Related Experiment Video
Updated: Jul 14, 2026
09:36
Outer-Boundary Assisted Segmentation and Quantification of Trabecular Bones by an Imagej Plugin
Published on: March 14, 2018
A local adaptive threshold strategy for high resolution peripheral quantitative computed tomography of trabecular
Andrew J Burghardt1, Galateia J Kazakia, Sharmila Majumdar
1Musculoskeletal Quantitative Imaging Research Group, Department of Radiology, University of California, San Francisco, CA 94158, USA. andrew.burghardt@radiology.ucsf.edu
Annals of Biomedical Engineering
|July 3, 2007
Summary
High resolution peripheral quantitative computed tomography (HR-pQCT) offers detailed bone analysis. A new local thresholding method improves the accuracy of trabecular bone microstructure quantification compared to current hybrid and global thresholding techniques.
Area of Science:
- Biomedical Engineering
- Orthopedics
- Medical Imaging
Background:
- High resolution peripheral quantitative computed tomography (HR-pQCT) is a key in vivo imaging technique for assessing human bone microstructure.
- Current methods for quantifying trabecular bone microstructure using HR-pQCT rely on hybrid models with densitometric and plate assumptions.
- These legacy methods may introduce inaccuracies in bone characterization.
Purpose of the Study:
- To compare a novel local thresholding scheme for HR-pQCT image analysis against existing hybrid and global thresholding methods.
- To evaluate the accuracy of direct 3D methods for trabecular bone microstructure quantification.
- To determine if advanced segmentation techniques can improve HR-pQCT analysis without model assumptions.
Main Methods:
- Human femoral trabecular bone samples were imaged using HR-pQCT (82 µm voxels) and micro-tomography (16 µm voxels).
- HR-pQCT images were analyzed using three segmentation approaches: manufacturer's hybrid method (global threshold), direct 3D method (global threshold), and direct 3D method (novel local threshold).
- Results were validated against microCT-derived standard trabecular indices.
Main Results:
- All HR-pQCT methods showed strong correlations with microCT for standard trabecular parameters.
- The hybrid method underestimated bone volume fraction (BV/TV) and trabecular thickness (Tb.Th).
- Direct methods with global thresholding overestimated BV/TV and Tb.Th, while the novel local threshold method provided optimal intermediate results and superior performance for trabecular number (Tb.N) and spacing (Tb.Sp).
Conclusions:
- Direct 3D analysis of HR-pQCT data using advanced local thresholding segmentation offers improved accuracy for trabecular bone microstructure quantification.
- This novel approach minimizes reliance on densitometric and model-based assumptions, potentially enhancing clinical applicability.
- Direct methods with advanced segmentation present a more accurate alternative for in vivo bone analysis.
