Analyzing and processing method for analyzing and processing magnetic resonance image of acute ischemic stroke

Abstract

The invention provides an analyzing and processing method for analyzing and processing the magnetic resonance image of the acute ischemic stroke the method specifically comprises the steps of according to the data of a magnetic resonance image, acquiring the data of the magnetic resonance image, conducting the gold standard treatment on the data of the magnetic resonance image, setting the data of the image after the gold standard treatment as a positive sample set, obtaining a negative sample based on the data of the image after the gold standard treatment, obtaining a negative sample based on the data of the image in an initial ischemic region, obtaining a negative sample set, extracting the feature vectors of the positive sample set and the negative sample set, combining the extracted feature vectors as a training dictionary, obtaining to-be-segmented voxels based on the data of the image in an initial ischemic candidate region, and judging whether to-be-segmented voxels are ischemic voxels or not according to the training dictionary. According to the technical scheme of the invention, based on the analyzing and processing method for analyzing and processing the magnetic resonance image of the acute ischemic stroke, an ischemic region can be segmented more accurately. Meanwhile, the segmentation accuracy is high and the influence of noises or artifacts is small.

Description

technical field [0001] The invention relates to the field of image analysis and processing, in particular to a method for analyzing and processing magnetic resonance images of acute ischemic stroke. Background technique [0002] Ischemic stroke is the most common type of stroke. How to make a qualitative and localized diagnosis of the ischemic area of ​​ischemic stroke within the shortest time after onset, so as to guide the correct clinical diagnosis and treatment, has always been the focus of neuroimaging research. [0003] Existing MRI diffusion-weighted imaging includes: T2-weighted image (B0 image) with a diffusion coefficient of 0, and a diffusion coefficient of 1000-1500s / mm 2 The diffusion-weighted image DWI and the apparent diffusion coefficient ADC calculated by the two can reflect the imaging of the diffusion characteristics of water molecules and are very sensitive to the diffusion movement of water molecules. It is obviously better than conventional MRI in the ...

AI Technical Summary

Problems solved by technology

In the first case, due to the influence of T2 penetration effect, some ischemic areas show abnormally high signal on DWI, which is not obvious, and the boundary is blurred and the distribution is not uniform

In the second case, the interference of artifacts

The false normalization of ADC may be manifested by the fact that the corresponding ADC value in the ischemic area does not decrease or the degree of decrease is not obvious. In addition, the area where ADC decreases does not necessarily correspond to ischemia

These will affect the segmentation accuracy of the ADC threshold segmentation method

[0006] Existing method 2: Prakash et al. use the grayscale histogram of the ischemic image to align the asymmetry of the sagittal plane, that is, the divergence (divergencemeasure, DM) to find the area where the cerebral ischemia is located, and then find the appropriate grayscale threshold to distinguish Ischemia and non-ischemia on DWI, so it is basically a global gray-level threshold segmentation method, which is difficult to adapt to the uneven gray level of DWI images in the hyperacute phase

Images