WO2022188490A1 - 一种基于影像基因组学的生存预测方法和系统 - Google Patents

一种基于影像基因组学的生存预测方法和系统 Download PDF

Info

Publication number
WO2022188490A1
WO2022188490A1 PCT/CN2021/137310 CN2021137310W WO2022188490A1 WO 2022188490 A1 WO2022188490 A1 WO 2022188490A1 CN 2021137310 W CN2021137310 W CN 2021137310W WO 2022188490 A1 WO2022188490 A1 WO 2022188490A1
Authority
WO
WIPO (PCT)
Prior art keywords
image
gene
modules
data
correlation
Prior art date
Application number
PCT/CN2021/137310
Other languages
English (en)
French (fr)
Inventor
张圣海
李志成
赵源深
孙秋畅
梁栋
Original Assignee
中国科学院深圳先进技术研究院
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 中国科学院深圳先进技术研究院 filed Critical 中国科学院深圳先进技术研究院
Publication of WO2022188490A1 publication Critical patent/WO2022188490A1/zh

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/0002—Inspection of images, e.g. flaw detection
    • G06T7/0012—Biomedical image inspection
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/23—Clustering techniques
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/11—Region-based segmentation
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16B—BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY
    • G16B20/00—ICT specially adapted for functional genomics or proteomics, e.g. genotype-phenotype associations
    • G16B20/20—Allele or variant detection, e.g. single nucleotide polymorphism [SNP] detection
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16B—BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY
    • G16B40/00—ICT specially adapted for biostatistics; ICT specially adapted for bioinformatics-related machine learning or data mining, e.g. knowledge discovery or pattern finding
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20081—Training; Learning
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20084—Artificial neural networks [ANN]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30004—Biomedical image processing
    • G06T2207/30096—Tumor; Lesion

Definitions

  • the present invention relates to the technical field of medical image processing, and more particularly, to a survival prediction method and system based on image genomics.
  • Cancer has a high morbidity and mortality, and has become the main cause of human death due to disease.
  • Survival prediction analysis of tumor patients can provide clinicians with necessary prognostic information, help clinicians establish a clear treatment plan, improve the cure rate of tumors, effectively reduce the burden on patients, and improve the quality of life of patients with prognosis.
  • the TNM staging system launched by the American Cancer Society has been widely used in tumor clinical practice and is an important tool to guide prognosis. However, many studies have found that TNM cannot effectively discriminate differences in survival outcomes among patients with different tumor characteristics.
  • Radiogenomics is primarily used to study potential links between medical imaging phenotypes and tumor genomes.
  • prognosis prediction the use of deep learning methods, combined with imaging markers and tumor genomes, can predict disease status and prognosis, and then non-invasively evaluate the biological behavior of tumors, which plays an important role in individualized tumor therapy.
  • Several studies have explored to correlate gene expression data with molecular information and clinical practice, but radiogenomics has rarely been applied to tumor survival prediction.
  • the purpose of the present invention is to overcome the above-mentioned defects of the prior art, provide a survival prediction method and system based on imaging genomics, and solve the problems of lack of biological interpretability and generalization ability of the imaging-based deep learning survival prediction method .
  • a survival prediction method based on radiogenomics includes the following steps:
  • Step S1 acquiring image data of multiple tumor patients, survival data and RNA data of each patient, and establishing a data set;
  • Step S2 segment the tumor area of each patient from the image data, and the tumor area includes the real tumor area and the suspected tumor area;
  • Step S3 input the image data of each patient into the neural network to extract image features and perform clustering to obtain multiple image modules;
  • Step S4 utilize RNA data to obtain the gene module of each patient
  • Step S5 screening is performed according to the correlation between the gene module and the image module, and a plurality of strongly correlated gene modules and image modules that satisfy the first correlation condition are selected;
  • Step S6 performing pathway enrichment on the genes in the screened gene modules to obtain gene pathways related to the image module;
  • Step S7 calculating the gene set variation analysis score of the gene pathway, retaining the strongly correlated gene pathway whose correlation with the image module satisfies the second correlation condition, and the gene pathway is used to interpret the image module at the biological level;
  • Step S8 using the final retained image features to perform survival prediction.
