Tuberculosis detection model construction method and applications

A detection model and construction method technology, applied in the field of medical imaging, can solve the time-consuming problems of tuberculosis and achieve the effects of reducing false positive rate, accurate prediction, and reducing missed and misdiagnosed rates

Active Publication Date: 2019-03-08
HUIYING MEDICAL TECH (BEIJING) CO LTD
View PDF5 Cites 9 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

These all pose great challenges for doctors, making the diagnosis of TB a difficult and time-consuming task

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
  • Tuberculosis detection model construction method and applications
  • Tuberculosis detection model construction method and applications
  • Tuberculosis detection model construction method and applications

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0055] Such as figure 1 As shown, the application provides a method for building a pulmonary tuberculosis detection model, including:

[0056] S101. Obtain a specified number of chest X-ray images marked with a label frame of a pulmonary tuberculosis lesion area.

[0057] Specifically, the chest X-ray images may use a large number of existing chest X-ray images of tuberculosis patients. For example, one existing source contains a dataset of 2443 chest X-ray images (in DICOM format). In this dataset, 1974 are randomly selected as the training set, and the rest are divided into a validation set and a test set. Wherein, the chest X-ray image has a pulmonary tuberculosis focus area marked by a doctor.

[0058] S102. Perform image preprocessing on the chest X-ray image to obtain preprocessed image data.

[0059] Specifically, use a clustering algorithm to generate standard values ​​of WW and WP from samples with window width WW and window position WP guidance values; Images ar...

Embodiment 2

[0092] The application also provides a tuberculosis detection method based on the above-mentioned tuberculosis detection model, comprising the steps of:

[0093] S601, acquiring a chest X-ray image of the patient;

[0094] S602. Perform preprocessing on the chest X-ray image to obtain preprocessed image data;

[0095] S603. Input the preprocessed image data into the trained tuberculosis detection model to detect tuberculosis.

[0096] S604, input the detection frame output by the tuberculosis detection model and its corresponding classification confidence score into the false positive reduction network;

[0097] S605. Obtain the detection frame and its corresponding classification confidence score output after being processed by the false positive reduction network;

[0098] S606. Output and display the detection frames whose classification confidence scores output after being processed by the false positive reduction network are higher than a specified threshold as valid ca...

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 a tuberculosis detection model construction method and applications. The method includes the following steps: A, obtaining chest X-ray images labeled with tuberculosis focus area labeling frames and with assigned quantities; B, performing image preprocessing on the chest X-ray images so that preprocessed image data can be obtained; and C, selecting a basic convolutional neural network model, and inputting the preprocessed image data into the basic convolutional neural network model to perform training so that a trained tuberculosis detection model can be obtained, wherein convolution kernels with different shapes can be adopted according to different anchor sizes in the basic convolutional neural network model. Thus, tuberculosis focus areas can be accurately predicted through the model, so that missed diagnosis and misdiagnosis rates can be reduced.

Description

technical field [0001] The invention relates to the field of medical imaging, in particular to a construction method and application of a pulmonary tuberculosis detection model. Background technique [0002] As an economical and convenient detection method, chest X-ray is the main means of tuberculosis screening. Doctors can use this X-ray to diagnose tuberculosis. The efficiency and accuracy of diagnosis depend on the doctor's experience level. However, due to the lack of a large number of professional medical imaging doctors, it is difficult to guarantee the efficiency and accuracy of diagnosis. [0003] And pulmonary tuberculosis has various X-ray manifestations. According to clinical research, multiple symptoms often appear together. All of these pose great challenges to doctors, making the diagnosis of TB a difficult and time-consuming task. [0004] Therefore, there is an urgent need for a pulmonary tuberculosis detection model that can more accurately predict the lo...

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): G16H50/50G06N3/04
CPCG16H50/50G06N3/045
Inventor 柴象飞郭娜黎安伟孟博文王成左盼莉
Owner HUIYING MEDICAL TECH (BEIJING) CO LTD
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