriptomics data to detect vascular niches in H&E slides, showing improved vascular quantification and prognostic value in astrocytoma.

Subphase-Labeled Mitotic Dataset for AI-powered Cell Division Analysis

The paper introduces a subphase-labeled mitotic dataset to improve mitosis detection and subtyping, enhancing domain generalization and enabling biological insights into mitotic subphases.

Direct pathway enrichment prediction from histopathological whole slide images and comparison with gene expression mediated models

This study compares direct pathway enrichment prediction from histopathological WSIs with gene expression-mediated models in TCGA-BRCA, finding direct prediction performs better.

A comprehensive benchmark of publicly available image foundation models for their usability to predict gene expression from whole slide images

The paper benchmarks five image foundation models for predicting gene expression from whole-slide images using the TCGA-BRCA cohort, finding domain-aligned models outperform general-purpose ones.