Time separation technique: Accurate solution for 4D C-Arm-CT perfusion imaging using a temporal decomposition model.

Area of Science:

• Medical Imaging
• Computational Imaging
• Radiology

Background:

• Perfusion imaging provides critical functional information for diagnosing conditions like stroke.
• Current dynamic CT perfusion imaging often requires high data acquisition and computational resources.
• Slowly rotating x-ray systems, such as C-arm cone beam CT (CB-CT), present challenges for dynamic imaging due to undersampling.

Purpose of the Study:

• To develop a temporal decomposition model for perfusion imaging using C-arm CB-CT.
• To integrate prior knowledge (mathematical models, clinical data) into dynamic CT reconstruction.
• To reduce the number of views, computational effort, and radiation dose in perfusion imaging.

Main Methods:

• A model-based approach using temporal basis functions to incorporate prior knowledge.
• Preprocessing in the projection space to decompose temporal and spatial domains.
• Transforming the high-dimensional dynamic CT problem into static CT tasks for easier reconstruction.

Main Results:

• Demonstrated accurate reconstruction of dynamic in silico data using conventional clinical parameters.
• The time separation technique achieved expected accuracy for model-based CT perfusion imaging.
• The 4D dynamic CT task was solved with computational speed comparable to static 3D tasks.

Conclusions:

• The presented algorithm enables model-based perfusion reconstruction for C-arm CB-CT without prohibitive computational costs.
• This approach makes model-based perfusion imaging potentially applicable in clinical practice.
• The study serves as a proof of concept for efficient dynamic CT perfusion imaging.