An automatic organ delineation algorithm based on deep learning

Abstract

The present invention discloses an automatic organ delineation algorithm based on deep learning, which specifically includes the following steps: (1) using an image part classification algorithm to screen out the image of the current part of the organ; (2) obtaining the window position and size of the organ in the image According to the window position and size of the organ in the image, the screened image is clipped to a fixed size; (3) the clipped image is interpolated to the size required by the neural network input; (4) the processed image is input to The trained 3DUnet convolutional neural network is used to predict the organ; (5) deinterpolate the predicted organ image to the size of the cropped image, and fill it with the original image size to obtain the real prediction of the organ; (6) extract the real prediction The edge line of the organ is the outline of the organ. On the basis of ensuring the delineation accuracy, the present invention significantly improves the delineation efficiency, and can effectively solve the problem of low delineation precision of small organs.

Description

technical field [0001] The invention belongs to the field of medical images, and is mainly used for the delineation of organs at risk in radiotherapy, and specifically relates to an automatic organ delineation algorithm based on deep learning. Background technique [0002] In the delineation of organs at risk in radiotherapy, the traditional method in hospitals is to let radiologists manually delineate organs at risk, which not only is prone to human error, but also affects the work efficiency of radiotherapists, which in turn affects the treatment time of patients. [0003] Deep learning is a kind of artificial intelligence algorithm, which is essentially a neural network. Because of its many hidden layers, it is called deep learning. A large number of hidden layers allow it to extract high-dimensional features of the data. After continuous training, it can have the ability to extract high-dimensional features corresponding to the labels of the data, and then obtain a mappi...

AI Technical Summary

Problems solved by technology

[0006] In order to solve the above technical problems, the present invention proposes an automatic organ delineation algorithm based on deep learning, which can effectively solve the problem of low delineation accuracy of small organs, and has high delineation accuracy on large and small organs

Images