Liver blood vessel segmentation method based on CT image

Abstract

The invention discloses a liver blood vessel segmentation method based on a CT image, and the method comprises the steps of: firstly obtaining an original 3D liver image, and carrying out the preprocessing, and obtaining a training set; then using the training set obtained through preprocessing for training a 3D convolutional neural network, wherein the adopted 3D convolutional neural network adopts a Unet network structure, an encoder is provided with a side output layer for deeply supervising the structure of the convolutional network system, an output end is provided with two parallel branches, the upper branch is used for extracting features, different from the background, of the hepatic vein and the portal vein, and the lower branch is used for extracting features for distinguishing the hepatic vein and the portal vein; and finally processing the 3D liver image by using the trained 3D convolutional neural network to obtain a liver blood vessel segmentation result. According to the method, a side output layer is added from an encoder part to help bottom layer features to extract more semantic information, and meanwhile, two parallel branches are arranged at the output end, so that the segmentation effect of the hepatic vein and the portal vein in the liver image is improved.

Description

technical field [0001] The invention belongs to the field of medical image processing, and in particular relates to a liver blood vessel segmentation method based on CT images. Background technique [0002] Liver vascular segmentation and 3D reconstruction help to accurately obtain the overall information of abdominal liver vascular tissue, which is the premise of computer-aided liver disease diagnosis and liver surgery planning, and is of great significance for liver disease diagnosis and liver surgery guidance. CT (computed tomography) technology is one of the most commonly used diagnostic techniques for liver and blood vessel effects. Through CT, doctors can obtain a series of two-dimensional CT slices with enhanced blood vessels. However, abdominal CT images often have unfavorable factors such as low contrast, high signal-to-noise ratio, blurred boundaries, and adhesion between the liver and other tissues with similar gray levels. Due to the complex structure of liver ve...

AI Technical Summary

Problems solved by technology

Through CT, doctors can obtain a series of two-dimensional CT slices with enhanced blood vessels. However, abdominal CT images often have unfavorable factors such as low contrast, high signal-to-noise ratio, blurred boundaries, and adhesion between the liver and other tissues with similar gray levels. Due to the complex structure of liver vessels , blood vessels are entangled with each other, and there are great differences between different individuals. Liver blood vessel segmentation is facing great challenges

[0003] The existing 3D liver vessel segmentation methods can generally be divided into two categories based on grayscale and gradient, but single grayscale or gradient-based segmentation methods, such as 3D region growing, fuzzy clustering, etc., cannot effectively extract low-contrast images. Hepatic vein and portal vein; in the existing liver vessel segmentation models such as the active contour model and its hybrid model, it is easy to cross the weak boundary of the vessel and cause severe over-segmentation, and it is necessary to provide the initial area of ​​​​the vessel, and it is difficult to handle the segmentation of small vessels

Images