Task-driven source-detector trajectories in cone-beam computed tomography: II. Application to neuroradiology

Area of Science:

• Medical Imaging
• Radiology
• Cone-Beam Computed Tomography

Background:

• Cone-beam computed tomography (CBCT) is crucial for interventional neuroradiology.
• Optimizing source-detector trajectories can enhance image quality and diagnostic performance.
• Current CBCT trajectory planning often lacks task-specific optimization.

Purpose of the Study:

• To apply a task-driven framework for optimizing CBCT source-detector trajectories in neuroradiology.
• To maximize the detectability index ($$) for specific imaging tasks.
• To evaluate the performance of task-driven trajectories in simulated and real-world interventional scenarios.

Main Methods:

• Utilized the task-driven imaging framework from Stayman et al. for trajectory optimization.
• Applied the methodology to simulated endovascular embolization of aneurysms and arteriovenous malformations.
• Validated the approach on a CBCT test bench and an interventional robotic C-arm system.

Main Results:

• Task-driven trajectories resulted in an average increase in the detectability index ($$) of 7% to 28%.
• Optimized trajectories favored higher fidelity, less noisy views.
• Improved conspicuity of imaging stimuli was achieved by noise reduction and altered noise correlation.

Conclusions:

• The task-driven imaging framework significantly improves CBCT imaging performance in neuroradiology.
• Demonstrated the potential of task-driven trajectories for enhancing visualization in interventional procedures.
• Task-driven optimization is a promising approach for motorized, multiaxis C-arms in neuroradiology.