Magnetic resonance diffusion weighted imaging method and device

The invention discloses a magnetic resonance diffusion weighted imaging method and device. A to-be-processed image is reconstructed based on DWI data collected by exciting a clinical tested tissue form times; a neural network is trained in advance and can output an output image with quality higherthan an input image, so that after the to-be-processed image with lower quality is input into the neural network, a high quality DWI image, that is, an output image corresponding to the to-be-processed image, can be acquired rapidly,; and the neural network improves the DWI image quality and shortensDWI image reconstruction time. Data for reconstruction of the to-be-processed image is DWI data collected by exciting the clinical tested tissue for m times, the excitation number of time m is a positive integer greater than or equal to 1, or can be very small, for example, the m is 4, 2, even 1, therefore, scanning time can be shortened effectively. Compared with the prior art, the method and the device can acquire high