  • a survival prediction system based on radiogenomics includes:
  • Data acquisition unit used to acquire image data of multiple tumor patients, survival data and RNA data of each patient, and establish a data set;
  • Tumor area extraction unit used to segment the tumor area of each patient from the image data, the tumor area includes the real tumor area and the suspected tumor area;
  • Image module extraction unit used to input the image data of each patient into the neural network to extract image features and perform clustering to obtain multiple image modules;
  • Gene module extraction unit used to obtain gene modules of each patient using RNA data
  • the first screening unit used for screening according to the correlation between the gene module and the image module, and selecting a plurality of strongly correlated gene modules and image modules that satisfy the first correlation condition;
  • Gene pathway analysis unit used for pathway enrichment of genes in the screened gene modules to obtain gene pathways related to the image module;
  • the second screening unit used to calculate the gene set variation analysis score of the gene pathway, and retain the strongly correlated gene pathway whose correlation with the image module satisfies the second correlation condition.
  • the gene pathway is used to analyze the biological level of the image module explain;
  • Survival prediction unit used for survival prediction using the final preserved image features.
  • the present invention has the advantages that: in the prior art, the survival prediction method based on deep learning images does not take into account the biological interpretability, in addition, due to insufficient sample data, training sample distribution and real samples Inconsistent distribution and other problems greatly limit the confidence of doctors in the survival prediction method based on deep learning imaging.
  • the survival prediction method based on imaging genomics proposed in the present invention can improve the biological interpretability of the model, and at the same time improve the Generalization ability of deep learning methods.
  • FIG. 1 is a flowchart of a survival prediction method based on videogenomics according to an embodiment of the present invention.
  • the survival prediction method based on video genomics includes the following steps.
  • Step S110 acquiring image data of multiple tumor patients, survival data and RNA data of each patient, and establishing a data set.
  • the patients in the established data set suffer from the same tumor disease, such as glioma.
  • Image data are images of patients before treatment, including but not limited to MRI, CI, and PET images. Survival was the time period between when the imaging data was generated and the patient died.
  • the patient's image data comes from the patient image data set jointly included in TCIA (The Cancer Imaging Archive) and TCGA (The Cancer Genome Atlas).
  • the present invention does not limit the number of image data of a patient. In general, the more the amount of data, the more accurate the correlation.
  • RNA (ribonucleic acid) data include, for example, nucleotide sequences, single nucleotide polymorphisms, structures, properties, and related descriptions.
  • Step S120 performing image normalization on the image data in the data set.
  • image normalization data of different magnitudes can be transformed into a unified measure for subsequent analysis.
  • methods of image data normalization include, but are not limited to, Z-Score normalization, maximum and minimum normalization, and decimal scaling methods.
  • step S130 the tumor region of each patient is segmented from the image data of each patient.
  • the tumor area includes the suspected tumor area in addition to the real tumor area.
  • step S140 a deep learning neural network is established to extract image features and acquire image modules.
  • Various types of neural networks may be employed, including but not limited to ResNet (residual network) and its derivatives, VGG (computer vision group) and its derivatives, and the like.
  • ResNet residual network
  • VGG computer vision group
  • the input of the last fully connected layer of the deep learning neural network is regarded as the image features extracted by the network from the image data, and these image features are extracted and clustered to obtain several image modules.
  • Clustering methods include, but are not limited to, K-Means clustering, mean-shift clustering and other methods.
  • Step S150 using RNA data to obtain a gene module.
  • Methods for obtaining gene modules include, but are not limited to, using WGCNA analysis (Weighted correlation network analysis, weighted gene co-expression network analysis).
  • WGCNA Weighted gene co-expression network analysis
  • WGCNA Weighted correlation network analysis
  • step S160 the correlation between the gene module and the image module is calculated, and the image module and the gene module are screened.
  • dimensionality reduction processing is performed on the image modules to obtain the feature value of each image module.
  • the dimensionality reduction methods used include but are not limited to PCA (Principal Component Analysis, principal component analysis), LDA (Linear Discriminant Analysis, Linear Discriminant Analysis) and so on.
