Capsule endoscopy image multi-focus detection method

A technology of capsule endoscopy and detection methods, applied in image enhancement, image analysis, image data processing, etc., can solve problems such as difficulty in feature extraction, low precision of multi-lesions, limited number of training set samples, etc., to reduce the burden of diagnosis, The effect of high detection accuracy and reduced calculation load

Pending Publication Date: 2020-10-16
CHANGSHA UNIVERSITY OF SCIENCE AND TECHNOLOGY
View PDF3 Cites 3 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] The existing computer-aided diagnosis system design methods cannot meet the actual needs. The main defects of these methods are: only for a single lesion or a few kinds of lesions, and a few methods for multiple lesions are not accurate.
The number of training set samples used to extract features is limited, which makes feature extraction difficult

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
  • Capsule endoscopy image multi-focus detection method
  • Capsule endoscopy image multi-focus detection method
  • Capsule endoscopy image multi-focus detection method

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0038] Such as figure 1 As shown, a multi-focal detection method for capsule endoscopic images based on visual feature fusion includes the following complete steps:

[0039] Capsule endoscopy image data collection: Capsule endoscopy images include normal images, bleeding images, ulcer images, polyp images, erosion images, vein exposure images, and edema images. Store these images in separate labeled folders.

[0040] Image preprocessing: Preprocessing the acquired images, removing image borders, deleting useless image information, and randomly shuffling the data set before feature extraction.

[0041] Color feature extraction: extract the third-order color moment color feature of the image in the HSV color space, and perform HSV color space conversion on the image; then divide the entire picture evenly into 25 non-overlapping sub-blocks:

[0042] Then the third-order color moment on each sub-block area includes: first-order moment (mean value), second-order moment (variance)...

Embodiment 2

[0049] Texture feature extraction:

[0050] Such as figure 1 The LBP texture histogram texture feature of the image extracted in the RGB color space is shown, and the LBP calculation formula is:

[0051]

[0052]

[0053] g in the formulac Represents the gray value of the center point pixel; g n Indicates the gray value of the pixels within the neighborhood radius, and R is the neighborhood radius. The image texture feature extraction is to extract the uniform local binary texture histogram of the image as the texture feature in the RGB color space. First, the image is divided into 100 complementary and overlapping sub-block images, and then the texture histogram of each sub-block is extracted separately. Finally, according to Sequentially concatenate the features of each region to form the texture features of the entire complete image, with a total of 5900 (59×100) dimensions.

[0054] Extract image convolutional neural network features: There are two methods for ext...

Embodiment 3

[0061] CNN feature and traditional feature fusion:

[0062] Such as figure 1 As shown, the present invention adopts four kinds of feature fusion methods, which are series fusion, series fusion and PCA dimensionality reduction, canonical correlation analysis fusion, and multi-layer perceptron fusion. Concatenation fusion refers to concatenating four features together. Concatenation and PCA dimensionality reduction fusion refers to concatenation of four features and then PCA dimensionality reduction. Canonical correlation analysis fusion means that two of the four features are fused into one feature through canonical correlation analysis, and then this feature is further fused with one of the remaining features until the final fusion feature is formed. Multilayer perceptron fusion refers to the establishment of a three-layer multilayer perceptron network. The fully connected layer with 512 neurons in the fully connected layer before the softmax layer is used as the feature ext...

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 a capsule endoscopy image multi-focus detection method, belonging to the field of computer image processing. Based on depth feature and human selection feature fusion and multi-classifier integration, the method comprises the following steps of image preprocessing, construction of color, texture and form feature vectors, acquisition of depth features by adopting a convolutional neural network, feature fusion, multi-classifier integration and focus recognition; and the above steps are repeated until all capsule endoscopic images are recognized. According to the capsule endoscopy image multi-focus detection method, the diagnosis burden of clinicians can be effectively reduced, and detection accuracy is improved.

Description

technical field [0001] The invention relates to the field of computer image processing, in particular to a multi-focus detection method for capsule endoscopic images. Background technique [0002] Capsule endoscopy, English name: capsule endoscopy, is a capsule-shaped endoscope, which is a medical instrument used to examine the human intestinal tract. Capsule endoscopes can enter the human body and are used to spy on the health status of the human stomach and esophagus. Can be used to help doctors diagnose patients. Capsule endoscopy actually shrinks the camera and implants a medical capsule to help doctors diagnose patients. A small capsule is a camera studio for exploring the human body, and may even become a "spaceship" for traveling in the human body; tiny fibers can be used to strengthen the heart arteries of the human body. From the outside, it is not much different from ordinary capsule medicine, but it is a miniature camera, which is used to spy on the health stat...

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
Patent Type & Authority Applications(China)
IPC IPC(8): G06T7/00G06K9/62G06N3/04G06N3/08
CPCG06T7/0012G06N3/08G06T2207/10024G06T2207/20081G06T2207/20221G06T2207/30028G06T2207/30092G06T2207/30096G06N3/045G06F18/2411
Inventor 贾智伟章黎明谢俊力田奕宏赖志强
Owner CHANGSHA UNIVERSITY OF SCIENCE AND TECHNOLOGY
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap