Liver tumor image segmentation method of dense connection network based on high and low layer feature fusion

Abstract

The invention provides a liver tumor image segmentation method of a dense connection network based on high and low layer feature fusion. The method comprises the steps of 1, collecting nuclear magnetic images of a liver tumor patient to form a data set, and dividing the data set into a training set and a test set; 2, creating a segmentation model, wherein the segmentation model selects a dense connection network as a framework, and the dense connection network comprises an encoder, a decoder and a global attention module located between the encoder and the decoder; wherein the global attention module performs feature fusion on low-level features of the liver tumor image and high-level features of the liver tumor image, and learns to automatically evaluate the importance of the high-level features, so that a decoder guides the low-level features to recover image details by using classification information contained in important high-level semantic features; and 3, using the image data in the training set to train a segmentation model, and then using the trained segmentation model to test the image data in the test set to obtain a segmentation result of the liver tumor image.

Description

technical field [0001] The invention relates to the technical field of image processing, in particular to a liver tumor image segmentation method based on a densely connected network of high-level and low-level feature fusion. Background technique [0002] The segmentation of liver tumors in magnetic resonance images (MRI) has important clinical application value for the accurate diagnosis and subsequent treatment of tumors. Usually liver tumor image segmentation is done manually layer by layer by radiologists using professional software, which is not only inefficient but also time-consuming. Therefore, a fully automatic method for liver tumor segmentation is needed clinically. In the MRI image, the boundary between the liver tumor area and normal liver tissue is blurred, and the gray scale of adjacent organ tissues is very close. The contrast between the tumor area and the surrounding tissue is poor, and the gray scale is uneven, resulting in artifacts and blurred boundarie...

AI Technical Summary

Problems solved by technology

However, the above methods only involve simple high-level and low-level feature ...

Images