Automatic cerebral artery sketching method based on deep neural network

Abstract

The invention discloses an automatic cerebral artery sketching method based on a deep neural network, belongs to the technical field of image data processing, particularly relates to the technical field of cerebral artery sketching, and aims to solve the problems that a network model consumes huge video memory and the calculation speed is obviously reduced when a graph convolutional neural network is adopted for modeling in the prior art. The method comprises the following steps: step 1, inputting MRI scanning image data of a patient; 2, constructing a six-class cerebral artery segmentation model; and step 3, model training and testing. A 3D cascaded novel network model is provided, and specific artificial feature constraint network training of continuous frames, blood vessel center lines and the like is adopted, so that calculation resources are saved, and cerebral arteries can be predicted and sketched more quickly and accurately; after the whole model is trained, prediction can be completely and automatically carried out, and an ROI image of seven types of labels (the background occupies one type) which is consistent with the original MRI output size is output.

Description

technical field [0001] A method for automatically delineating cerebral arteries based on a deep neural network. The invention belongs to the technical field of image data processing, and specifically relates to the technical field of delineating cerebral arteries. Background technique [0002] Brain tumor resection in neurosurgery is often limited by the intracranial arteries adjacent to the brain. At present, the intraoperative lesion and important brain tissue structure navigation system is not yet mature. Doctors can only use magnetic resonance (MRI), magnetic resonance angiography (MRA) and other technical means to pre-judge the resection area in the preoperative stage and then specify the surgical plan. The total number of slices from the transverse, sagittal, and coronal perspectives of each MRI scanning sequence can reach as many as 600. The ratio (signal-to-noise ratio) of brain arteries to other tissue areas is about 1:500, and the area of ​​brain arteries is in the...

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to provide a method for automatically drawing cerebral arteries based on a deep neural network, so as to solve the problem that the existing network model uses a graph convolutional neural network for modeling, and the network model consumes a lot of video memory and the calculation speed is obviously slowed down.

Images