Cyclic hopping deep learning network

Abstract

The invention relates to a cyclic hopping deep learning network, which is mainly used for accurately extracting different interested targets in medical images. According to the network, a new reverse short hopping link and an attention-guided convolution module are introduced on the basis of an existing BiO-Net segmentation network, so that a cyclic hopping deep learning network is constructed, and then local OCT image data and public fundus image data are sampled to verify the segmentation performance of the network. According to the invention, different interested targets in the image can be effectively extracted, and the network has segmentation performance superior to that of existing networks such as U-Net, AU-Net and BiO-Net.

Description

technical field [0001] The invention relates to the technical field of image segmentation, in particular to a loop-jumping deep learning network. Background technique [0002] Image segmentation is a process that divides an image into different regions and makes each region have completely different imaging characteristics (such as pixel gray distribution, tissue contrast, and anatomical shape, etc.), thereby reducing the difficulty of analysis and measurement of the region of interest, and providing a basis for the diagnosis and treatment of related diseases. Lesion location and morphological quantification, disease analysis, clinical diagnosis and prognosis monitoring provide key guidance, so it has important academic research value. In order to accurately extract the desired region of interest, a large number of image segmentation algorithms have been developed, such as threshold-based methods, active contour-based methods, and atlas-based methods. These segmentation alg...

AI Technical Summary

Problems solved by technology

[0006] (1) The design of reverse short jump links: In order to alleviate the problem of information loss caused by multiple image downsampling, existing segmentation networks (such as U-Net and BiO-Net) usually use skip links to encode and decode images. The convolutional features are connected in series to ensure that the decoding convolution module has enough information input and learns and captures the required target features from the input information.

Images