Method and device for segmenting blood vessel wall and blood flow region in vascular oct image

Abstract

Embodiments of the present invention provide a method and device for segmenting a blood vessel wall and a blood flow area in a blood vessel OCT image, wherein the method includes: based on a blood vessel Doppler OCT intensity map, using the first step in a cascaded fully convolutional neural network model The first-level fully convolutional neural network is used to segment the contour map of the blood vessel wall; based on the contour map of the blood vessel wall, the background noise processing is performed on the blood vessel Doppler OCT phase map, and based on the processed blood vessel Doppler OCT phase Figure, using the second-level fully convolutional neural network in the cascaded fully convolutional neural network model to segment the contour map of the bleeding area. The embodiment of the present invention uses a cascaded fully convolutional network to perform Doppler OCT blood vessel feature extraction, and combines the Doppler signal to make the segmentation of the blood flow region more accurate, and can effectively reduce the difficulty of segmentation of the blood vessel wall and the blood flow region in the blood vessel OCT image. , it is convenient and effective to segment the blood vessel wall and blood flow area in the image.

Description

technical field [0001] Embodiments of the present invention relate to the technical field of image processing, and more specifically, to a method and device for segmenting blood vessel walls and blood flow regions in a blood vessel OCT image. Background technique [0002] Optical coherence tomography (OCT) is an imaging technique with micron-level spatial resolution and millimeter-level imaging depth, which matches the characteristic size of vascular anastomosis. 3D imaging. In addition, Doppler OCT can simultaneously provide three-dimensional structure and blood flow information of blood vessels. [0003] At present, there are many biomedical image segmentation technologies based on deep learning. When the Doppler OCT system is used to obtain mouse arterial images, due to the blood scattering in the blood vessels, and as the depth increases, the sensitivity of the system decreases and the image quality Severe noise, low contrast and other factors, especially the blurred l...

AI Technical Summary

Problems solved by technology

[0003] At present, there are many biomedical image segmentation technologies based on deep learning. When the Doppler OCT system is used to obtain mouse arterial images, due to the blood scattering in the blood vessels, and as the depth increases, the sensitivity of the system decreases, and the image is blurred. Severe noise, low contrast and other factors, especially the blurred lower edge of the vessel wall in the vessel intensity map, make it difficult to completely and accurately segment the vessel wall, and a large amount of background noise appears in the phase map, making it difficult to accurately segment the blood flow area

Images