SEA-Net: Structure-Enhanced Attention Network for Limited-Angle CBCT Reconstruction of Clinical Projection Data

• Published in: IEEE transactions on instrumentation and measurement 2023, Vol.72, p.1-13
• Main Authors: Hu, Dianlin, Zhang, Yikun, Li, Wangyao, Zhang, Weijie, Reddy, Krishna, Ding, Qiaoqiao, Zhang, Xiaoqun, Chen, Yang, Gao, Hao
• Format: Article
• Language: English
• Subjects: Clinical projection data; Computed tomography; cone-beam computed tomography (CBCT); Convolutional neural networks; Feasibility studies; Feature extraction; Image enhancement; Image quality; Image reconstruction; Image restoration; Iterative methods; limited-angle (LA) computed tomography (CT); Medical imaging; Radiation; Radiation dosage; Radiation therapy; spatial attention (SA); structure enhancement; Texture; Tomography
• ISSN: 0018-9456; 1557-9662
• DOI: 10.1109/TIM.2023.3318712

This work aims to improve limited-angle (LA) cone-beam computed tomography (CBCT) by developing deep learning (DL) methods for real clinical CBCT projection data, which is the first feasibility study of clinical-projection-data-based LA-CBCT, to the best of our knowledge. In radiation therapy (RT), CBCT is routinely used as the onboard imaging modality for patient setup. Compared to diagnostic computed tomography (CT), CBCT has a long acquisition time, e.g., 60 s for a full 360° rotation, which is subject to the motion artifact. Therefore, the LA-CBCT, if achievable, is of great interest for the purpose of RT, for its proportionally reduced scanning time in addition to the radiation dose. However, LA-CBCT suffers from severe wedge artifacts and image distortions. Targeting at real clinical projection data, we have explored various DL methods such as image/data/hybrid-domain methods and finally developed the so-called structure-enhanced attention network (SEA-Net) method that has the best image quality from clinical projection data among the DL methods we have implemented. Specifically, the proposed SEA-Net employs a specialized structure enhancement subnetwork to promote texture preservation. Based on the observation that the distribution of wedge artifacts in reconstruction images is nonuniform, the spatial attention (SA) module is utilized to emphasize the relevant regions while ignoring the irrelevant ones, which leads to more accurate texture restoration.