XCal: model-based approach to X-ray CT spectral calibration.

Area of Science:

• Medical Imaging
• Physics
• Computational Science

Background:

• Accurate 3D object reconstruction in X-ray computed tomography (CT) relies on precise system calibration to account for the X-ray spectrum.
• Existing spectral calibration methods are often ill-posed and necessitate frequent recalibration when scanner parameters are altered, posing practical challenges.

Purpose of the Study:

• To develop a robust and adaptable spectral calibration approach for X-ray CT systems.
• To overcome the limitations of existing methods that require recalibration upon changes in scanner settings.

Main Methods:

• Propose XCal, a multi-energy, model-based spectral calibration technique for X-ray CT.
• Model the effective X-ray spectrum using a separable, physics-based CT system model.
• Estimate model parameters by fitting calibration data from known objects across multiple energy levels.

Main Results:

• XCal enables spectral calibration without the need for recalibration when scanner settings (e.g., source voltage, filters) are modified.
• Evaluations using simulated and measured datasets show XCal significantly enhances the accuracy of spectrum estimation compared to conventional methods.

Conclusions:

• XCal offers a significant advancement in X-ray CT spectral calibration, providing improved accuracy and operational flexibility.
• The proposed method simplifies the calibration process, making quantitative 3D reconstruction more reliable and efficient across varying experimental conditions.