An accurate method for direct dual-energy calibration and decomposition

Area of Science:

• Medical Physics
• Image Processing
• Computational Imaging

Background:

• Dual-energy imaging relies on accurate decomposition of material components.
• Conventional methods often require extensive calibration data and can be sensitive to noise.
• Existing approximations may lack smoothness or correct asymptotic behavior.

Purpose of the Study:

• To develop and validate conic and cubic surface equations for direct approximation of dual-energy equations and their inverses.
• To assess the accuracy, speed, and robustness of these novel approximation methods.
• To compare the performance against traditional polynomial approximations.

Main Methods:

• Direct approximation of dual-energy integral equations using conic (second-order) and cubic (third-order) surface equations.
• Evaluation using simulated and real calibration data from split-detector systems.
• Analysis of noise effects and derivation of accuracy-related formulas.

Main Results:

• The conic surface equation (eight-term rational form) offers a fast, accurate, and noise-robust decomposition algorithm.
• Calibration requires as few as 9 points, or 16 points for robust calibration with an accuracy check.
• The cubic surface equation (eighteen-term form) theoretically achieves extreme accuracy (<10⁻⁶ log-signal, 1 micron thickness) with a closed-form solution.

Conclusions:

• Conic and cubic surface equations provide efficient and accurate solutions for dual-energy decomposition.
• These methods significantly reduce calibration requirements and enhance robustness to noise.
• The proposed approach offers a superior alternative to conventional polynomial approximations for dual-energy analysis.