Maximum-Likelihood Calibration of an X-ray Computed Tomography System

Area of Science:

• Medical Imaging
• Computational Imaging
• Physics

Background:

• Accurate geometrical calibration is crucial for high-quality X-ray computed tomography (CT) imaging.
• Traditional calibration methods may use reduced datasets or require initial parameter estimates, which can be problematic.
• Adaptive CT systems introduce dynamic geometrical changes during acquisition, necessitating robust calibration techniques.

Purpose of the Study:

• To introduce a novel maximum-likelihood (ML) method for calibrating the geometrical parameters of X-ray CT systems.
• To develop a calibration technique that utilizes the complete image dataset for improved accuracy.
• To address the challenges of calibrating CT systems with dynamically changing geometries, such as adaptive CT.

Main Methods:

• A maximum-likelihood (ML) estimation framework was employed for geometrical parameter calibration.
• The method utilizes the full image data, not a reduced subset, for comprehensive analysis.
• A contracting-grid algorithm was implemented, eliminating the need for initial parameter guesses.

Main Results:

• The proposed ML method effectively calibrates X-ray CT system geometry.
• Utilizing full image data enhances the robustness and accuracy of the calibration process.
• The contracting-grid algorithm successfully performs estimation without requiring initial values, simplifying the process.

Conclusions:

• The presented ML calibration method offers a powerful tool for X-ray CT systems, especially adaptive ones.
• This approach improves calibration accuracy and reliability by leveraging complete image data.
• The algorithm's independence from initial values makes it practical and broadly applicable in various CT scenarios.