Spatial-temporal image-constrained lung 4D-CT reconstruction for radiotherapy planning

Thoracic radiotherapy planning is increasingly dependent on 4D computed tomography (CT), which acquires axial images in multiple respirator phases and reconstructs them into 3D CT images based on respiratory signals. However, large reconstruction errors or artifacts may be observed due to poor reproducibility of breathing cycles. In this paper, 4D-CT reconstruction of helical mode CT scanning is achieved by incorporating spatial continuity and longitudinal smoothness of anatomical structures, such as chest surface, bone, vessel, and lung fields. The objective is to optimize the assignment of each axial image into different respiratory phases so that the artifacts or spatial discontinuity of anatomical structures are minimized, and the anatomical structures maintain their longitudinal consistency. In experiments, we compared our results visually and quantitatively with the current surrogate-based, image-matchingbased, and chest surface-constrained methods. The results showed that the proposed algorithm yields better helical mode 4D-CT than other proposed methods