Feasibility study for 3D cone-beam computed tomography reconstruction with few projection data using MLEM algorithm with total variation minimization

Abstract

Since, the CBCT (cone-beam computed tomography) was developed in 1967, it has been widely used in medical field. The CBCT has relatively simple geometry and less time consuming characteristics to obtain three-dimensional volume data. On the other hand, the major disadvantage of CBCT system is higher radiation exposure to the patient. Thus, it is necessary to apply proper reconstruction method to reduce patient dose.

In this study, simulated sparse projection data was obtained by GATE (Geant4 application for tomography emission) v6.0 simulation tool and those two dimensional projections were reconstructed by MLEM (maximum-likelihood expectationmaximization) algorithm. The total variation minimized image processing method was applied to improve the image quality. The images were evaluated by CNR and FWHM. Furthermore, absorbed dose in simulated phantom was evaluated by GATE v6.0 dose actor.

The images reconstructed by MLEM algorithm and total variation minimized method result in high image quality despite of sparse of projection. Especially image obtained by 60 projections showed comparable image quality with that from 360 projections.

In conclusion, sparse-projection reconstruction with MLEM and total variation minimization is a useful approach to reduce radiation dose with maintaining quality of reconstructed images.