AI in cancer detection

Buyer Beware: confounding factors and biases abound when predicting omics-based biomarkers from histological images

This paper shows that deep learning models for predicting biomarkers from histology images are confounded by interdependencies among biomarkers, leading to less accurate and less reliable predictions.

Detecting Extrachromosomal DNA from Routine Histopathology

This paper develops a deep learning framework to detect extrachromosomal DNA from histopathology images, identifying ecDNA-amplified tumours across twelve cancer types.

Comparative Analysis of Pathology Foundation Models for Automated Detection of Tertiary Lymphoid Structures in H&E-Stained Digital Pathology Images

This study evaluates pathology foundation models for detecting tertiary lymphoid structures in cancer tissues, finding strong performance in PDAC but challenges in HNSCC due to tumor heterogeneity.

Transforming Histology into Virtual Multiplex Immunofluorescence to Decode Prognostic Spatial Immunity in Hepatocellular Carcinoma

HCCExplorer translates H&E slides into virtual multiplex immunofluorescence to identify prognostic spatial immune features in hepatocellular carcinoma.

Impact of tissue staining and scanner variation on the performance of pathology foundation models: a study of sarcomas and their mimics

Across 14 sarcoma and mimic subtypes, six pathology foundation models beat ResNet50 but lost accuracy under multi-institutional staining; adding 121 externally stained slides narrowed the gap.

Detection of prostate cancer in 3D pathology datasets via generative immunolabeling

This paper develops a 3D computational pipeline called SIGHT to detect prostate cancer in pathology datasets using generative immunolabeling, achieving an F1 score comparable to pathologist agreement.

X-SPATIO: An Explanatory Deep Learning Pipeline for the Prediction and Visualization of Spatially Resolved Biomarker Expression in Triple-Negative Breast Cancer

X-SPATIO predicts and visualizes spatially resolved biomarker expression in TNBC by linking H&E morphology with spatial transcriptomic and proteomic data.

Normal Breast Tissue (NBT)-Classifiers: Advancing Compartment Classification in Normal Breast Histology

CNN patch classifiers trained only on normal breast H&E slides label epithelium, stroma and adipocytes with AUCs of 0.98–1.00 on three external cohorts and feed a lobule-detection pipeline.

StainStyleSampler: Clustering-based sampling of whole slide image appearances

StainStyleSampler clusters and samples stain styles from whole slide images to improve machine learning robustness across different staining appearances.

Neural Networks as Entropic Systems: Applications in Digital Pathology

This paper characterizes neural network learning dynamics in digital pathology by measuring activation structure, weight evolution, and entropy changes during training.

Performance of Naiive Spectral Geometric Models in Histopathology AI

This paper evaluates spectral models in histopathology tasks and finds they underperform CNN baselines but offer interpretability benefits.

Genomic Characterization of Lung Cancer in Never-Smokers Using Deep Learning

This paper develops a deep learning model to predict molecular alterations in never-smoker lung adenocarcinomas from histological images, achieving high accuracy for most features.

HistoSweep enables cellular-resolution tissue quality control for gigapixel images in digital pathology and spatial omics

HistoSweep improves tissue quality control in gigapixel images by removing artifacts and preserving biologically relevant structures, enhancing analysis accuracy in digital pathology and spatial omics.

ShapeUpLMD: An Automated Pipeline for Spatially Optimized Laser Microdissection in Multi-omic Tissue Profiling

ShapeUpLMD automates ROI optimization for laser microdissection, improving tumor purity and spatial accuracy in multi-omic tissue profiling.

LIMPACAT : Multi-Omics Attention Transformer for Immune Prediction in Liver Cancer Using Whole-Slide Imaging

LIMPACAT uses whole-slide images and multi-omics data to predict immune cell levels in liver cancer, achieving 80% accuracy in prognosis classification.

HistoSB-Net: Semantic Bridging for Data-Limited Cross-Modal Histopathological Diagnosis

HistoSB-Net adapts pre-trained vision-language models for histopathological diagnosis with limited data, improving performance through semantic bridging.

Mitosis detection in the wild using detection transformers

This paper develops a deep learning model for mitotic figure detection that generalizes across diverse histological regions and tumor types, achieving an F1-score of 0.7769.

Generative AI Enables Breast Cancer Genomic Subtype Prediction from Histology Images

PATH-IC uses generative AI to predict breast cancer genomic subtypes from histology images, achieving high accuracy and correlating with clinical outcomes.

FourierMIL: Fourier filtering-based multiple instance learning for whole slide image analysis

FourierMIL uses Fourier filtering and multiple instance learning to analyze whole-slide images, outperforming existing methods in tumor classification.

Label Noise in Pathological Segmentation is Overlooked, Leading to Potential Overestimation of Artificial Intelligence Models

This study investigates how label noise in pathology segmentation affects AI model performance, finding that boundary-altering noise leads to overestimation of model accuracy.

Multiple instance learning on tile level-pathologyimages provides accurate and interpretableclassification for breast cancer molecular subtypes

This paper presents a weakly supervised deep learning approach for breast cancer molecular subtyping using tile-level pathology images without tumor annotations, achieving high accuracy and interpretability.

Diffusion Models vs. DCGANs for Class-Imbalanced Lung Cancer CT Classification: A Comparative Study

This study compares diffusion models and DCGANs for class-imbalanced lung cancer CT classification, finding diffusion models achieve higher accuracy and better recall for benign cases.

Ensemble-Based Deep Learning for Breast Cancer Detection and Classification in Histopathological Images

This paper develops a deep learning framework for breast cancer detection and classification in histopathological images, achieving high accuracy using ensemble methods and comprehensive preprocessing.

Digital Spatial Pathway Mapping Reveals Prognostic Tumor States in Head and Neck Cancer

This paper introduces a method to infer signaling pathway activities from H&E slides in HNSCC, revealing two prognostic tumor states.

Predicting microsatellite instability from whole slide images using texture features

This study predicts microsatellite instability from whole slide images using texture features, achieving high AUC scores on both tile and sample levels.

Self-supervised AI reveals a lethal discohesive phenotype in lung adenocarcinoma

This paper uses self-supervised AI to identify under-appreciated stromal morphologies in lung adenocarcinoma linked to poor prognosis, independent of current clinical schemas.

Machine Learning Models Based on Histological Images from Healthy Donors Identify ImageQTLs and Predict Chronological Age

This paper develops models linking histological images to genotype, transcriptome, and age, identifying 906 imageQTLs and predicting gene expression and chronological age from images.

Image Analysis for Non-Neoplastic Kidney Disease: Utilizing Morphological Segmentation to Improve Quantification of Interstitial Fibrosis

This paper improves quantification of interstitial fibrosis in kidney biopsies by using morphological segmentation, showing better correlation with expert grading and reduced bias.

Annotation-Free Prediction of Cancer Cells and Glands and Spatial Analysis of Immune Cells.

The paper develops TOPAZ, a tool that predicts cancerous cells/glands and analyzes immune cell spatial relationships in prostate cancer, finding that immune cell distance from cancer glands correlates with recurrence risk.

Star-Motifs: Revealing Single-Cell Spatiotypes from Routine Histology Using Star-Convex Neighborhoods

Star-Motifs identifies spatial patterns in histology slides to reveal clinically relevant tumor microenvironment features and survival correlations.