Medical image noise reduction method based on generative adversarial network and 3D residual encoding and decoding

Abstract

The invention discloses a medical image noise reduction method based on a generative adversarial network and 3D residual coding and decoding, and the method comprises the steps: carrying out the classification collection of training data, carrying out the preprocessing of the training data, wherein the training data comprises a low-quality image and a high-quality image; constructing a convolutional neural network based on a generative adversarial network and 3D residual encoding and decoding; taking a low-quality image with scanning time of 75s and size of N * 9 * 64 * 64 * 1 as training input, taking a high-quality image with scanning time of 150s and size of N * 9 * 64 * 64 * 1 as a training label, and training the network; and carrying out noise reduction on the high-noise image by using the trained convolutional neural network to obtain a high-quality image. By applying the technical scheme of the invention, any high-noise positron emission computer tomography image can be accurately and rapidly denoised after the model is trained by using a small amount of data.

Description

technical field [0001] The present invention relates to positron emission computed tomography image processing, in particular, to a medical image noise reduction method based on generative confrontation network and 3D residual coding and decoding. Background technique [0002] Positron emission tomography (PET) is a functional imaging mode that observes the molecular activity level in tissues by injecting a specific radioactive tracer. 18F-FDG is a commonly used radioactive tracer. This substance will interact with the human body Negative electrons in the tissue are annihilated, and a pair of positrons with equal energy but opposite flight directions are emitted. The detector can realize the imaging function by detecting the electron trajectory. When a disease occurs in a certain part of the human body, more active physiological activities will increase the absorption of the substance, which is different from other normal tissues. PET is widely used in clinical practice, in...

AI Technical Summary

Problems solved by technology

[0004] In view of the problems in the prior art that image details are easily lost and the imaging speed is slow, the present invention provides a medical image noise reduction method based on generative confrontation network and 3D residual coding and decoding. After using a small amount of data to train the model Accurate and fast denoising of any noisy positron emission tomography image

Images