Magnetic resonance imaging method and device based on generative adversarial network

Abstract

The invention provides a magnetic resonance imaging method and device based on a generative adversarial network, and the magnetic resonance imaging method comprises the steps: building a correspondingrelation between undersampled MRI data and the image features of an MRI image through the self-learning capability of an artificial neural network; specifically, determining the correlation between data segments corresponding to adjacent time sequences; determining a target spatial feature in the under-sampled MRI data; determining a corresponding relationship between the under-sampled MRI data and the image features of the MRI image according to the correlation and the target spatial features; obtaining current under-sampling MRI data of a current detected person; determining image featuresof a current MRI image corresponding to the current under-sampled MRI data through the corresponding relationship; and specifically, determining image features of a current MRI image corresponding tocurrent undersampled MRI data, and determiniing the image features of the MRI image corresponding to the undersampled MRI data the same as the current undersampled MRI data in the corresponding relation as the image features of the current MRI image. A large amount of residual noise is prevented from being generated in the reconstruction process.

Description

technical field [0001] The present application relates to the field of medical detection, in particular to a magnetic resonance imaging method and device based on a generative adversarial network. Background technique [0002] As a medical imaging method widely used in clinic, Magnetic Resonance Imaging (MRI, Magnetic Resonance Imaging) can provide repeatable, non-invasive and quantitative tissue measurement, and provide information including structure, anatomy and function. A major drawback currently preventing MRI from being more widely used is the long acquisition time, which is mainly related to its inherent acquisition speed. MRI acquisition data is acquired through the frequency domain information of K space (including the frequency information acquired line by line) rather than in the image domain. Excessively long acquisition times make MRI results susceptible to patient motion and physiological movements, such as heart beats, respiratory excursions, and gastrointes...

AI Technical Summary

Problems solved by technology

[0004] The main shortcomings of the existing technical solutions are as follows: the end-to-end training method ignores the correlation between adjacent two-dimensional slices in the MRI data sequence; the dynamic training method lacks the k-space information that cannot effectively use a single image ; based on compressed sensing reconstruction produces a lot of residual noise

Images