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 Experiment Video

Updated: Jun 13, 2026

A Live-cell Image-Based Machine Learning Strategy to Monitor Pluripotent Stem Cell Differentiation
11:38

A Live-cell Image-Based Machine Learning Strategy to Monitor Pluripotent Stem Cell Differentiation

Published on: October 4, 2024

Continual Learning for Histopathology Image Classification in Class-Incremental Learning.

Yuanyuan Wu1, Yu Zhao1, Anca Ralescu1

  • 1Department of Computer Science, University of Cincinnati, 2901 Woodside Drive, Cincinnati, OH 45221, USA.

Diagnostics (Basel, Switzerland)
|June 12, 2026
PubMed
Summary
This summary is machine-generated.

Related Concept Videos

Classification of Leukocytes01:30

Classification of Leukocytes

Leukocytes are classified into two groups based on the presence or absence of cytoplasmic granules. Granular leukocytes, which contain granules, belong to the myeloid lineage and are divided into three subtypes: neutrophils, eosinophils, and basophils. These cells are roughly spherical and characterized by the granules in their cytoplasm.
Neutrophils are the most abundant type of granular leukocytes, comprising 50-70% of all leukocytes. They feature small, evenly distributed granules and a...

You might also read

Related Articles

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

Sort by
Same author

Automated detection of spreading depolarizations in electrocorticography.

Scientific reports·2025
Same author

Bio-informatic analysis of CRISPR protospacer adjacent motifs (PAMs) in T4 genome.

BMC genomic data·2022
Same author

Ketogenesis Attenuates KLF5-Dependent Production of CXCL12 to Overcome the Immunosuppressive Tumor Microenvironment in Colorectal Cancer.

Cancer research·2022
Same author

A fast algorithm to compute precise type-2 centroids for real-time control applications.

IEEE transactions on cybernetics·2014
Same author

Piglets cloned from induced pluripotent stem cells.

Cell research·2012
Same author

Telmisartan improves insulin resistance of skeletal muscle through peroxisome proliferator-activated receptor-δ activation.

Diabetes·2012
Same journal

Correction: Luca et al. Global and Regional Diagnostic Results of Progress Toward Cervical Cancer Elimination, According to the WHO Strategy: A Systematic Literature Review with Narrative Synthesis. <i>Diagnostics</i> 2026, <i>16</i>, 1224.

Diagnostics (Basel, Switzerland)·2026
Same journal

Association Between Systemic Inflammatory Response Biomarkers and Disease Activity in Systemic Lupus Erythematosus: A Multi-Center Retrospective Study.

Diagnostics (Basel, Switzerland)·2026
Same journal

Vertebrogenic Low Back Pain and Basivertebral Nerve Ablation: A Review of Mechanisms, Imaging-Driven Selection, and Clinical Outcomes.

Diagnostics (Basel, Switzerland)·2026
Same journal

Multivalvular Carcinoid Heart Disease: The Role of Echocardiography in Diagnosis and Selection for Heterotopic Bicaval Valve Implantation.

Diagnostics (Basel, Switzerland)·2026
Same journal

Data-Efficient and Explainable Multimodal Survival Prediction in NSCLC Using Deep Image Embeddings, Clinical Variables, and Gradient-Boosted Trees.

Diagnostics (Basel, Switzerland)·2026
Same journal

Anomalous Left Coronary Artery from the Pulmonary Artery: Cinematic Volume Rendering Technique for Enhanced Anatomic Visualization.

Diagnostics (Basel, Switzerland)·2026
See all related articles

Continual learning (CL) in histopathology AI is challenging but feasible. Replay methods offer high accuracy with ample resources, while prompt-based methods provide efficient, privacy-preserving alternatives for class-incremental learning (CIL).

Area of Science:

  • Artificial Intelligence
  • Computational Pathology
  • Medical Imaging

Background:

  • Continual learning (CL) is crucial for adaptive clinical AI but faces challenges in histopathology due to privacy, evolving categories, and staining variability.
  • Class-incremental learning (CIL) scenarios, where new diagnostic categories are introduced sequentially, are particularly relevant for histopathology AI development.
  • Investigating CL methods is essential to overcome these hurdles and improve AI model adaptability in digital pathology.

Purpose of the Study:

  • To benchmark various Continual Learning (CL) methods for histopathology image classification within a class-incremental learning (CIL) framework.
  • To evaluate the impact of normalization strategies, replay buffer size, and training epochs on CL performance.
  • To analyze the clinical relevance, error patterns, and computational efficiency of different CL approaches.
Keywords:
catastrophic forgettingclass-incremental learningcontinual learningdigital pathologyhistopathologyprompt-based learning

Related Experiment Videos

Last Updated: Jun 13, 2026

A Live-cell Image-Based Machine Learning Strategy to Monitor Pluripotent Stem Cell Differentiation
11:38

A Live-cell Image-Based Machine Learning Strategy to Monitor Pluripotent Stem Cell Differentiation

Published on: October 4, 2024

Main Methods:

  • Benchmarking of regularization-, replay-, architecture-, and prompt-based CL methods on the NCT-CRC-HE-100K and CRC-HE-7K datasets.
  • Comparison of four normalization strategies, including dataset-level normalization.
  • Analysis of performance metrics such as average accuracy and forgetting, alongside clinical relevance and error analysis using confusion matrices and ROC curves.

Main Results:

  • Dataset-level normalization consistently yielded the best performance across evaluated strategies.
  • DER++ (replay-based) achieved high accuracy (94.77%) and low forgetting (3.66%) but required significant memory and training time.
  • DualPrompt (prompt-based) offered competitive performance (88.97% accuracy, 7.70% forgetting) with fewer epochs, lower computational cost, and smoother training.

Conclusions:

  • Replay-based CL methods excel in accuracy and reduced forgetting when computational resources and data storage are available, albeit with higher costs.
  • Prompt-based CL methods present a viable, exemplar-free alternative for privacy- and resource-constrained histopathology CIL applications.
  • Dataset-level normalization is a key factor for achieving stable and reliable CL performance in histopathology CIL settings.