Enhanced bone structural analysis through pQCT image preprocessing.

Area of Science:

• Orthopedics
• Medical Imaging
• Bone Densitometry

Background:

• Peripheral quantitative computed tomography (pQCT) is crucial for assessing bone health.
• Image quality, influenced by preprocessing, impacts the accuracy of pQCT-derived bone traits.
• High noise levels in raw pQCT images present a significant challenge for reliable analysis.

Purpose of the Study:

• To evaluate the effectiveness of two novel image preprocessing techniques for pQCT bone analysis.
• To compare these novel methods against traditional median filtering (3x3 and 5x5).
• To determine the impact of preprocessing on the reliability of cortical and trabecular bone measurements.

Main Methods:

• Utilized repeated scans of liquid phantoms and in vivo scans of distal tibiae (n=25).
• Compared two novel preprocessing methods (histogram down-sampling, statistical approximation) with 3x3 and 5x5 median filtering.
• Assessed signal-to-noise ratio (SNR) and reliability coefficients (R) for raw and preprocessed images.

Main Results:

• Raw pQCT images exhibited low SNR (~20dB), challenging preprocessing.
• Cortical bone analysis reliability increased from 67% (raw) to 94-97% post-preprocessing, with no significant difference between methods.
• Trabecular density analysis showed high R-values (~99%) in raw images, with minimal improvement after preprocessing.
• Preprocessing revealed coarse structural patterns in trabecular regions, absent in raw images.

Conclusions:

• Image preprocessing significantly enhances the reliability of cortical bone analysis using pQCT.
• Preprocessing offers limited improvement for trabecular bone density measurements due to inherent image noise.
• Novel preprocessing methods can facilitate structural analysis of trabecular bone, complementing density assessments.