Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

Image data expansion method for deep learning model training and learning

An image data and model training technology, applied in image data processing, image analysis, character and pattern recognition, etc., can solve the problems of insufficient image data heterogeneity, insufficient data volume, unfavorable image analysis of rare cases, etc. The effect of overfitting, reducing the burden on doctors, improving accuracy and generalization ability

Pending Publication Date: 2019-05-17
NANJING UNIV OF INFORMATION SCI & TECH
View PDF1 Cites 21 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, the existing methods are obviously insufficient in solving the heterogeneity of image data, and cannot fundamentally solve the problem of insufficient data volume, which is not conducive to image analysis of some rare cases

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
  • Image data expansion method for deep learning model training and learning
  • Image data expansion method for deep learning model training and learning
  • Image data expansion method for deep learning model training and learning

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0030] Example 1: as Figure 1-6 As shown, this image image data expansion method for deep learning model training and learning includes the following steps:

[0031] (1) Determine the data type and identify CT or MRI image data;

[0032](2) For the image data, determine whether there is a defined ROI, and combine the size of the tumor area, select the corresponding method to complete the construction of the image data set;

[0033] (3) Use the basic image transformation method to train the image data set to obtain a preliminary training data set. Obtain a preliminary training data set, and the initially expanded data set has enabled the deep learning network to achieve a certain performance increase. At this time, for the data after the initial expansion, a further richer data expansion can be performed using a Generative Adversarial Network (GAN). After that, the data set can be sent to the deep learning network model for parameter training. In the parameter training pro...

Embodiment 2

[0045] Embodiment 2: as Figure 1-6 As shown, the video image data expansion method for deep learning model training and learning includes the following steps:

[0046] (1) Determine the data type and identify CT or MRI image data;

[0047] (2) For the image data, judge whether there is a defined ROI, and combine the size of the tumor area, select the corresponding method to complete the construction of the image data set; that is, use the MRI data expansion method based on ROI, first select the region of interest in the original MRI image , and then use conventional methods to sequentially perform pruning, translation transformation, rotation, noise removal, resampling, mirror flip, reflection transformation, PCA dithering, and color dithering on the image of the region of interest to obtain the MR image dataset.

[0048] Data augmentation based on ROI is to randomly select blocks in the image data set of a given Region of Interest (ROI) derived by existing algorithms, that ...

Embodiment 3

[0054] Embodiment 3: as Figure 1-6 As shown, the video image data expansion method for deep learning model training and learning includes the following steps:

[0055] (1) Determine the data type and identify CT or MRI image data;

[0056] (2) For the image data, judge whether there is a defined ROI, and combine the size of the tumor area, select the corresponding method to complete the construction of the image data set; that is, use the MRI data expansion method based on ROI, first select the region of interest in the original MRI image , and then use conventional methods to sequentially perform pruning, translation transformation, rotation, noise removal, resampling, mirror flip, reflection transformation, PCA dithering, and color dithering on the image of the region of interest to obtain the MR image dataset.

[0057] Data augmentation based on ROI is to randomly select blocks in the image data set of a given Region of Interest (ROI) derived by existing algorithms, that ...

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 relates to an image data expansion method for deep learning model training and learning, and belongs to the technical field of computer medical image calculation. The method comprises the following steps: firstly, judging a data type, and identifying CT or MRI image data; For the image data, judging whether an ROI (Region Of Interest) is defined or not, and selecting a correspondingmethod to complete the construction of an image data set in combination with the size of a tumor region; Training the image data set by adopting a basic image transformation method to obtain a preliminary training data set; And finally, carrying out data expansion on the preliminary training data set, carrying out deep training by adopting a network model, and finally, carrying out probability prediction. Based on artificial intelligence deep learning, a series of data expansion methods are applied to learning of deep model training in the field of medical image processing, the influence of medical image data heterogeneity on abnormal data is solved, computer-aided diagnosis is facilitated, and the diagnosis efficiency and accuracy are improved.

Description

technical field [0001] The invention relates to an image image data expansion method for deep learning model training and learning, and belongs to the technical field of computer medical image computing. Background technique [0002] In the era of big data, a high-performance deep model often requires a large amount of high-quality data, but it is not easy to obtain high-quality data, so this also means that a robust (Robust) model is not easy to obtain. Secondly, compared with natural images, the acquisition of medical image data is often more difficult. The root cause is that it is difficult to obtain case data. At the same time, the use of data also has certain ethical and privacy restrictions, which are difficult to solve from a practical point of view. For example, the National Institutes of Health (NIH) announced in September 2017 that the chest imaging (CT) image data set contains a total of 110,000 image data. On the issue of desensitizing information, the NIH team u...

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): G06T7/00G06K9/32G06K9/62
Inventor 徐军谢嘉伟蔡程飞
Owner NANJING UNIV OF INFORMATION SCI & TECH
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products

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

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com