  • the correlation between the gene module and the image module is calculated through the eigenvalues obtained by dimensionality reduction, and the function for calculating the correlation includes, but is not limited to, the Pearson correlation coefficient, the Spearman correlation coefficient, and the like.
  • step S170 pathway enrichment is performed on the genes in the screened gene modules to obtain gene pathways related to the image module.
  • Pathway enrichment methods include but are not limited to using Metascape website, Cytosacpe software, ClusterProfiler R package, etc.
  • the pathways obtained by pathway enrichment can be used to characterize image modules, that is, to interpret the image modules at the biological level. For example, an image module is related to a gene module, and the gene module is related to metabolism, then it can be considered that the biological significance of the image module is strongly related to metabolism.
  • step S180 the GSVA score of the gene pathway is calculated, the gene pathway related to the image module is further screened, and the image module is explained at the level of biological significance.
  • GSVA Gene Set Variation Analysis converts the gene expression matrix into a pathway enrichment score matrix, and the score in the pathway enrichment score matrix represents the activation degree of a pathway in a case. Select the GSVA scores of each pathway in the pathway enrichment in step S170, and use these GSVA scores to perform correlation analysis with the image modules obtained after screening in step S160.
  • the function for calculating correlation includes but is not limited to Pearson correlation coefficient, Spearman correlation coefficient, etc.
  • an appropriate correlation threshold may be set according to the requirements of data volume, execution speed and prediction accuracy, and the correlation threshold in step S160 does not have to be the same.
  • Step S190 using the finally retained image features to perform survival prediction.
  • the image features in the final image module are the required biologically meaningful image features.
  • These biologically meaningful images Features can make neural networks more interpretable and generalizable. After finding image modules that are strongly correlated with genes, the neural network is retrained. In practical applications, various training schemes can be used.
  • the feature values of the image module are used to replace the original image features for retraining. Specifically, keep the weight of the network before the fully connected layer unchanged, perform dimensionality reduction processing on the filtered image modules to obtain the eigenvalues of each image module, and then use the eigenvalues of the image module to replace the original image features. The weights of the fully connected layers are retrained.
  • the dimensionality reduction methods used here include but are not limited to PCA (Principal Component Analysis, principal component analysis), LDA (Linear Discriminant Analysis, Linear Discriminant Analysis) and so on. The number of feature values used by this method is significantly lower than that of the original image features, thereby enhancing the generalization ability.
  • the neural network model obtained through the above steps of training and retraining can be used to make predictions for individual cases. For example, for the input case image data, the network finally outputs a risk value, which can be used as a predictor. Further, combining the survival data of each case, construct a regression model (such as using Lasso Cox regression to build predictive models) to predict patient survival.
  • a regression model such as using Lasso Cox regression to build predictive models
  • the present invention also provides a survival prediction system based on video genomics, which is used to realize one or more aspects of the above method.
  • the system includes: a data acquisition unit for acquiring image data of multiple tumor patients, survival data and RNA data of each patient, and establishing a data set; a tumor region extraction unit for segmenting image data from image data The tumor area of each patient, the tumor area includes the real tumor area and the suspected tumor area; the image module extraction unit is used to input the image data of each patient into the neural network to extract image features and perform clustering to obtain multiple image modules Gene module extraction unit, which is used to obtain the gene module of each patient by using RNA data;
  • the first screening unit is used for screening according to the correlation between the gene module and the image module, and selects the ones that satisfy the first correlation condition.
  • a plurality of strongly correlated gene modules and image modules A plurality of strongly correlated gene modules and image modules; a gene pathway analysis unit, which is used for pathway enrichment of genes in the screened gene modules to obtain gene pathways related to the image modules; a second screening unit, which uses To calculate the gene set variation analysis score of the gene pathway, retain the strongly correlated gene pathway whose correlation with the image module satisfies the second correlation condition, and this gene pathway is used to interpret the image module at the biological level; survival prediction unit , which is used for survival prediction using the final preserved image features.
