Review of Medical Imaging Cross-Modal Generation Methods

Abstract: In the rapidly developing field of medical artificial intelligence, image generation algorithm based on deep learning has become one of the research hotspots. This paper aims to review the development status of four major image generation algorithms, namely autoregressive model, variational autoencoder, generative adversarial network and diffusion model, and analyze the application of generative model in medical multimodal image conversion from three modes: computed tomography, magnetic resonance imaging and computed tomography angiography.Generative model not only has broad application prospects in the field of medical imaging, but also has great value potential.