Dual energy CT (DECT) has become a hot topic for its high detection precision and robust material identification ability in the field of nuclear safety and security inspection. However, the high cost of the system becomes a big limitation for its wide usage. To solve this problem, in 2009, we have proposed a dual energy CT reconstruction method with reduced data (DECT-RD) requiring much fewer data to reduce the cost of detectors. However, it is a simple idea without more analyzing in the process of solving ill-posed equations. In this paper, we tried to solve ill-posed equations with constraint condition (DECT-RDCC) and least squares (DECT-RDLS) respectively. Numerical simulations are done by using DECT-RD, DECT-RDCC and DECT-RDLS in the same situation, only 7 dual energy detector bins instead of 256 complete bin sampling in each projection. Results demonstrated that DECT-RDCC with relative error less than 1.1% is better than DECT-RD with relative error less than 1.8% while DECT-RDLS plays a more exact and steady role with relative error less than 0.6% than DECT-RDCC. Hence, DECT-RDLS is a better method used to obtain much lower system cost. We believe this work will drive DECT into wide usage.
Modified kernel MLAA using autoencoder for PET-enabled dual-energy CT