  • the technical solution based on image genomics proposed by the present invention combines deep learning technology and bioinformatics technology for the first time, and solves the lack of biological interpretability and generalization of the current deep learning survival prediction method based on image. question of ability. It can be applied to tumor survival prediction, which is helpful for doctors to better understand the strengths and weaknesses of survival prediction, and to clarify the knowledge boundary of survival prediction methods, understand under what circumstances they are effective, so as to appropriately trust and use survival prediction methods , for clinical guidance.
  • the present invention may be a system, method and/or computer program product.
  • the computer program product may include a computer-readable storage medium having computer-readable program instructions loaded thereon for causing a processor to implement various aspects of the present invention.
  • a computer-readable storage medium may be a tangible device that can hold and store instructions for use by the instruction execution device.
  • the computer-readable storage medium may be, for example, but not limited to, an electrical storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination of the foregoing.
  • Non-exhaustive list of computer readable storage media include: portable computer disks, hard disks, random access memory (RAM), read only memory (ROM), erasable programmable read only memory (EPROM) or flash memory), static random access memory (SRAM), portable compact disk read only memory (CD-ROM), digital versatile disk (DVD), memory sticks, floppy disks, mechanically coded devices, such as printers with instructions stored thereon Hole cards or raised structures in grooves, and any suitable combination of the above.
  • RAM random access memory
  • ROM read only memory
  • EPROM erasable programmable read only memory
  • flash memory static random access memory
  • SRAM static random access memory
  • CD-ROM compact disk read only memory
  • DVD digital versatile disk
  • memory sticks floppy disks
  • mechanically coded devices such as printers with instructions stored thereon Hole cards or raised structures in grooves, and any suitable combination of the above.
  • Computer-readable storage media are not to be construed as transient signals per se, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through waveguides or other transmission media (eg, light pulses through fiber optic cables), or through electrical wires transmitted electrical signals.
  • the computer readable program instructions described herein may be downloaded to various computing/processing devices from a computer readable storage medium, or to an external computer or external storage device over a network such as the Internet, a local area network, a wide area network, and/or a wireless network.
  • the network may include copper transmission cables, fiber optic transmission, wireless transmission, routers, firewalls, switches, gateway computers, and/or edge servers.
  • a network adapter card or network interface in each computing/processing device receives computer-readable program instructions from a network and forwards the computer-readable program instructions for storage in a computer-readable storage medium in each computing/processing device .
  • Computer program instructions for carrying out operations of the present invention may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, state setting data, or instructions in one or more programming languages.
  • Source or object code written in any combination including object-oriented programming languages, such as Smalltalk, C++, Python, etc., and conventional procedural programming languages, such as the "C" language or similar programming languages.
  • the computer readable program instructions may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer, or entirely on the remote computer or server implement.
  • the remote computer may be connected to the user's computer through any kind of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (eg, using an Internet service provider through the Internet connect).
  • LAN local area network
  • WAN wide area network
  • custom electronic circuits such as programmable logic circuits, field programmable gate arrays (FPGAs), or programmable logic arrays (PLAs)
  • FPGAs field programmable gate arrays
  • PDAs programmable logic arrays
  • Computer readable program instructions are executed to implement various aspects of the present invention.
  • These computer readable program instructions may be provided to a processor of a general purpose computer, special purpose computer or other programmable data processing apparatus to produce a machine that causes the instructions when executed by the processor of the computer or other programmable data processing apparatus , resulting in means for implementing the functions/acts specified in one or more blocks of the flowchart and/or block diagrams.
  • These computer readable program instructions can also be stored in a computer readable storage medium, these instructions cause a computer, programmable data processing apparatus and/or other equipment to operate in a specific manner, so that the computer readable medium on which the instructions are stored includes An article of manufacture comprising instructions for implementing various aspects of the functions/acts specified in one or more blocks of the flowchart and/or block diagrams.
  • Computer readable program instructions can also be loaded onto a computer, other programmable data processing apparatus, or other equipment to cause a series of operational steps to be performed on the computer, other programmable data processing apparatus, or other equipment to produce a computer-implemented process , thereby causing instructions executing on a computer, other programmable data processing apparatus, or other device to implement the functions/acts specified in one or more blocks of the flowcharts and/or block diagrams.
