Nodule recognition model training method, nodule recognition method and device

A technology for identifying models and training methods, applied in the field of image processing, can solve the problems of low nodule recognition accuracy and inability to adapt to nodules, and achieve the effects of avoiding difficult training, reducing human operation errors, and avoiding error interference

Active Publication Date: 2021-07-27
INFERVISION MEDICAL TECH CO LTD
View PDF15 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, due to the uncertainty of the size, shape and density of nodules, the currently trained deep neural network has low recognition accuracy for nodules and cannot adapt to different nodules.

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
  • Nodule recognition model training method, nodule recognition method and device
  • Nodule recognition model training method, nodule recognition method and device
  • Nodule recognition model training method, nodule recognition method and device

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0032] figure 1 It is a schematic flow chart of a nodule recognition model training method provided by Embodiment 1 of the present invention. This embodiment is applicable to the case of high-precision training of the nodule recognition model. The model training device can be implemented by means of software and / or hardware, and the device can be integrated into an electronic device such as a server or a computer. The method specifically includes the following steps:

[0033] S110. Acquire a sample image, wherein each pixel in the sample image is provided with an identifier, and the identifier includes a positive sample identifier, a negative sample identifier, and a non-sample identifier.

[0034] S120. Input the sample image into a nodule recognition model to be trained, and determine a loss function of each pixel based on a recognition result of the nodule recognition model to be trained.

[0035] S130. Determine a loss function corresponding to the sample image according...

Embodiment 2

[0062] Figure 4 It is a schematic flowchart of a nodule identification method provided by an embodiment of the present invention, and the method specifically includes:

[0063] S210. Acquire a lung image to be identified.

[0064] S220. Input the lung image into a pre-trained nodule recognition model to obtain a nodule recognition result of the lung image, wherein the nodule recognition model is based on the nodule recognition model provided in the above-mentioned embodiment The training method is trained to get.

[0065] In this embodiment, the lung image is input into a pre-trained nodule recognition model to obtain the nodule recognition result of the lung image, which includes: processing the three-dimensional lung image slice to be recognized into a plurality of binary two-dimensional lung image; for any two-dimensional lung image to be identified, the two-dimensional lung image to be identified, the first preset number of two-dimensional lung images before the two-dim...

Embodiment 3

[0069] Figure 5 It is a schematic structural diagram of a nodule recognition model training device provided in Embodiment 3 of the present invention, and the device includes:

[0070] A sample acquisition module 310, configured to acquire a sample image, wherein each pixel in the sample image is provided with an identification, and the identification includes a positive sample identification, a negative sample identification and a non-sample identification;

[0071] The first loss function determination module 320 is configured to input the sample image into the nodule recognition model to be trained, and determine the loss function of each pixel based on the recognition result of the nodule recognition model to be trained;

[0072] The second loss function determination module 330 is configured to determine a loss function corresponding to the sample image according to the identification of each pixel in the sample image and the loss function of each pixel;

[0073] The mod...

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 nodule recognition model training method, a nodule recognition method and a device. Wherein the nodule recognition model training method includes: obtaining a sample image, wherein each pixel in the sample image is provided with a mark, and the mark includes a positive sample mark, a negative sample mark and a non-sample mark; the sample image is input to the nodule recognition to be trained In the model, the loss function of each pixel is determined based on the recognition results of the nodule recognition model to be trained; the loss function corresponding to the sample image is determined according to the identification of each pixel in the sample image and the loss function of each pixel; based on the sample image The corresponding loss function is used to train the nodule recognition model to be trained to generate a nodule recognition model. It ensures the balance of positive and negative samples, avoids the problem of poor recognition accuracy, and ignores the supervision of non-sample identification during the training process, which can avoid the wrong interference to model training caused by the omission of nodule regions during the pixel labeling process.

Description

technical field [0001] The embodiments of the present invention relate to the technical field of image processing, and in particular, to a nodule recognition model training method, a nodule recognition method and a device. Background technique [0002] Pulmonary nodules are one of the early symptoms of lung cancer, and the characteristics of the lesion can be inferred from the lesion characteristics of the nodules. Due to the uncertainty of the size, shape and density of nodules, traditional medical detection methods are difficult to meet the detection accuracy of pulmonary nodules. [0003] In recent years, with the development of artificial intelligence and deep learning algorithms, the processing of medical images has also been involved. The abstract features in CT images are extracted through deep neural networks. Compared with manual feature extraction, artificial subjective factors are avoided in different environments. different degrees of impact. However, due to th...

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 & AuthorityPatents(China)
IPC IPC(8): G06T7/00G06N3/08G06N3/04
CPCG06T7/0012G06N3/08G06T2207/20084G06T2207/20081G06T2207/30064G06N3/045
Inventor刘恩佑张欢赵朝炜李新阳陈宽王少康
OwnerINFERVISION MEDICAL TECH CO LTD