dPIRPLE: a joint estimation framework for deformable registration and penalized-likelihood CT image reconstruction...

Area of Science:

• Medical Imaging
• Image Reconstruction
• Computational Anatomy

Background:

• Sequential imaging studies utilize prior scans for patient-specific anatomical information, aiding in radiation dose reduction.
• Patient motion between scans causes misregistration, degrading image quality and reconstruction accuracy.
• Existing methods often use rigid registration, which is insufficient for complex anatomical motion.

Purpose of the Study:

• To develop a joint framework for estimating 3D deformation and reconstructing images using prior information.
• To improve image quality and accuracy in sequential imaging by addressing misregistration due to patient motion.

Main Methods:

• Proposed a joint framework: deformable prior image registration, penalized-likelihood estimation (dPIRPLE).
• Integrated a 3D B-spline-based free-form-deformation model into the registration-reconstruction objective function.
• Employed an alternating maximization strategy for joint optimization of registration and reconstruction.

Main Results:

• dPIRPLE demonstrated superior reconstruction accuracy and image quality compared to traditional methods (FBP, PLE, PIPLE, PIRPLE).
• Experiments were conducted on a cone-beam CT testbench simulating lung nodule surveillance.
• The method showed effectiveness across various sampling sparsities and exposure levels.

Conclusions:

• The dPIRPLE framework effectively handles complex anatomical motion for improved image reconstruction.
• This approach offers a significant advancement for radiation dose reduction and maintaining image quality in sequential imaging.
• dPIRPLE shows promise for clinical applications requiring accurate anatomical representation from serial imaging.