  • each block in the flowchart or block diagrams may represent a module, segment, or portion of instructions, which comprises one or more functions for implementing the specified logical function(s) executable instructions.
  • the functions noted in the blocks may occur out of the order noted in the figures. For example, two blocks in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved.
  • each block of the block diagrams and/or flowchart illustrations, and combinations of blocks in the block diagrams and/or flowchart illustrations can be implemented in dedicated hardware-based systems that perform the specified functions or actions , or can be implemented in a combination of dedicated hardware and computer instructions. It is well known to those skilled in the art that implementation in hardware, implementation in software, and implementation in a combination of software and hardware are all equivalent.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • General Physics & Mathematics (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Computation (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • General Health & Medical Sciences (AREA)
  • General Engineering & Computer Science (AREA)
  • Evolutionary Biology (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Medical Informatics (AREA)
  • Biophysics (AREA)
  • Molecular Biology (AREA)
  • Software Systems (AREA)
  • Biomedical Technology (AREA)
  • Computing Systems (AREA)
  • Mathematical Physics (AREA)
  • Spectroscopy & Molecular Physics (AREA)
  • Biotechnology (AREA)
  • Computational Linguistics (AREA)
  • Public Health (AREA)
  • Epidemiology (AREA)
  • Databases & Information Systems (AREA)
  • Bioethics (AREA)
  • Chemical & Material Sciences (AREA)
  • Analytical Chemistry (AREA)
  • Genetics & Genomics (AREA)
  • Proteomics, Peptides & Aminoacids (AREA)
  • Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
  • Radiology & Medical Imaging (AREA)
  • Quality & Reliability (AREA)
  • Image Analysis (AREA)

Abstract

本发明公开了一种基于影像基因组学的生存预测方法和系统。该方法包括：获取肿瘤患者的影像数据及各患者的生存期数据和RNA数据，建立数据集合；从影像数据中分割出各患者的肿瘤区域；将各患者的影像数据输入神经网络，以提取影像特征并进行聚类，得到多个影像模块；利用RNA数据获取各患者的基因模块；根据基因模块与影像模块的相关性进行筛选，选出多个强相关的基因模块和影像模块；对所选出的基因模块中的基因进行通路富集，得到与影像模块相关的基因通路；计算基因通路的基因集变异分析分数，保留与影像模块具有强相关的基因通路；使用保留的影像特征进行生存预测。本发明可提高生存预测方面的生物学可解释性、并同时提高深度学习的泛化能力。

Description

一种基于影像基因组学的生存预测方法和系统 技术领域
本发明涉及医学图像处理技术领域，更具体地，涉及一种基于影像基因组学的生存预测方法和系统。
 背景技术
肿瘤的发病率和死亡率很高，已经成为人类因疾病死亡的主要原因。对于肿瘤病人的生存预测分析可以给临床医生提供必要的预后信息，帮助临床医生建立明确的治疗方案，提高肿瘤的治愈率，同时有效降低患者的负担，改善患者预后的生活质量。美国癌症联合会推出的TNM分期系统在肿瘤临床实践中得到了广泛应用，是指导预后的重要工具。但是，许多研究发现TNM不能对不同肿瘤特征的患者的生存结局差异进行有效的区分。
影像基因组学主要用于研究医学成像表型和肿瘤基因组之间的潜在联系。在预后预测方面，利用深度学习的方法，通过影像学标记结合肿瘤基因组，可以预测疾病状态和预后情况，进而无创地评估肿瘤的生物学行为，对肿瘤个体化治疗起到重要作用。目前已有一些研究探索将基因表达数据与分子信息和临床关联起来，但影像基因组学还很少被应用到肿瘤生存预测方面。
现在的单纯基于深度学习影像的生存预测方法虽然已取得巨大进展，但在临床实践中仍然面临一些亟待解决的难题。例如，深度学习方法可以自动提取抽象的影像特征，但是其预测过程是端到端的，只有直接结果，无法提供诊断依据和病因病理，不能被医生完全信任和接受。以脑胶质瘤生存预测为例，医生可通过各种检测方式，结合患者的临床症状来进行生存预测；然而深度学习方法是通过人工神经网络学习大量带有标记的训练数据进行提取特征，得到的模型在临床上很难解释输入和输出间的因果关系。而可解释性对于医学领域非常重要，可解释性有利于医生更好地理解生存预测方法的强项与不足，并明确生存预测方法的知识边界，了解其在何种情况下有效，从而恰当地信任和使用生存预测方法来进行预测。因此，在现有技术中，深度学习方法缺乏可解释性的问题导致了医生对其得出结果的置信度较低，难以支持生存预测研究中的因果推理。而且，在医学领域，数据驱动的深度学习算法由于样本数据不足、训练样本分布与真实样本分布不一致等问题，均可能导致算法的性能下降严重，而现有的深度学习方法缺乏可解释性又会进一步导致泛化能力变差。因此，目前基于深度学习影像的生存预测方法泛化能力受到很大的质疑和挑战。
 技术问题
本发明的目的是克服上述现有技术的缺陷，提供一种基于影像基因组学的生存预测方法和系统，解决了基于影像的深度学习生存预测方法缺乏生物学可解释性以及缺乏泛化能力的问题。
 技术解决方案
根据本发明的第一方面，提供了一种基于影像基因组学的生存预测方法。该方法包括以下步骤：
步骤S1，获取多个肿瘤患者的影像数据及各患者的生存期数据和RNA数据，建立数据集合；
步骤S2，从影像数据中分割出各患者的肿瘤区域，该肿瘤区域包括真实肿瘤区域和疑似肿瘤区域；
步骤S3，将各患者的影像数据输入神经网络，以提取影像特征并进行聚类，得到多个影像模块；
步骤S4，利用RNA数据获取各患者的基因模块；
步骤S5，根据基因模块与影像模块之间的相关性进行筛选，选出满足第一相关性条件的多个强相关的基因模块和影像模块；
步骤S6，对所筛选出的基因模块中的基因进行通路富集，得到与影像模块相关的基因通路；
步骤S7，计算基因通路的基因集变异分析分数，保留与影像模块的相关性满足第二相关性条件的强相关的基因通路，该基因通路用于对影像模块在生物学层面上进行解释；
步骤S8，使用最终保留的影像特征进行生存预测。
根据本发明的第二方面，提供一种基于影像基因组学的生存预测系统。该系统包括：
数据采集单元：用于获取多个肿瘤患者的影像数据及各患者的生存期数据和RNA数据，建立数据集合；
肿瘤区域提取单元：用于从影像数据中分割出各患者的肿瘤区域，该肿瘤区域包括真实肿瘤区域和疑似肿瘤区域；
影像模块提取单元：用于将各患者的影像数据输入神经网络，以提取影像特征并进行聚类，得到多个影像模块；
基因模块提取单元：用于利用RNA数据获取各患者的基因模块；
第一筛选单元：用于根据基因模块与影像模块之间的相关性进行筛选，选出满足第一相关性条件的多个强相关的基因模块和影像模块；
基因通路分析单元：用于对所筛选出的基因模块中的基因进行通路富集，得到与影像模块相关的基因通路；
