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

Epicardial adipose tissue signatures in Asian coronary artery disease: Insights from cardiac CT.

American journal of preventive cardiology·2026
Same author

DiffBulk: Enhancing Spatial Transcriptomic Prediction With Diffusion-Based Training.

IEEE transactions on medical imaging·2026
Same author

Can 3D T1 Post-Contrast MRI in A Radiomics-Machine Learning Model Distinguish Infective from Neoplastic Ring-Enhancing Brain Lesions? An Exploratory Study.

Diagnostics (Basel, Switzerland)·2026
Same author

Deep learning-based quantification of epicardial adipose tissue volume from non-contrast computed tomography images: a multi-centre study.

European heart journal. Digital health·2025
Same author

Systematic benchmarking of 13 AI methods for predicting cyclic peptide membrane permeability.

Journal of cheminformatics·2025
Same author

The haplotype-resolved T2T genome for Bauhinia × blakeana sheds light on the genetic basis of flower heterosis.

GigaScience·2025

Related Experiment Video

Updated: Aug 1, 2025

Machine Learning Algorithms for Early Detection of Bone Metastases in an Experimental Rat Model
07:15

Machine Learning Algorithms for Early Detection of Bone Metastases in an Experimental Rat Model

Published on: August 16, 2020

6.9K

Distribution based MIL pooling filters: Experiments on a lymph node metastases dataset.

Mustafa Umit Oner1, Jared Marc Song Kye-Jet2, Hwee Kuan Lee3

  • 1A*STAR Bioinformatics Institute, 30 Biopolis Street, Singapore 138671, Singapore; School of Computing, National University of Singapore, 13 Computing Drive, Singapore 117417, Singapore; Artificial Intelligence Engineering, Bahcesehir University, Besiktas, Istanbul 34349, Turkey.

Medical Image Analysis
|April 30, 2023
PubMed
Summary

This study introduces novel distribution-based pooling filters for multiple instance learning (MIL) in digital histopathology. These filters enhance cancer diagnosis by improving the analysis of gigapixel slides with weak labels.

Keywords:
Distribution poolingMIL pooling filtersMultiple instance learning (MIL)Point estimate based pooling

More Related Videos

Pathological Analysis of Lung Metastasis Following Lateral Tail-Vein Injection of Tumor Cells
08:54

Pathological Analysis of Lung Metastasis Following Lateral Tail-Vein Injection of Tumor Cells

Published on: May 20, 2020

9.1K
Comparison of Predictive Performance of Three Lymph Node Staging Systems in Colorectal Signet Ring Cell Carcinoma Based on Machine Learning Model
07:13

Comparison of Predictive Performance of Three Lymph Node Staging Systems in Colorectal Signet Ring Cell Carcinoma Based on Machine Learning Model

Published on: April 18, 2025

153

Related Experiment Videos

Last Updated: Aug 1, 2025

Machine Learning Algorithms for Early Detection of Bone Metastases in an Experimental Rat Model
07:15

Machine Learning Algorithms for Early Detection of Bone Metastases in an Experimental Rat Model

Published on: August 16, 2020

6.9K
Pathological Analysis of Lung Metastasis Following Lateral Tail-Vein Injection of Tumor Cells
08:54

Pathological Analysis of Lung Metastasis Following Lateral Tail-Vein Injection of Tumor Cells

Published on: May 20, 2020

9.1K
Comparison of Predictive Performance of Three Lymph Node Staging Systems in Colorectal Signet Ring Cell Carcinoma Based on Machine Learning Model
07:13

Comparison of Predictive Performance of Three Lymph Node Staging Systems in Colorectal Signet Ring Cell Carcinoma Based on Machine Learning Model

Published on: April 18, 2025

153

Area of Science:

  • Computational pathology
  • Machine learning
  • Digital histopathology

Background:

  • Histopathology is vital for cancer diagnosis, analyzing gigapixel slides.
  • Multiple Instance Learning (MIL) is suitable for digital histopathology due to its handling of large slides and weak labels.
  • MIL treats slides as bags of patches with slide-level labels.

Purpose of the Study:

  • To introduce and evaluate distribution-based pooling filters for MIL in digital histopathology.
  • To demonstrate the superior expressiveness of distribution-based pooling compared to point estimate-based methods.
  • To improve the accuracy of cancer diagnosis using computational pathology.

Main Methods:

  • Developed distribution-based pooling filters to estimate marginal distributions of instance features for bag-level representation.
  • Formally proved the enhanced expressiveness of distribution-based pooling filters over 'max' and 'mean' pooling.
  • Empirically validated models with distribution-based pooling on the CAMELYON16 dataset.

Main Results:

  • Distribution-based pooling filters capture more information than point estimate-based filters.
  • Models using distribution-based pooling performed equally well or better than traditional methods on real-world MIL tasks.
  • Achieved an Area Under the ROC Curve of 0.9325 for tumor vs. normal slide classification on the CAMELYON16 dataset.

Conclusions:

  • Distribution-based pooling filters offer a more expressive and effective approach for MIL in digital histopathology.
  • This method shows significant promise for improving automated cancer diagnosis from histopathology slides.
  • The proposed filters advance the field of computational pathology and machine learning applications in cancer research.