GPU based parallel acceleration for fast C-arm cone-beam CT reconstruction.

Area of Science:

• Medical Imaging
• Computer-Aided Surgery
• Radiology

Background:

• Cone-beam CT (CBCT) with Flat Panel Detector technology is a novel imaging modality.
• C-arm CBCT is particularly suitable for image-guided interventional surgeries.
• Acquiring high-resolution, high-quality 3D images in real-time remains a challenge for CBCT.

Purpose of the Study:

• To develop a GPU-based accelerated method for rapid C-arm CBCT 3D image reconstruction.
• To enhance system performance and reduce reconstruction delay in clinical applications.
• To optimize the image reconstruction process for interventional surgery.

Main Methods:

• Implementation of a GPU-based parallel acceleration technique for a filtered back projection algorithm.
• Optimization of the Filtered Back Projection (FBP) method for C-arm CBCT.
• Design of a distributed system to mask reconstruction delay by utilizing image acquisition time.

Main Results:

• The GPU-accelerated FDK algorithm significantly improves reconstruction efficiency.
• System performance is enhanced by 26% and reconstruction delay is reduced 2.1 times (90 projections).
• Further acceleration of 39% in performance and 3.3 times in delay reduction achieved with 120 projections.

Conclusions:

• The proposed GPU-based acceleration method substantially increases C-arm CBCT system efficiency.
• The distributed system design effectively hides reconstruction delay, crucial for real-time applications.
• This approach addresses the challenge of real-time, high-quality 3D image reconstruction in image-guided surgery.