Related Experiment Video
Updated: Jan 11, 2026
09:36
Outer-Boundary Assisted Segmentation and Quantification of Trabecular Bones by an Imagej Plugin
Published on: March 14, 2018
9.7K
Tracking temporal progression of benign bone tumors through X-ray based detection and segmentation.
Se-Yeol Rhyou1,2, Chohee Bang3, Yong Jin Cho4
1Department of Electrical and Computer Engineering, College of Information and Communication Engineering, Sungkyunkwan University, Suwon, 440-746, South Korea.
Scientific Reports
|November 11, 2025
Summary
This study introduces FusionX-BBTNet, an AI tool for automated analysis of benign bone tumors (BBTs) in X-rays. It accurately measures tumor size and shape changes over time, aiding clinical decisions.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Oncology
Background:
- X-ray imaging is crucial for diagnosing bone tumors but lacks objective longitudinal analysis.
- Manual assessment of tumor size and shape progression is subjective and time-consuming.
Purpose of the Study:
- To develop an automated deep learning framework, FusionX-BBTNet, for analyzing benign bone tumors (BBTs) in X-ray images.
- To enable quantitative, time-sequential assessment of BBT size and shape progression.
Main Methods:
- FusionX-BBTNet integrates YOLO object detection and U-Net segmentation for automated BBT detection and segmentation.
- A novel wavelet-enhanced dataset improves contour accuracy, and an OCR module extracts scale bars for real-world measurements.
- Centroid-based alignment visualizes changes in tumor size and area over time.
Main Results:
- The framework achieved high performance with a mean IoU of 0.9376 and a boundary F1 score of 0.9827 on 466 expert-annotated X-rays.
- Automated quantification of tumor size and area in millimeters was demonstrated.
- The system provides intuitive shape and area comparisons for clinical decision support.
Conclusions:
- FusionX-BBTNet offers a reliable, automated solution for longitudinal analysis of benign bone tumors from X-ray images.
- This AI-driven approach can enhance diagnostic efficiency and complement expert interpretation, particularly in resource-limited settings.
