Low‐dose 4DCT reconstruction via temporal nonlocal means

• DOI: 10.1118/1.3547724
• Publication Date: 2011-03-04T14:55:04Z
• Authors: Zhen Tian; Xun Jia; Bin Dong; Yifei Lou; Steve B. Jiang
• Keywords: Regularization; Reconstruction algorithm; Image-guided radiation therapy

Purpose: Four‐dimensional computed tomography (4DCT) has been widely used in cancer radiotherapy for accurate target delineation and motion measurement tumors the thorax upper abdomen areas. However, its prolonged scanning duration causes a considerable increase of radiation dose compared to conventional CT, which is major concern clinical application. This work develop new algorithm reconstruct 4DCT images from undersampled projections acquired at low mA s levels order reduce imaging dose. Methods: Conventionally, each phase reconstructed independently using filtered backprojection (FBP) algorithm. The basic idea authors' that by utilizing common information among different phases, input required image high quality, thus dose, can be reduced. authors proposed temporal nonlocal means (TNLM) method explore interphase similarity. All phases are simultaneously minimizing cost function consisting data fidelity term TNLM regularization term. utilized modified forward‐backward splitting Gauss–Jacobi iteration efficiently solve minimization problem. was also implemented on graphics processing unit (GPU) improve computational speed. reconstruction tested digital NCAT phantom three scenarios: with level, level. Results: In all scenarios, generates visually much better CT containing less noise streaking artifacts standard FBP Quantitative analysis shows comparing algorithm, contrast‐to‐noise ratio improved factor 3.9–10.2 signal‐to‐noise 2.1–5.9, depending cases. situation projection data, majority streaks suppressed total time ten slice ranges 40 90 an NVIDIA Tesla C1060 GPU card. Conclusions: experimental results indicate outperforms effectively reducing due undersampling suppressing