Program Information
KV X-Ray Spectrum Estimation From Transmission Measurements Using Quadratic Programming
B YANG*, H GENG , T Chiu , W Lam , W Wong , K Cheung , S Yu , Hong Kong Sanatorium & Hospital, 2 Village Road, Happy Valley, Hong Kong
Presentations
SU-J-CAMPUS-IT-3 (Sunday, July 30, 2017) 4:00 PM - 5:00 PM Room: Imaging ePoster Theater
Purpose: In diagnostic x-ray imaging, knowledge of x-ray spectrum is important for radiation dose calculation, beam-hardening correction and dual energy computed tomography. One attractive measurement-based method is to estimate the x-ray spectrum by transmission measurements of a phantom of known composition. In this study, we propose to use quadratic programming to solve the problem.
Methods: Transmission method usually consists of two steps. The first step is to measure the transmission data for one material with different thicknesses. In our study, aluminum was chosen. The second step is to reconstruct the spectrum by solving a linear equation that is formularized by discretizing the problem. Two reference spectra were calculated by monte carlo simulation and spectral model using interpolating plynomials, respectively. The transmission data for aluminum were simulated from the two reference spectra. In addition, Gaussian random white noises within [-1, 1] with a level of 0.005 were added to the calculated attenuation curve for simulating the real situation. By applying proper constraints and regulations, quadratic programming was then used to solve the linear system.
Results: All of the solved spectra were close to the reference spectra with reasonable shape. The root mean square deviations (RMSD) of the solved spectra from their reference spectra were 0.05% and 0.21%, respectively. The transmission data were also calculated based on the solved spectra, which were in excellent agreement with those ideal transmission data with noise added. The RMSD of the derived transmission data were 0.22% and 0.24%.
Conclusion: This study demonstrated that quadratic programming works well in estimating the x-ray spectrum from the transmission measurements and is robust enough when noisy data are involved.
Contact Email: