CT projection geometric correction method based on deep learning

Abstract

The invention discloses a CT projection geometric correction method based on deep learning, and the method comprises the following steps: A1, obtaining CT original projection data through CT scanning;A2, enabling the CT original projection data to pass through a neural network, and outputting a geometric offset vector; A3, correcting and reconstructing the CT original projection data according tothe geometric offset vector to obtain a CT image without motion artifacts, so that the geometric correction offset can be automatically obtained in real time, and the method can be used for correcting a projection or reconstruction system matrix, removing artifacts caused by mechanical or patient motion and realizing more accurate diagnosis and treatment.

Description

technical field [0001] The invention relates to the field of CT imaging, in particular to a CT projection geometry correction method based on deep learning. Background technique [0002] Mechanical shaking or patient breathing, heartbeat and other movements during CT scanning will cause motion artifacts in the reconstructed image, seriously affecting diagnosis and treatment. For rigid motion, external motion detection equipment is usually used for geometric correction or calculation based on projection image information. Geometric offset; for non-rigid movements such as breathing and heartbeat, the mainstream solution is to use a respiratory detector and an electrocardiogram to detect breathing and heartbeat, so that CT can scan at the same time when breathing is still and heartbeat. The disadvantage is that an external phantom is needed ,higher cost. [0003] At present, there are also some technologies that do not require external detectors, but use algorithms to perform ...

AI Technical Summary

Problems solved by technology

[0005] The technical problem to be solved by the present invention is that the existing geometric correction method has the problem of time-consuming calculation or scanning, and the problem that real-time correction cannot be realized. The present invention provides a method based on The CT projection geometric correction method based on deep learning, the offset of CT projection mutual misalignment can be predicted by the neural network based on deep learning, and the geometric correction offset can be obtained automatically in real time, which can be used to correct the projection or reconstruct the system matrix to realize the CT projection. Geometric correction, removing artifacts caused by machinery or patient movement, to achieve more accurate diagnosis and treatment; using the sinusoidal characteristics of CT projections, and the sensitivity of regression-like convolutional neural networks to positions, the projection deviations at various angles are automatically obtained in real time The offset is used to correct the projection and eliminate the geometric motion artifact of the reconstructed image. The present invention can improve the calculation speed of the geometric correction, and does not need to use an external phantom, and the cost is lower, which is used to solve the defects caused by the existing technology.

Images