Reconstruction-of-difference (RoD) imaging for cone-beam CT neuro-angiography

Area of Science:

• Medical Imaging
• Radiology
• Image Reconstruction

Background:

• Timely neurovascular evaluation using CT angiography (CTA) is crucial for detecting pathologies like ischemic stroke.
• Cone-beam CTA (CBCT-A) offers point-of-care advantages but faces challenges with gantry speed affecting image quality, data consistency, and sparsity.
• Existing reconstruction methods like filtered back-projection (FBP) and penalized likelihood (PL) have limitations in addressing these CBCT-A challenges.

Purpose of the Study:

• To describe and evaluate a novel reconstruction-of-difference (RoD) approach for CBCT-A.
• To assess the RoD method's robustness against challenges posed by slow gantry rotation speeds in CBCT-A systems.
• To compare the performance of RoD against FBP and PL under various imaging conditions.

Main Methods:

• Developed a fast digital simulation framework to test RoD performance across diverse imaging conditions, focusing on data consistency, sparsity, and peak contrast trade-offs.
• Conducted experiments using a CBCT prototype and an anthropomorphic neurovascular phantom to validate simulation findings with real-world data.
• Evaluated performance using normalized root mean square error (NRMSE) against ground truth, optimizing reconstruction parameters for fair comparison.

Main Results:

• The RoD approach reduced NRMSE by up to 50%-53% compared to FBP and 29%-31% compared to PL in simulations.
• Real-world experiments confirmed NRMSE reductions of 34% with RoD versus FBP and 17% versus PL.
• Identified optimal scan protocols for RoD, balancing data consistency, sparsity, and peak contrast for effective neurovascular imaging.

Conclusions:

• The RoD method demonstrates superior performance and robustness in 3D angiography compared to FBP and PL.
• RoD enables accurate reconstruction even with data sparsity and inconsistency, crucial for CBCT-A.
• This suggests the potential for CBCT-A on low-cost, mobile platforms for point-of-care neurovascular imaging.