MRI motion artifact correction method based on local optimization generative adversarial network

Abstract

The invention discloses an MRI motion artifact correction method based on a local optimization generative adversarial network, and the method adds skip layer connection between an upper sampling module and a lower sampling module, enables the finally constructed motion artifact to be better in features, achieves the removal of motion artifacts, and can achieve the removal of motion artifacts better. The local loss of each group of photos is calculated based on the local optimization loss, so that the output image not only has the minimum global loss, but also can be optimal in a local area, the local consistency of the output image is reserved, and additional components are not required to be added. In addition, the model is optimized by combining an adversarial loss function, a gradient penalty loss function and a content loss function, so that the image corrected by the model has vivid texture information, structural information and richer detail information.

Description

technical field [0001] The invention relates to the technical field of medical image processing, in particular to an MRI motion artifact correction method based on a local optimization generation confrontation network. Background technique [0002] Medical imaging is widely used in modern medicine. Among them, MRI is widely used in clinical examination because it does not produce radiation to the human body and has a good detection effect on tumors. However, MRI takes a long time, and a full-body examination usually takes about half an hour. Therefore, compared with other types of medical imaging, MRI is more likely to be troubled by human motion during MRI. At the same time, since MRI is extremely sensitive to human body motion during acquisition, motion artifacts are often prone to appear in the final imaging results. The generated motion artifacts will adversely affect the doctor's diagnosis in clinical treatment, such as the loss or blurring of pathological information,...

AI Technical Summary

Problems solved by technology

However, MRI takes a long time, and a full-body examination usually takes about half an hour. Therefore, compared with other types of medical imaging, MRI is more likely to be disturbed by human movement

At the same time, since MRI is extremely sensitive to the movement of the human body during acquisition, motion artifacts are often prone to appear in the final imaging results.

The generated motion artifacts will adversely affect the doctor's diagnosis in clinical treatment, such as the loss or blurring of pathological information, which increases the risk of misdiagnosis, so it is necessary to avoid motion artifacts in the clinical application of MRI

Although a clear MRI can be obtained through re-acquisition, the time cost and economic cost of MR acquisition are very high

Images