A liver image segmentation method based on a dense feature pyramid network

Abstract

The invention discloses a liver image segmentation method based on a dense feature pyramid network, and relates to the technical field of image processing. The method adopts a dense feature pyramid network to process a segmentation problem of a multi-stage liver CT image, the network introduces a feature pyramid based on a full-volume integral segmentation network, uses dense connection to enhancea feature flow, and realizes pixel-level liver segmentation on multi-stage CT; According to the method, the Dice value in a public 3DIRCADb database is 95.0%, so that the segmentation performance isimproved; A common data set and a clinical data set prove that the DPFN trained by using the CT image with the larger layer thickness can be seamlessly popularized to the CT image with the smaller layer thickness.

Description

technical field [0001] The invention relates to the technical field of image processing, in particular to a liver image segmentation method based on a dense feature pyramid network. Background technique [0002] Automatic liver segmentation in multiphase CT images is an important step in liver diagnosis and treatment planning. Accurate liver segmentation is essential in many clinical applications, such as diagnosis of liver diseases, functional testing and surgical planning. Manual segmentation is a heavy, error-prone, and time-consuming task, especially on large amounts of CT data. Therefore, automatic segmentation of the liver is very necessary. However, this is a challenging task due to the complex anatomy, blurred boundaries, and various shapes of the liver in CT images. [0003] At present, scholars have proposed several segmentation methods based on CT images. It can be mainly divided into non-machine learning and machine learning methods. Non-machine learning met...

AI Technical Summary

Problems solved by technology

However, the inventor found in the process of studying the segmentation of liver CT images that the current segmentation method still has low segmentation performance, and the segmentation models trained by CT images with different slice thicknesses are not compatible with each other.

Images