Three-dimensional medical image super-resolution reconstruction method and system

Abstract

The invention discloses a three-dimensional medical image super-resolution reconstruction method and system, and belongs to the field of medical image processing. According to the method, a three-dimensional super-resolution problem is decomposed into combination of super-resolution reconstruction of a single slice and content correlation between adjacent slices, super-resolution reconstruction is carried out on three-dimensional medical image data through the multi-channel two-dimensional convolutional neural network, and high correlation between the adjacent slices is fully considered in a multi-channel network structure. Parameters of an image super-resolution reconstruction part are trained through a large number of two-dimensional high-resolution medical images, then the parameters of the reconstruction part are frozen, and a small amount of three-dimensional data is used for training weight parameters of a multi-channel output layer. Because the multi-channel network only needs to train parameters such as the weight of the sequence image of the new output layer, network training can be completed only through a small number of three-dimensional high-resolution images. The two-dimensional network is adopted essentially, compared with a three-dimensional network, the required three-dimensional data is greatly reduced, and the training difficulty is also greatly reduced.

Description

technical field [0001] The invention belongs to the field of medical image processing, and more particularly relates to a method and system for super-resolution reconstruction of three-dimensional medical images. Background technique [0002] Three-dimensional medical image super-resolution reconstruction is an important content in medical image processing. Three-dimensional medical images currently exist widely in clinics, and their usual imaging acquisition methods are tomographic imaging, such as CT and UCT. However, for 3D tomography, the spacing between scanning layers of 3D images is generally significantly greater than the interval between adjacent pixels in a layer, that is, the resolution between scanning layers is significantly lower than that of images within a scanning layer, and due to human motion and mechanical errors, Misalignment often occurs between scanned slice layers and layers, so in the non-scanned plane, the image will appear jagged and the image res...

AI Technical Summary

Problems solved by technology

Among them, the traditional two-dimensional convolutional network can improve the quality of the image itself to a certain extent, but usually cannot improve the dislocation phenomenon and ensure better spatial continuity. Algorithms such as VDSR and EDSR cannot well consider three-dimensional data when applied to three-dimensional reconstruction. Although the three-dimensional neural network has sufficient super-resolution accuracy, such as DDSR and ReCNN, feature extraction is performed through the three-dimensional convolution kernel, but the three-dimensional convolutional network has problems such as insufficient data, high training overhead, and low efficiency. , so the three-dimensional convolutional network has great limitations in application

And for the field of medical image processing, it is difficult to obtain clinical high-resolution 3D medical images

Images