Multi-site phantomless bone mineral density from clinical quantitative computed tomography in males

Area of Science:

• Radiology
• Orthopedics
• Biomedical Engineering

Background:

• Quantitative CT (QCT) is standard for volumetric bone mineral density (vBMD) assessment.
• Traditional QCT requires phantom rods, which are not always available.
• Phantomless methods, including internal-reference phantomless (IPL) and Hounsfield unit (HU) measurements, offer alternatives.

Purpose of the Study:

• To evaluate the accuracy of phantomless QCT methods (IPL and regression-based phantomless (RPL) using HU) compared to phantom-based (PB) methods.
• To assess the performance of these methods across multiple skeletal sites.
• To validate HU as a proxy for vBMD calculation.

Main Methods:

• QCT scans from 100 male post-mortem human subjects (PMHS) were analyzed.
• vBMD was calculated using site-specific PB calibration.
• IPL and RPL vBMD were determined using a development sample (50/100 PMHS) for calibration and validated on the remaining subjects.

Main Results:

• PB and phantomless vBMD (IPL/RPL) showed no significant differences (p > .05).
• Univariate regressions between PB and phantomless vBMD were significant (p < .05), except for IPL Rad-30 (p = .078).
• RPL vBMD demonstrated strong correlation with PB vBMD (R² = 99.17%) with smaller differences compared to IPL vBMD (R² = 97.99%).

Conclusions:

• Phantomless QCT methods, particularly RPL, are accurate and comparable to PB methods for vBMD assessment.
• Site-specific calibration is recommended for phantomless methods to enhance accuracy.
• RPL calibration improves the utility of opportunistic QCT for bone quality and fracture risk assessment.