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 Video

Updated: Apr 16, 2026

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

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

Published on: May 20, 2020

Deep Learning-Based Tumor Cell Classification in Lung Adenocarcinoma With a Case-by-Case Human-in-the-Loop Approach.

Martin Stampe^1,2, Pia Klausen¹, Ida Skovgaard Christiansen¹

¹Department of Pathology, Rigshospitalet, Copenhagen, Denmark.

APMIS : Acta Pathologica, Microbiologica, Et Immunologica Scandinavica

|April 14, 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

Are lazertinib and osimertinib comparable for treating advanced <i>EGFR</i>-mutant non-small cell lung cancer?-insights and limitations from the MARIPOSA study.

Journal of thoracic disease·2026

Same author

Survival and relapse of Danish patients with thymic epithelial tumors.

Acta oncologica (Stockholm, Sweden)·2026

Same author

Histological fidelity and microenvironmental kinome signatures of metastatic patient-derived organoids.

Frontiers in bioengineering and biotechnology·2026

Same author

Regulatory elements in the Sox9 locus license the initiation of pancreatic ductal adenocarcinoma.

Cell reports·2026

Same author

Correlation of TP53 Mutations With p53 Expression and Ki-67 Index in Pulmonary Large Cell Neuroendocrine Carcinomas.

APMIS : acta pathologica, microbiologica, et immunologica Scandinavica·2026

Same author

Rationale and Design of the MIA Trial: A Proof-of-Concept Study of Minimally Invasive Autopsy in an Infectious Cause of Death in a Scandinavian ICU Setting.

APMIS : acta pathologica, microbiologica, et immunologica Scandinavica·2026

Same journal

Immune Cell Composition and Prognosis in Node-Positive, Irradiated Breast Cancer Patients in the DBCG-IMN2 Study.

APMIS : acta pathologica, microbiologica, et immunologica Scandinavica·2026

Same journal

Urinary Secretory-IgA and IgG Responses and Inflammatory Markers in Patients With Pseudomonas aeruginosa Urinary Tract Infections.

APMIS : acta pathologica, microbiologica, et immunologica Scandinavica·2026

Same journal

Predictive Immunohistochemistry in Cholangiocarcinoma: Current Clinical Utility, Practical Limitations, and Emerging Directions.

APMIS : acta pathologica, microbiologica, et immunologica Scandinavica·2026

Same journal

Active Chlorine Antiseptics in Wound Care: Recent Advances and Perspectives for Military Medicine.

APMIS : acta pathologica, microbiologica, et immunologica Scandinavica·2026

Same journal

A Review on the Logistics, Financial, Ethical, and Regulatory Frameworks of Artificial Intelligence in Digital Pathology.

APMIS : acta pathologica, microbiologica, et immunologica Scandinavica·2026

Same journal

UE, CT, and ACR TI-RADS Grading in the Combined Diagnosis of Benign and Malignant Thyroid Nodules.

APMIS : acta pathologica, microbiologica, et immunologica Scandinavica·2026

See all related articles

This summary is machine-generated.

A human-in-the-loop deep learning approach enhances cell classification in lung adenocarcinoma (LUAD) histology. This method, fine-tuned by pathologists, achieves accuracy comparable to human experts for digital pathology applications.

Area of Science:

Digital Pathology
Computational Pathology
Machine Learning in Histology

Background:

Deep learning networks for classifying single cells in hematoxylin and eosin-stained (H&E) sections show limited accuracy.
Pathologist accuracy for cell classification is typically 98%-100%, setting a high benchmark for digital methods.

Purpose of the Study:

To develop and evaluate a human-in-the-loop approach for fine-tuning deep learning cell classification in H&E-stained histological slides.
To achieve digital classification accuracy comparable to that of a pathologist in routine pathology settings.

Main Methods:

A human-in-the-loop strategy was implemented, allowing pathologists to fine-tune cell classification during microscopy.
The approach was tested on H&E-stained slides of pulmonary adenocarcinomas (LUADs), using networks pretrained on typical cells.

Keywords:

cell classification convolutional neural network digital pathology histology

More Related Videos

Orthotopic Transplantation of Syngeneic Lung Adenocarcinoma Cells to Study PD-L1 Expression

Orthotopic Transplantation of Syngeneic Lung Adenocarcinoma Cells to Study PD-L1 Expression

Published on: January 19, 2019

Related Experiment Videos

Last Updated: Apr 16, 2026

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

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

Published on: May 20, 2020

Orthotopic Transplantation of Syngeneic Lung Adenocarcinoma Cells to Study PD-L1 Expression

Orthotopic Transplantation of Syngeneic Lung Adenocarcinoma Cells to Study PD-L1 Expression

Published on: January 19, 2019

Performance was evaluated on high-power fields (HPFs) and a dataset from The Cancer Genome Atlas (TCGA).

Main Results:

Pretrained networks achieved an average true positive rate (TPR) of 0.989 for LUAD tumor cell classification within HPFs.
Consistent results across three HPFs showed mean Dice coefficient of 0.991, TPR of 0.987, precision of 0.996, and specificity of 0.988.
Networks trained and tested on TCGA LUAD data yielded mean Dice of 0.951, TPR of 0.944, precision of 0.96, and specificity of 0.951.

Conclusions:

Deep learning classification of LUAD tumor cells, when trained and tested on the same data source, is comparable to pathologist accuracy.
The human-in-the-loop approach yields interpretable, validatable, and applicable results for routine digital pathology.
This method significantly improves the relevance and accuracy of automated cell classification in histological analysis.