The Feasibility of the Dual-Dictionary Method for Breast Computed Tomography Based On Photon-Counting Detectors
B Zhao1*, H Ding1, Y Lu2, G Wang3, J Zhao2, S Molloi1, (1) University of California Irvine, Irvine, CA, (2) Shanghai Jiao-Tong University, Shanghai, China, (3) Virginia Tech, BLACKSBURG, VATU-E-217BCD-9 Tuesday 2:00:00 PM - 3:50:00 PM Room: 217BCD
To investigate the feasibility of an Iterative Reconstruction (IR) method utilizing the algebraic reconstruction technique coupled with dual-dictionary learning for the application of dedicated breast computed tomography (CT) based on a photon-counting detector.
Postmortem breast samples were scanned in an experimental fan beam CT system based on a Cadmium-Zinc-Telluride (CZT) photon-counting detector. Images were reconstructed from various numbers of projections with both IR and Filtered-Back-Projection (FBP) methods. Contrast-to-Noise Ratio (CNR) between the glandular and adipose tissue of postmortem breast samples were calculated to evaluate the quality of images reconstructed from IR and FBP. In addition to CNR, the spatial resolution was also used as a metric to evaluate the quality of images reconstructed from the two methods. This is further studied with a high-resolution phantom consisting of a 14 cm diameter, 10 cm length polymethylmethacrylate (PMMA) cylinder. A 5 cm diameter coaxial volume of Interest insert that contains fine Aluminum wires of various diameters was used to determine spatial resolution.
The spatial resolution and CNR were better when identical sinograms were reconstructed in IR as compared to FBP. In comparison with FBP reconstruction, a similar CNR was achieved using IR method with up to a factor of 5 fewer projections.
The results of this study suggest that IR method can significantly reduce the required number of projections for a CT reconstruction compared to FBP method to achieve an equivalent CNR. Therefore, the scanning time of a CZT-based CT system using the IR method can potentially be reduced.