TY - JOUR
T1 - Statistical Iterative CBCT Reconstruction Based on Neural Network
AU - Chen, Binbin
AU - Xiang, Kai
AU - Gong, Zaiwen
AU - Wang, Jing
AU - Tan, Shan
N1 - Funding Information:
Manuscript received April 8, 2018; revised April 17, 2018; accepted April 17, 2018. Date of publication April 25, 2018; date of current version May 31, 2018. This work was supported by the National Natural Science Foundation of China under Grant 61375018 and Grant 61672253. The work of J. Wang was supported in part by the Cancer Prevention and Research Institute of Texas under Grant RP160661 and in part by the National Institute of Biomedical Imaging and Bioengineering under Grant R01 EB020366. (Binbin Chen and Kai Xiang are co-first authors.) (Corresponding authors: Jing Wang; Shan Tan.) B. Chen, K. Xiang, Z. Gong, and S. Tan are with the Key Laboratory of Image Processing and Intelligent Control of Ministry of Education of China, Huazhong University of Science and Technology, Wuhan 430074, China, and also with the School of Automation, Huazhong University of Science and Technology, Wuhan 430074, China (e-mail: ace7chan@hust.edu.cn; u201314635@hust.edu.cn; m201472350@hust.edu.cn; shantan@hust.edu.cn).
Publisher Copyright:
© 2018 IEEE.
PY - 2018/6
Y1 - 2018/6
N2 - Cone-beam computed tomography (CBCT) plays an important role in radiation therapy. Statistical iterative reconstruction (SIR) algorithms with specially designed penalty terms provide good performance for low-dose CBCT imaging. Among others, the total variation (TV) penalty is the current state-of-the-art in removing noises and preserving edges, but one of its well-known limitations is its staircase effect. Recently, various penalty terms with higher order differential operators were proposed to replace the TV penalty to avoid the staircase effect, at the cost of slightly blurring object edges. We developed a novel SIR algorithm using a neural network for CBCT reconstruction. We used a data-driven method to learn the "potential regularization term" rather than design a penalty term manually. This approach converts the problem of designing a penalty term in the traditional statistical iterative framework to designing and training a suitable neural network for CBCT reconstruction. We proposed using transfer learning to overcome the data deficiency problem and an iterative deblurring approach specially designed for the CBCT iterative reconstruction process during which the noise level and resolution of the reconstructed images may change. Through experiments conducted on two physical phantoms, two simulation digital phantoms, and patient data, we demonstrated the excellent performance of the proposed network-based SIR for CBCT reconstruction, both visually and quantitatively. Our proposed method can overcome the staircase effect, preserve both edges and regions with smooth intensity transition, and provide reconstruction results at high resolution and low noise level.
AB - Cone-beam computed tomography (CBCT) plays an important role in radiation therapy. Statistical iterative reconstruction (SIR) algorithms with specially designed penalty terms provide good performance for low-dose CBCT imaging. Among others, the total variation (TV) penalty is the current state-of-the-art in removing noises and preserving edges, but one of its well-known limitations is its staircase effect. Recently, various penalty terms with higher order differential operators were proposed to replace the TV penalty to avoid the staircase effect, at the cost of slightly blurring object edges. We developed a novel SIR algorithm using a neural network for CBCT reconstruction. We used a data-driven method to learn the "potential regularization term" rather than design a penalty term manually. This approach converts the problem of designing a penalty term in the traditional statistical iterative framework to designing and training a suitable neural network for CBCT reconstruction. We proposed using transfer learning to overcome the data deficiency problem and an iterative deblurring approach specially designed for the CBCT iterative reconstruction process during which the noise level and resolution of the reconstructed images may change. Through experiments conducted on two physical phantoms, two simulation digital phantoms, and patient data, we demonstrated the excellent performance of the proposed network-based SIR for CBCT reconstruction, both visually and quantitatively. Our proposed method can overcome the staircase effect, preserve both edges and regions with smooth intensity transition, and provide reconstruction results at high resolution and low noise level.
KW - CBCT
KW - image reconstruction
KW - neural network
KW - regularization term
UR - http://www.scopus.com/inward/record.url?scp=85045960259&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85045960259&partnerID=8YFLogxK
U2 - 10.1109/TMI.2018.2829896
DO - 10.1109/TMI.2018.2829896
M3 - Article
C2 - 29870378
AN - SCOPUS:85045960259
VL - 37
SP - 1511
EP - 1521
JO - IEEE Transactions on Medical Imaging
JF - IEEE Transactions on Medical Imaging
SN - 0278-0062
IS - 6
ER -