Technical Note: Intrinsic raw data-based CT misalignment correction without redundant data

Area of Science:

• Medical Imaging
• Image Reconstruction
• Computational Imaging

Background:

• Accurate geometry is crucial for CT image reconstruction; misalignment causes artifacts and degrades image quality.
• Traditional methods often require dedicated calibration phantoms or redundant data, limiting applicability.
• Intrinsic methods offer a phantom-less approach but traditionally need extensive angular data.

Purpose of the Study:

• To propose and evaluate an intrinsic, raw data-based method for computed tomography (CT) misalignment correction.
• To enable misalignment correction without requiring a calibration phantom or redundant data.
• To develop a method applicable to CT systems with limited angular scan ranges.

Main Methods:

• A nonlinear transform is applied to the reconstructed volume to introduce raw data inconsistencies.
• These inconsistencies are then used to estimate geometric parameters for correction.
• The method is validated using simulations (FORBILD head phantom) and experimental micro-CT data (mouse scan).

Main Results:

• The proposed method successfully corrects misalignment artifacts in both simulated and real-world noisy data.
• The correction is achieved without using redundant data, preserving raw data fidelity.
• Evaluations confirm the method's effectiveness even with limited angular data.

Conclusions:

• The developed intrinsic method effectively corrects CT misalignment artifacts using raw data.
• It significantly extends the applicability of intrinsic correction methods to limited angular ranges ($$).
• This advancement is particularly beneficial for CT systems with restricted scan capabilities.