Classification model construction method and device for macular lesion region segmentation

Abstract

The invention discloses a method for constructing a classification model for segmenting macular lesion regions of fundus images, which comprises the following steps: selecting multiple fundus images, performing grayscale processing on them to obtain multiple grayscale images, and performing grayscale processing on the grayscale images Samples are obtained by sampling the foreground and background of the sample respectively; the transformation matrix is ​​obtained by using the generalized low-rank approximation method, and the dimensionality reduction process is performed on the sample based on the transformation matrix to obtain the low-rank approximation matrix of the sample; the low-rank approximation matrix of the sample is added to The label information is used as supervision to construct the manifold regularization item; the objective function is constructed by combining the generalized low-rank approximation method and the manifold regularization item, and the iterative optimization method is used to solve the objective function to obtain the optimal transformation matrix and the optimal low-rank approximation of the sample matrix; constructing a classification model based on the optimal low-rank approximate matrix and label information. The classification model of the present invention can extract feature descriptors that are both low-dimensional and highly distinguishable, and can improve segmentation accuracy.

Description

technical field [0001] The invention relates to the field of medical image processing, in particular to a classification model construction method, device and image segmentation method for macular degeneration region segmentation in fundus images. Background technique [0002] Eyes are the most important organ for human beings to obtain information. The macula is located at the back of the eyeball and is an important tissue for people to perceive external light and objects. If the lesion occurs in this part, it will cause vision loss or even blindness, which is one of the important causes of blindness in the elderly. When doctors diagnose the macular degeneration region (drusen) in fundus images, there are disadvantages such as low accuracy, poor repeatability, and many subjective factors. Therefore, there is an urgent need for the application and research of macular degeneration region segmentation technology to meet the clinical needs of auxiliary medical care such as sc...

AI Technical Summary

Problems solved by technology

[0005] However, there are still relatively few researches on the application of supervised feature learning methods in fundus image segmentation.

However, other hand-designed features for fundus image segmentation do not have strong distinguishing and descriptive capabilities, and cannot obtain more accurate segmentation results.

Images