Establishing error bounds for internal calibration of quantitative computed tomography

Area of Science:

• Radiology and Imaging Science
• Biomedical Engineering
• Osteoporosis Research

Background:

• Opportunistic computed tomography (CT) scans can identify osteoporosis but lack calibration phantoms.
• Internal calibration methods for CT scans face challenges with beam hardening and heterogeneous tissues.
• Accurate bone mineral density (BMD) assessment is crucial for osteoporosis diagnosis.

Purpose of the Study:

• To introduce an advanced internal calibration method for CT scans that accounts for variations.
• To estimate error bounds for BMD measurements derived from internally calibrated CT scans.
• To assess the accuracy and reliability of the proposed internal calibration technique.

Main Methods:

• Developed a modified internal calibration method incorporating Monte Carlo simulation and error propagation.
• Applied the method to 138 clinical abdominal CT scans, using phantom calibration as ground truth.
• Evaluated 10 different internal reference tissue combinations to determine optimal referents.

Main Results:

• The internal calibration method, using air, skeletal muscle, and cortical bone, provided accurate BMD estimates.
• Error bounds were established and successfully contained the ground truth phantom-calibrated BMD.
• Mean BMD differences were minimal (2.12 mg/cc and 1.13 mg/cc for left and right femur).

Conclusions:

• The proposed internal calibration method with error bounds enhances confidence in opportunistic CT-based BMD measurements.
• This approach effectively addresses sources of error like beam hardening and tissue heterogeneity.
• Provides a reliable tool for osteoporosis screening using existing CT infrastructure.