Simulating Reduced-Dose CT Imaging Using An Image-Based Virtual Sinogram Approach: Comparisons with Real Sinogram-Based Simulation
C Kim1*, J Kim2, M McNitt-Gray3, D Zhang4, (1) Interdisciplinary Program of Bioengineering Major Seoul National University College of Engineering,Seoul,(2) Seoul National University College of Medicine, Seoul, ,(3) UCLA School of Medicine, Los Angeles, CA, (4) UCLA School of Medicine, Los Angeles, CATH-E-217BCD-8 Thursday 1:00:00 PM - 2:50:00 PM Room: 217BCD
Low dose CT simulation is promising technique which can be applied to dose optimization of the clinical CT images without repeating CT examinations for patients to obtain images at different dose levels. We present a virtual sinogram approach which enables low-dose CT simulation study using only conventional CT image data without requiring raw sinogram data. Also presented are comparison results of virtual sinogram-based and real sinogram-based simulated reduced-dose CT images.
A virtual sinogram is generated by performing line integral of the CT number-based attenuation value, and a separate noise sinogram is generated using a noise model which incorporating X-ray photon flux depending on mAs, system electronic noise, and the virtual sinogram. A synthetic noise CT data is generated by applying FBP of the noise sinogram using an appropriate filter depending on reconstruction kernel of original CT. Finally, a simulated low-dose CT image is generated by adding the synthetic noise CT data to the original CT data.
Two phantoms for use in QC study were scanned with different mAs levels, and virtual sinogram-based and real sinogram-based approaches were applied to generate simulated reduced dose images. For quantitative assessment, non-subtracted noise power spectrum (NPS) and standard deviation on selected ROIs were compared.
Visual comparison made it clear that the degree of streak noise pattern in image as well as the level of noise magnitude agrees very well. The %difference of standard deviations between real and simulated images ranged from 0 to 7.5%. Also NPS measurements validate the proper working of the proposed low-dose simulator.
Proposed method could produce realistic reduced-dose CT images and its similarity was confirmed visually and quantitatively. Proposed method might be a useful tool for assessing low-dose CT application in various clinical settings even when raw sinogram data is not available.