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: Jan 15, 2026

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

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

Published on: August 16, 2020

Utility of Same-Modality, Cross-Domain Transfer Learning for Malignant Bone Tumor Detection on Radiographs: A

Joe Hasei¹, Ryuichi Nakahara², Yujiro Otsuka^3,4,5

¹Department of Medical Informatics and Clinical Support Technology Development, Graduate School of Medicine, Dentistry and Pharmaceutical Sciences, Okayama University, Okayama 700-8558, Japan.

|October 16, 2025

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

Successful implant salvage of a delayed tibial megaprosthesis infection using continuous local antibiotic perfusion: a case report.

International journal of surgery case reports·2026

Same author

Larger preoperative medial meniscus extrusion is associated with insufficient pain relief following pullout repair for medial meniscus posterior root tears.

Journal of experimental orthopaedics·2026

Same author

A modified Delphi consensus on tenosynovial giant cell tumour and giant cell tumour of bone : a report from the Birmingham Orthopaedic Oncology Meeting (BOOM).

The bone & joint journal·2026

Same author

Distinct epidemiology and treatment outcomes between skeletal and extraskeletal Ewing sarcoma in Japan: a population-based study.

Japanese journal of clinical oncology·2026

Same author

Complications of PI to PIII hemipelvic resections for intermediate and malignant tumours : a systematic review and meta-analysis.

Bone & joint open·2026

Same author

Dynamic Positional Changes in the Popliteal Artery and Vastus Medialis and Lateralis Muscles During Knee Flexion and Extension: An Open MRI-Based Anatomical Study.

Diagnostics (Basel, Switzerland)·2026

Summary

This summary is machine-generated.

Transfer learning for AI in rare diseases like bone tumors enhances practical performance. While not improving overall accuracy, it reduces false positives and increases detection rates, making AI tools more clinically useful.

Area of Science:

Radiology
Artificial Intelligence
Medical Imaging

Background:

Developing AI for rare diseases like malignant bone tumors is challenging due to limited annotated data.
Same-modality cross-domain transfer learning is explored to overcome data scarcity.
This study compares AI models for detecting bone tumors using transfer learning versus training from scratch.

Purpose of the Study:

To evaluate the effectiveness of transfer learning in AI models for detecting malignant bone tumors on knee radiographs.
To compare the performance of an AI model pretrained on chest radiographs with one trained from scratch.

Main Methods:

Two YOLOv5-based AI detectors were used: one initialized with transfer learning (YOLO-TL) and one trained from scratch (YOLO-SC).
Models were trained and validated on institutional data and tested on an external dataset of 743 knee radiographs.

Keywords:

YOLO artificial intelligence cross-domain learning diagnostic imaging malignant bone tumors radiographs transfer learning

More Related Videos

Modeling Primary Bone Tumors and Bone Metastasis with Solid Tumor Graft Implantation into Bone

Modeling Primary Bone Tumors and Bone Metastasis with Solid Tumor Graft Implantation into Bone

Published on: September 9, 2020

Related Experiment Videos

Last Updated: Jan 15, 2026

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

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

Published on: August 16, 2020

Modeling Primary Bone Tumors and Bone Metastasis with Solid Tumor Graft Implantation into Bone

Modeling Primary Bone Tumors and Bone Metastasis with Solid Tumor Graft Implantation into Bone

Published on: September 9, 2020

Performance was primarily assessed by Area Under the Curve (AUC) and evaluated at specific operating points (high-sensitivity, high-specificity, Youden-optimal).

Main Results:

Overall AUC was similar between YOLO-TL (0.954) and YOLO-SC (0.961).
At high-sensitivity and high-specificity thresholds, YOLO-TL demonstrated significantly improved specificity and sensitivity, respectively.
YOLO-TL reduced false positives by approximately 17 cases among 475 normal radiographs.

Conclusions:

Transfer learning can enhance the practical performance of AI models for rare diseases, even if overall AUC is not improved.
The transfer learning approach offers superior clinical utility by maintaining high detection rates and reducing false positives.
Same-modality cross-domain transfer learning is an efficient strategy for developing robust AI systems for rare disease detection.