Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Huachansu triggers mitochondrial apoptosis and ER stress to inhibit hepatocellular carcinoma progression.

Saudi pharmaceutical journal : SPJ : the official publication of the Saudi Pharmaceutical Society·2026
Same author

Topology-Driven Activation of the GLP-1 Receptor Promotes Adipose Tissue Browning.

Protein & cell·2026
Same author

RND3 Enhances Cardiac Glucose Metabolism Through Inhibiting ACAT1-Dependent PDHA1 Acetylation and Protects Against Ischemia-Reperfusion Injury.

Circulation·2026
Same author

Dual-Stimulus Chiroptical Switch of Tetrastable [3]Rotaxanes.

Organic letters·2026
Same author

Unveiling medication patterns in traditional Chinese medicine for the prevention of colorectal cancer recurrence: from potential combinations to validation of components and targets.

Chinese medicine·2026
Same author

Drug-Coated Balloons Versus Drug-Eluting Stents for Patients With Long De Novo Coronary Artery Lesions: Insights From the REC-CAGEFREE I Trial.

Catheterization and cardiovascular interventions : official journal of the Society for Cardiac Angiography & Interventions·2026

Related Experiment Video

Updated: Jun 7, 2025

Swin-PSAxialNet: An Efficient Multi-Organ Segmentation Technique
04:48

Swin-PSAxialNet: An Efficient Multi-Organ Segmentation Technique

Published on: July 5, 2024

369

Positional encoding-guided transformer-based multiple instance learning for histopathology whole slide images

Jun Shi1, Dongdong Sun2, Kun Wu3

  • 1School of Software, Hefei University of Technology, Hefei, 230601, Anhui Province, China.

Computer Methods and Programs in Biomedicine
|November 16, 2024
PubMed
Summary

This study introduces a novel positional encoding-guided transformer-based multiple instance learning (PEGTB-MIL) method for whole slide image classification. PEGTB-MIL enhances pathological diagnosis by incorporating spatial information into patch features, improving accuracy in cancer subtyping and gene mutation prediction.

Keywords:
Cancer subtypingDigital pathologyGene mutation predictionMultiple instance learningPosition encodingWhole slide image

More Related Videos

A Swin Transformer-Based Model for Thyroid Nodule Detection in Ultrasound Images
04:23

A Swin Transformer-Based Model for Thyroid Nodule Detection in Ultrasound Images

Published on: April 21, 2023

1.7K
Automated Slide Scanning and Segmentation in Fluorescently-labeled Tissues Using a Widefield High-content Analysis System
09:33

Automated Slide Scanning and Segmentation in Fluorescently-labeled Tissues Using a Widefield High-content Analysis System

Published on: May 3, 2018

8.1K

Related Experiment Videos

Last Updated: Jun 7, 2025

Swin-PSAxialNet: An Efficient Multi-Organ Segmentation Technique
04:48

Swin-PSAxialNet: An Efficient Multi-Organ Segmentation Technique

Published on: July 5, 2024

369
A Swin Transformer-Based Model for Thyroid Nodule Detection in Ultrasound Images
04:23

A Swin Transformer-Based Model for Thyroid Nodule Detection in Ultrasound Images

Published on: April 21, 2023

1.7K
Automated Slide Scanning and Segmentation in Fluorescently-labeled Tissues Using a Widefield High-content Analysis System
09:33

Automated Slide Scanning and Segmentation in Fluorescently-labeled Tissues Using a Widefield High-content Analysis System

Published on: May 3, 2018

8.1K

Area of Science:

  • Computational pathology
  • Digital pathology
  • Machine learning in healthcare

Background:

  • Whole slide image (WSI) classification is crucial for computer-aided pathological diagnosis.
  • Weakly supervised methods, particularly multiple instance learning (MIL), are gaining attention due to high annotation costs.
  • Existing MIL methods often overlook spatial relationships between image patches, potentially limiting feature discriminative ability.

Purpose of the Study:

  • To propose a novel positional encoding-guided transformer-based multiple instance learning (PEGTB-MIL) method for histopathology WSI classification.
  • To encode spatial positional properties of patches into semantic features for improved WSI classification performance.
  • To explore correlations among patches by integrating spatial and semantic information.

Main Methods:

  • Extract deep features from WSI patches and encode their 2D spatial information using a position encoder.
  • Incorporate semantic features with spatial embeddings.
  • Apply multi-head self-attention (MHSA) to analyze contextual and spatial dependencies, enhanced by an auxiliary reconstruction task for spatial-semantic consistency.

Main Results:

  • The PEGTB-MIL method was validated on public TCGA datasets (TCGA-LUNG, TCGA-BRCA) and in-house clinical datasets (USTC-EGFR, USTC-GIST).
  • Demonstrated effectiveness in cancer subtyping and gene mutation status prediction.
  • Achieved superior performance over state-of-the-art methods, with AUCs of 97.13±0.34%, 86.74±2.64%, 83.25±1.65%, and 72.52±1.63% on the respective datasets.

Conclusions:

  • PEGTB-MIL effectively utilizes positional encoding to guide and enhance MIL for improved WSI classification.
  • The auxiliary reconstruction module successfully preserves spatial-semantic consistency of patch features.
  • The study highlights the significance of positional information in disease diagnosis, offering a promising research direction.