Patient specific respiratory motion model using two static CT images

Abstract

Computed Tomography (CT) has been extensively used for guiding percutaneous lung biopsy during image-guided intervention. However, due to respiratory motion, there having a difference between static images and current lung. Current studies are using the global model to predict lung movement in real time. This model extracts common features of lung motion based on group-level of imaging data, which overlook the information comes from the randomness of lung movements and individual differences on the features of lung motion and consequently, limit the precision of the model. In order to resolve this issue, patient specific model is proposed and can effectively acquire individual features of lung movement but required 4D CT image that increases the risk of radiation damage. This paper developed a new patient specific respiratory motion model to establish the mathematical relationship between lung internal motion and external chest surface motion. Only two static 3D CT are needed for a specific patient, and there is no need to collect 4D CT images. The proposed method combining the advantages of global and patient specific model will ensure better prediction of lung motions and lower the risk of radiation damage. The qualitative and quantitative result show that the proposed model achieved a better balance between predictive accuracy and radiation doses.