Unlock instant, AI-driven research and patent intelligence for your innovation.

Automatic segmentation method and system for retinal image lesion area

A lesion area, automatic segmentation technology, applied in the direction of image analysis, image enhancement, image data processing, etc., can solve the problems of long time, inaccurate segmentation results, etc., achieve low system cost, improve generalization ability, and low hardware requirements Effect

Inactive Publication Date: 2019-07-09
SHANDONG NORMAL UNIV
View PDF6 Cites 2 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

The inventor found during the research and development process that the existing manual segmentation method is used to segment the CSC lesion area in the OCT retinal image, but it takes a lot of time and relies on manual experience, resulting in inaccurate segmentation results

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Automatic segmentation method and system for retinal image lesion area
  • Automatic segmentation method and system for retinal image lesion area
  • Automatic segmentation method and system for retinal image lesion area

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0037] This embodiment provides a method for automatically segmenting retinal image lesion regions based on a fully convolutional neural network. The FCN-8S fully convolutional neural network structure is used to simultaneously segment two regions of interest, and a two-step compensation method is proposed, which improves the model in Classification accuracy in data sets with large differences, and the Jaccard distance loss function is extended to three classifications; this method has high accuracy, and uses a two-step compensation method to improve the generalization ability of the training model. The requirements are low, the system cost is low, and it can be reused.

[0038] Please refer to the attached figure 1 , the method for automatically segmenting retinal image lesion regions comprises the following steps:

[0039] S101. Collect OCT retinal images and classify them.

[0040] Specifically, in order to provide a reliable training classification standard for the neura...

Embodiment 2

[0076] This embodiment provides an automatic retinal image lesion region segmentation system based on a fully convolutional neural network, which simultaneously segments two regions of interest through an image compensation segmentation module, adopts a two-step compensation method, and improves the performance of the model in datasets with large differences. The classification accuracy is high, and the Jaccard distance loss function is extended to three classifications, which has low requirements for hardware, low system cost, and can be reused.

[0077] Please refer to the attached Figure 6 The system for automatically segmenting retinal image lesions includes an image acquisition module, a classification module, a model training module, an image compensation segmentation module, an image threshold segmentation module, and an ellipsoid zone layer and Bruch's membrane layer segmentation module.

[0078] Specifically, the image collection module is used to collect 757 retinal...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention discloses an automatic segmentation method and system for a retinal image lesion area, and aims to simultaneously segment a central serous fluid choroid lesion area and an ellipsoid zone. Brush film regions. The method comprises the following steps: acquiring a retina image, and configuring a training set and a test set; configuring a full convolutional neural network model, modifying a loss function, and training the full convolutional neural network model by using the image data in the training set; performing preliminary segmentation on the image data in the test set by usingthe trained full convolutional neural network model, and performing two-step compensation on a preliminary segmentation result to obtain a probability graph; segmenting the probability graph again toobtain EZ-; a BM region and a CSC lesion region; eZ-extraction by adopting binary image edge detection method Ellipsoidal band of bm region And obtaining a segmentation result of the ellipsoid belt layer and the Brusher film layer through upper and lower boundaries of the Brusher film area.

Description

technical field [0001] The present disclosure relates to the field of image segmentation, in particular to a method and system for automatic segmentation of ellipsoidal bands and Bruch's membrane regions in OCT retinal images based on fully convolutional neural networks. Background technique [0002] Central serous chorioretinopathy (CSC) is a common retinopathy, especially in young and middle-aged men, that causes neurosensory retinal serous detachment secondary to the retinal pigment epithelium (RPE). Serous detachment between the ellipsoidal zone (EZ) and Bruch's membrane (BM) often occurs, resulting in abnormal protrusion of the ellipsoidal zone. [0003] Optical coherence tomography (OCT) is a commonly used technique for observing retinal morphology and is a good way to observe changes in diseases. The technology has the advantages of non-invasive, depth resolution, volume imaging and so on. During the research and development process, the inventor found that the exis...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
IPC IPC(8): G06T7/11G06T7/136G06T7/13G06N3/04
CPCG06T7/11G06T7/136G06T7/13G06T2207/10101G06T2207/20081G06T2207/30041G06N3/045
Inventor 李登旺王卓牛四杰孔问问吴敬红薛洁陈美荣刘婷婷黄浦赵睿
Owner SHANDONG NORMAL UNIV