Search research articles

ABOUT JoVE

Overview Leadership Blog JoVE Help Center

AUTHORS

Publishing Process Editorial Board Scope & Policies Peer Review FAQ Submit

LIBRARIANS

Testimonials Subscriptions Access Resources Library Advisory Board FAQ

RESEARCH

JoVE Journal Methods Collections JoVE Encyclopedia of Experiments Archive

EDUCATION

JoVE Core JoVE Business JoVE Science Education JoVE Lab Manual Faculty Resource Center Faculty Site

Terms & Conditions of Use

Search research articles

Related Experiment Videos

From unlabeled to labeled: self-supervised deep learning in computational pathology.

Taymaz Akan¹, Richa Aishwarya², Md Shenuarin Bhuiyan²

¹Department of Medicine, LSU Health Shreveport, Shreveport, LA, USA.

|July 16, 2026

Summary

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

Association of clinical characteristics to mortality in acute respiratory failure patients treated with ECCO<sub>2</sub>R.

The International journal of artificial organs·2026

Same author

Endothelial mitochondrial homeostasis in the development and progression of atherosclerosis.

Frontiers in cardiovascular medicine·2026

Same author

Expanding the OMOP common data model to support extracorporeal life support research.

JAMIA open·2026

Same author

Chemical exposure and hepatic steatosis: How environmental factors impact liver health.

Food and chemical toxicology : an international journal published for the British Industrial Biological Research Association·2026

Same author

Combined effects of heavy metals and volatile organic compounds on oxidative stress and cardiovascular risk.

American heart journal plus : cardiology research and practice·2026

Same author

Computational pathology with dynamic convolutional and adaptive kernels.

Journal of pathology informatics·2026

Self-supervised learning (SSL) models like SSL-HistoNet can identify disease-specific tissue structures from unannotated histopathological images, reducing manual labeling needs. This approach shows high performance, comparable to supervised methods, for analyzing muscle pathologies.

Area of Science:

Computational pathology
Machine learning in histopathology
Digital pathology advancements

Background:

Pathology decision support systems require large, manually annotated datasets, hindering clinical implementation.
Self-supervised learning (SSL) offers a solution by extracting features from unannotated images.

Purpose of the Study:

Introduce SSL-HistoNet, a novel model for learning disease-relevant morphological representations from histopathological images using SSL.
Demonstrate the model's capability in an annotation-free setting.

Main Methods:

Applied SSL-HistoNet to WGA-stained skeletal muscle tissues from mouse models of amyotrophic lateral sclerosis (ALS) and Type I diabetes.
Pretrained the SSL encoder on unlabeled data, then integrated it with an attention-guided classifier to detect pathological alterations.

Keywords:

Computational pathology Contrastive learning Self-supervised learning Skeletal muscle WGA staining

Related Experiment Videos

Main Results:

SSL-HistoNet achieved high performance with 0.98 precision, 0.98 recall, and 0.98 AUC, outperforming state-of-the-art supervised models.
Feature analyses revealed consistent, class-level changes in morphology-related patterns.

Conclusions:

SSL-HistoNet provides an annotation-free framework for identifying disease-specific tissue structures.
The model reduces manual labeling demands and inter-/intra-observer variability in histology.