Medical image segmentation method based on deep learning

Abstract

The invention belongs to the technical field of image segmentation, and relates to a medical image segmentation method based on deep learning, which comprises the following steps: firstly, using a novel convolution module in the stages of an encoder and a decoder, and secondly, designing a residual bottleneck structure containing an attention mechanism, and using the residual bottleneck structurein skip layer connection, thereby reducing the semantic difference between the encoder and the decoder, and enabling the neural network to pay more attention to a target area to be segmented in the training process, so that finer semantic features can be extracted. The method is simple, the blurred boundary can be better recognized, the effect of the segmented image is more coherent, the noise interference resistance is high, and the generalization ability is high.

Description

Technical field: [0001] The invention belongs to the technical field of image segmentation, and relates to a medical image segmentation method based on deep learning, and a method for medical image segmentation using deep learning technology. Background technique: [0002] In the early days of image segmentation, most of them were based on graph theory or pixel clustering methods, and many classic algorithms such as K-Means algorithm were born. In terms of medical images, it is often based on edge detection and template matching, such as using Hough transform for optic disc segmentation. However, medical images are often derived from different imaging techniques, such as computed tomography (CT), X-ray, and magnetic resonance imaging (MRI), so these methods fail to remain robust when tested on large amounts of data. After the emergence of deep learning technology, it has solved the problem of lack of semantic information in traditional image segmentation methods to a certai...

AI Technical Summary

Problems solved by technology

However, continuous pooling operations may lead to the loss of some spatial information, and the convolution kernel scale in the convolution layer is too single to extract finer semantic features, which makes U-Net in some actual medical image segmentation In the scene, it is easy to be disturbed by noise, thus ignoring some details. For example, CN201910158251.5 discloses a brain tumor medical image segmentation method based on deep learning, including training a segmentation model, receiving brain tumor medical image data information to be segmented, Four processes are performed on the received brain tumor medical image data information to be segmented and the output of the segmented results; CN201810852143.3 discloses an image segmentation method based on deep learning, including step a: normalizing the original image, Step b: Input the normalized image into the ResUNet network model, the ResUNet network model extracts the feature map containing global semantic information in the input image, and performs upsampling and feature map stacking processing on the feature map to obtain The final feature map, step c: classify the feature map after the upsampling and stacking processing pixel by pixel, and output the image segmentation result; CN201910521449.5 discloses a lung tissue image segmentation method based on deep learning, through The improved deep learning method of Deeplabv3+ realizes the segmentation of X-ray chest X-ray lung tissue, and X-ray chest X-ray images are input into the segmentation model, wherein, the segmentation model is obtained by training with multiple sets of training data, and the multiple sets of training Each set of training data in the data includes: an X-ray chest image and a corresponding gold standard for identifying lung tissue; obtaining output information of the model, wherein the output information includes the X-ray chest image Segmentation results of lung tissue; CN201911355349.6 discloses a liver CT image segmentation method and device based on a deep learning neural network, the method includes: constructing a U-shaped DenseNet based on a dense convolutional network DenseNet and a U-shaped network Unet Two-dimensional 2D network and U-shaped DenseNet three-dimensional 3D network; based on the automatic context auto-context method, the U-shaped DenseNet 3D network is integrated into the U-shaped DenseNet 2D network, and the U-shaped DenseNet hybrid network is obtained for deep learning training; through training The U-shaped DenseNet hybrid network for liver CT image segmentation

Images