Supercharge Your Innovation With Domain-Expert AI Agents!

Method for predicting glioma molecular subtypes and prognosis based on depth image features

A molecular subtype and depth image technology, applied in image analysis, image enhancement, image data processing, etc., can solve the problems of expensive, invasive, etc., and achieve the effect of improving accuracy

Pending Publication Date: 2021-01-12
TIANJIN UNIV
View PDF7 Cites 2 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Traditionally, for example, in the article "Emerging insights into the molecular and cellular basis of glioblastoma" by Dunn et al., identification of tumor subtypes has depended on brain biopsy followed by mRNA profiling, an expensive and invasive procedure

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
  • Method for predicting glioma molecular subtypes and prognosis based on depth image features
  • Method for predicting glioma molecular subtypes and prognosis based on depth image features
  • Method for predicting glioma molecular subtypes and prognosis based on depth image features

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0051] A method for predicting the molecular subtype and prognosis of glioma based on deep image features, comprising the following steps:

[0052] Step 1, constructing a data set: the data set includes magnetic resonance imaging (MRI) of multiple cases of primary glioma patients.

[0053] Step 2, build a feature extractor based on RA-UNet neural network:

[0054] 2-1. Use RA-UNet to segment the magnetic resonance imaging MRI of each patient with primary glioma in step 1 to generate segmentation results, and generate n pieces of magnetic resonance imaging MRI for each patient with primary glioma Split results.

[0055] 2-2. For the segmentation results obtained in step 2-1, use the feature extractor to extract high-dimensional features. The high-dimensional feature dimensions extracted from each segmentation result are (64, 64, 64, 32), integrated into each case The high-dimensional feature dimensions of MRI in primary glioma patients are (n, 64, 64, 64, 32);

[0056] Step ...

Embodiment 2

[0070] A method for predicting the molecular subtype and prognosis of glioma based on deep image features, comprising the following steps:

[0071] Step 1, data preprocessing

[0072] 1-1 Prepare the data set, which comes from the pretreatment MRI of primary glioma patients in The Cancer Genome Atlas. Each MRI includes four modes of T1, T1Gd, T2, and FLAIR, that is, four scan modes .

[0073] Molecular subtype information and survival data for glioma patients were obtained from cBioPortal, a public domain clinical and molecular data repository of The Cancer Genome Atlas patient cohorts.

[0074] In the embodiment of the present invention, the data set has a total of 71 cases, that is, the pretreatment MRIs of 71 patients with primary glioma. Among them, the four molecular subtypes of classical, neuronal, proneural and mesenchymal included 15, 9, 18 and 29 patients, respectively. We divided patient survival into short (short) - less than 6 months, medium (medium) - 6 to 18 m...

Embodiment 3

[0099] We sort the 2166 features obtained in step 3-3 in descending order according to their F-score values, and set the selected feature set as an empty set initially, and select a feature with the largest F-score value from the unselected features to add To the selected feature set, and then use the selected feature set to train SVM, each training will get an accuracy rate, until 2166 features are added to the selected set, and the iteration ends. Finally, experimental data show that for molecular subtype prediction problems such as figure 2 As shown, the number of features achieves the best effect when the number of features is 370, and the accuracy reaches 82%. In the survival time prediction problem, such as image 3 As shown, the number of features achieves the best effect when the number of features is 174, and the accuracy reaches 94.8%.

[0100] The training process of the SVM classifier adopts the method of 10-fold cross-validation. For simplicity and intuition, ...

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 provides a method for predicting glioma molecular subtypes and prognosis based on depth image features. The method comprises the following steps: step 1, constructing a data set, whereinthe data set comprises magnetic resonance imaging (MRI) of multiple primary glioma patients; step 2, constructing a feature extractor based on an RANet neural network; step 3, constructing a featureoptimizer; step 4, constructing a DeepRA model based on depth image features, including a primary glioma molecular subtype predictor and a patient survival time predictor; step 5, performing performance evaluation on the primary glioma molecular subtype predictor and the patient survival time predictor. step 6, predicting the magnetic resonance imaging (MRI) of the patient suffering from the primary glioma by using a molecular subtype predictor and a survival time predictor. According to the method, the primary glioma molecular subtype and the survival time of the patient are predicted by using the depth image features, and therefore the accuracy of predicting the primary glioma molecular subtype and the survival time of the patient is improved to a certain extent compared with the conventional method.

Description

technical field [0001] The invention relates to the technical field of deep learning and medical image processing, in particular to a method for predicting molecular subtypes and prognosis of glioma based on deep image features. Background technique [0002] Glioma is the malignant transformation of astrocytoma. It is the most malignant type of astrocytoma and the most common malignant intracranial tumor, accounting for 40 to 50 percent of all intracranial tumors. Glioma grows rapidly, is highly malignant, and generally progresses rapidly, with an average survival period of only about 12 months. Therefore, the research and prognosis of glioma are of great significance to people. [0003] Gliomas are classified into four molecular subtypes: classical, stromal, neuronal, and preneuronal, each of which can be treated differently. Traditionally, for example, in the article "Emerging insights into the molecular and cellular basis of glioblastoma" by Dunn et al., identification ...

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): G06T7/00G06T7/10G06K9/62G06N3/04
CPCG06T7/0012G06T7/10G06T2207/10088G06T2207/30096G06V2201/03G06N3/045G06F18/2135G06F18/2411
Inventor 苏苒刘晓莹
Owner TIANJIN UNIV
Features
  • R&D
  • Intellectual Property
  • Life Sciences
  • Materials
  • Tech Scout
Why Patsnap Eureka
  • Unparalleled Data Quality
  • Higher Quality Content
  • 60% Fewer Hallucinations
Social media
Patsnap Eureka Blog
Learn More

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

© 2025 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com