Iterative CT Reconstruction Using Curvelet-Based Regularization

Zusammenfassung

There is a critical need to reconstruct clinically usable images at a low dose. One way of achieving this is to reconstruct with as few projections as possible. Due to the undersampling, streak artifacts degrade image quality for traditional CT reconstruction. Compressed sensing (CS) [1] theory uses sparsity as a prior and improves the reconstruction quality considerably using only few projections. CS formulates the reconstruction problem to an optimization problem. The objective function consists of one data fidelity term and one regularization term which enforce the sparsity under a certain sparsifying transform. Curvelet is an effective sparse representation for objects [2]. In this work, we introduce to use curvelet as the sparsifying transform in the CS based reconstruction framework. The algorithm was evaluated with one physical phantom dataset and one in vitro dataset and was compared against and two state-of-art approach, namely, wavelet-based regularization (WR) [3] and total variation based regularization methods (TVR) [4]. The results show that the reconstruction quality of our approach is superior to the reconstruction quality of WR and TVR.