A 4D-CBCT Imaging Method Based on Motion Compensated Learning

Abstract

The invention discloses a 4D-CBCT imaging method based on motion compensation, belonging to the field of computer tomography. The present invention first obtains high-quality 4D-CBCT data of the patient, and divides it into sample and label data; then it is necessary to construct a motion-compensated learning convolutional neural network for 4D-CBCT data, which is used to establish the mapping between images of different phases; Next, the sample and label data are used as input to train the network to obtain the optimal network parameter weights; finally, the network is used to assist in the reconstruction of 4D-CBCT projection data under clinical scans to obtain high-quality reconstruction images. The present invention can greatly reduce reconstruction blur caused by respiratory movement and noise and artifacts caused by lack of data acquisition angle, shorten the scanning cycle and reduce the radiation damage suffered by the examinee, and meet the requirements of clinical analysis and diagnosis. Quality requirements to improve the tracking efficiency of lung tumors.

Description

technical field [0001] The present invention relates to the technical field of computed tomography, and more specifically, relates to a 4D-CBCT imaging method based on motion compensation learning. Background technique [0002] With the rapid development of medical technology, many new technologies are applied to the prevention and treatment of tumors, such as radiofrequency ablation, biological cell therapy, gene therapy, etc., but radiotherapy is still one of the three major methods for treating malignant tumors. In recent decades, the mode of clinical implementation of radiotherapy has undergone several major technological innovations. In the late 20th century, with the emergence of technologies such as Computed Tomography (CT) and Magnetic Resonance Imaging (MRI), radiation therapy plans began to change from two-dimensional to three-dimensional. In the 21st century, intensity-modulated radiation therapy has basically realized a fully automatic computer control mode. Ho...

AI Technical Summary

Problems solved by technology

[0007] The purpose of the present invention is to overcome the problem that the star-stripe artifact and noise cannot be effectively removed in the 4D-CBCT imaging method in the prior art, and provide a 4D-CBCT imaging method based on motion compensation learning, which is called motion compensation learning reconstruction (Motion Compensation Learning Reconstruction, referred to as MCLR), without increasing the cost of the existing CBCT hardware, through the study of key reconstruction techniques, combined with the training of the motion compensation network, to suppress image blurring and stripes caused by breathing motion and angle loss To improve the image quality of 4D-CBCT, so that image-guided radiotherapy can "see more clearly and grasp more accurately", thereby improving the accuracy of tumor radiotherapy and increasing treatment benefits

Images