Eureka AIR delivers breakthrough ideas for toughest innovation challenges, trusted by R&D personnel around the world.

Corona virus disease 2019 screening method and system based on deep learning

A technology of deep learning and pneumonia, applied in the medical field, can solve the problems of inapplicable detection of new coronary pneumonia, large resource consumption, cumbersome operation, etc., and achieve the effect of fast diagnosis, convenient operation, and improved detection rate

Pending Publication Date: 2020-09-11
ZHEJIANG UNIV
View PDF7 Cites 9 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, during its training, it is necessary to calculate the mask of the lesion area, which requires a lot of resources for marking, and the operation is relatively cumbersome. Therefore, the above method is not suitable for improving the early detection rate of COVID-19. Testing for Novel Coronavirus (COVID-19)

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

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0032] A method for screening new coronary pneumonia based on deep learning, specifically comprising the following steps:

[0033] S1. Extract the mask of the effective area of ​​​​the lungs

[0034] S2. Use the extracted mask to segment the effective area of ​​the lungs, reduce the influence of irrelevant areas, and normalize the Hu value of the original CT to 0-255;

[0035] S3. Use the 3D convolutional neural network to segment out the lung lesion area, and output the location information of the lesion area, including the coordinates (x, y, z) of the center point and the diameter d of the lesion area;

[0036] S4. To detect the two-dimensional image of the lesion area in the middle layer of the lesion area, and obtain the images adjacent to the upper and lower sides at the same time, a total of three images

[0037] S5. Send the above-mentioned classified images into a two-dimensional classification network for training to determine whether it is COVID-19, influenza A or n...

Embodiment 2

[0043] This embodiment is a web interface-based screening system for new coronary pneumonia designed on the basis of Embodiment 1, including:

[0044] Image acquisition module, used to collect CT original compressed files;

[0045] The decompression module is used to decompress the CT original compressed file collected by the image acquisition module;

[0046] A format judging module is used for judging the format of the decompressed file obtained by decompressing the decompression module, and screening and obtaining the CT image in the lung window format in the CT;

[0047] The data processing module is used to calculate and process the CT image in lung window format in the CT screened by the format judging module, and calculate the overall disease type and probability of a single patient;

[0048] The display module is used to display the processing result of the data processing module.

[0049] This embodiment develops a web application interface for uploading the origina...

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 corona virus disease 2019 screening method and system based on deep learning. The corona virus disease 2019 screening method comprises the steps: detecting a lung lesion areaof CT by using a deep learning detection model, and conveying the lung lesion area of CT into a three-classification network, wherein three classifications comprise COVID-19, influenza A and non-infection symptoms; and through calculation processing, outputting a CT diagnosis result and a disease probability. According to the method, based on deep learning, characteristics of CT images are automatically learned to distinguish COVIDI-19, influenza A and healthy people, so that the method is high in accuracy, the overall accuracy of current testing reaches 86.7%, the diagnosis speed is high, and only 30-60 S is needed for one set of CT according to different slice numbers; and a user can upload a CT image file and calculate and output a CT diagnosis result and the disease probability through the corona virus disease 2019 screening system, so that operation is convenient, the speed is high, and the detection rate of COVID-19 is greatly increased.

Description

technical field [0001] The invention belongs to the field of medical technology, and in particular relates to a deep learning-based screening method for COVID-19 and a screening system for COVID-19. Background technique [0002] In the early screening of the new type of coronavirus pneumonia (COVID-19), at present, it is mainly detected by means of kits, but its detection rate is low. In order to improve the detection rate of early screening of COVID-19, some other methods need to be adopted Auxiliary methods for judgment, such as CT image judgment methods, etc. [0003] There are some detection methods of lung lesion tissue in the prior art. For example, the patent document with publication number CN110599448A discloses a lung lesion tissue detection system based on MaskScoring R-CNN network migration learning, which inputs the lung CT to be detected For the image, the network outputs the recognized image, frames and masks the lesion tissue identified by the mask, and mark...

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): G16H50/20G06N3/04G06N3/08G16H50/80G06K9/62
CPCG16H50/20G16H50/80G06N3/08G06N3/047G06N3/045G06F18/24155
Inventor 吴炜李旭锟杜鹏
Owner ZHEJIANG UNIV
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
Eureka Blog
Learn More
PatSnap group products