Compressed-sensing-based three-dimensional image reconstruction algorithm for C-arm vascular imaging

X-ray C-arm is an important imaging tool in interventional surgery, road-mapping and radiation therapy. It provides accurate description of vascular anatomy and therapy end point. The C-arm scanner produces two-dimensional (2D) x-ray projection data obtained with flat-panel detector by rotating the source around the patient. The number of 2D projections acquired is several hundreds, which results in significant amount of radiation dose. Unlike the conventional fluoroscopic imaging, three-dimensional (3D) C-arm computed tomography (CT) provides more accurate cross-sectional images which are valuable for therapy planning, guidance and evaluation in interventional radiology. However, 3D vascular imaging using the conventional C-arm fluoroscopy is a challenging task. First, the rotation orbit of the C-arm gantry is usually limited to a range less than those of CT scanners. Second, in several commercial models (including the one of consideration in this study), the x-ray source and detector are shifted from the gantry isocenter to enlarge the scanner field-of-view (FOV), which is so-called the offset scan. Finally, it is difficult to acquire sufficient projection views required for stable 3D reconstruction using manually controlled gantry motion. Inspired by the theory of compressed sensing, we developed an image reconstruction algorithm for the conventional angiography C-arm scanners. The main challenge in this image reconstruction problem is the projection data limitations. We consider a small number of views (less than 10 views) acquired from a short orbit with the offset scan geometry. The proposed method is developed using the alternating direction method of multipliers (ADMM) and results obtained from simulated data and real data are encouraging. The proposed method can significantly contribute to the reduction of patient dose and provides a framework to generate 3D vascular images using the conventional C-arm scanners.