第二筛选单元：用于计算基因通路的基因集变异分析分数，保留与影像模块的相关性满足第二相关性条件的强相关的基因通路，该基因通路用于对影像模块在生物学层面上进行解释；
生存预测单元：用于使用最终保留的影像特征进行生存预测。
 有益效果
与现有技术相比，本发明的优点在于：在现有技术中，基于深度学习影像的生存预测方法没有考虑到生物学上的可解释性，另外由于样本数据不足、训练样本分布与真实样本分布不一致等问题，极大地限制了医生对于基于深度学习影像的生存预测方法的置信度，本发明所提出的基于影像基因组学的生存期预测方法可以提高模型的生物学可解释性、并同时提高深度学习方法的泛化能力。
通过以下参照附图对本发明的示例性实施例的详细描述，本发明的其它特征及其优点将会变得清楚。
 附图说明
被结合在说明书中并构成说明书的一部分的附图示出了本发明的实施例，并且连同其说明一起用于解释本发明的原理。
图1是根据本发明一个实施例的基于影像基因组学的生存预测方法的流程图。
 本发明的实施方式
现在将参照附图来详细描述本发明的各种示例性实施例。应注意到：除非另外具体说明，否则在这些实施例中阐述的部件和步骤的相对布置、数字表达式和数值不限制本发明的范围。
以下对至少一个示例性实施例的描述实际上仅仅是说明性的，决不作为对本发明及其应用或使用的任何限制。
对于相关领域普通技术人员已知的技术、方法和设备可能不作详细讨论，但在适当情况下，所述技术、方法和设备应当被视为说明书的一部分。
在这里示出和讨论的所有例子中，任何具体值应被解释为仅仅是示例性的，而不是作为限制。因此，示例性实施例的其它例子可以具有不同的值。
应注意到：相似的标号和字母在下面的附图中表示类似项，因此，一旦某一项在一个附图中被定义，则在随后的附图中不需要对其进行进一步讨论。
参见图1所示，本发明提供的基于影像基因组学的生存预测方法包括以下步骤。
步骤S110，获取多个肿瘤患者的影像数据及各患者的生存期数据和RNA数据，建立数据集合。
具体地，多个患者的影像数据为患者的影像集合，例如表示为V={v i,i=1,...,N}，N为患者的数量，每个个体数据v i表示为一个影像样本。
在一个实施例中，所建立的数据集合中的患者患有同一种肿瘤疾病，如脑胶质瘤。影像数据为患者进行治疗前的影像，包括但不限于MRI、CI及PET等影像。生存期为影像数据产生时至患者死亡之间的时间段。患者的影像数据例如来源于TCIA（The Cancer Imaging Archive，癌症影像存档）和TCGA（The Cancer Genome Atlas，癌症基因图谱）中共同收录的患者影像数据集。本发明对患者的影像数据个数不做限定，一般情况下，数据量越多，关联关系越准确。
RNA（核糖核酸）数据例如包括的核苷酸序列，单核苷酸多态性、结构、性质以及相关描述等。
步骤S120，对数据集合中的图像数据进行图像标准化。
通过图像标准化，可以将不同量级的数据转化为统一量度，以便于后续分析。例如图像数据标准化的方法包括但不限于Z-Score标准化、最大最小标准化及小数定标法等方法。
步骤S130，从各患者的影像数据中分割出各患者的肿瘤区域。
例如，为了提高生存预测的准确性，肿瘤区域除了包括真实肿瘤区域外，还包括疑似肿瘤区域。
步骤S140，建立一个深度学习神经网络，提取影像特征并获取影像模块。
可采用多种类型的神经网络，包括但不限于ResNet（残差网络）及其衍生网络、VGG（计算机视觉组）及其衍生网络等。将该深度学习神经网络的最后一个全连接层的输入视为网络从影像数据中提取到的影像特征，提取这些影像特征并进行聚类，得到若干个影像模块。聚类的方法包括但不限于K-Means聚类、均值漂移聚类等方法。
步骤S150，利用RNA数据获取基因模块。
获取基因模块的方法包括但不限于使用WGCNA分析（Weighted correlation network analysis，加权基因共表达网络分析）。加权基因共表达网络分析（WGCNA，Weighted correlation network analysis）是一种系统生物学方法，用来描述基因间的相关模式。
步骤S160，计算基因模块与影像模块之间的相关性，筛选影像模块和基因模块。
具体地，首先，对影像模块做降维处理，得到每个影像模块的特征值。
使用的降维方法包括但不限于PCA（Principal Component Analysis，主成分分析）、LDA（Linear Discriminant Analysis，线性判别分析）等。
然后，通过降维得到的特征值计算基因模块与影像模块之间的相关性，用于计算相关性的函数包括但不限于皮尔逊相关系数、斯皮尔曼相关系数等。
最后，根据基因模块和影像模块之间的相关性，可以筛选出与若干个强相关的基因模块和影像模块，对于相关性不大的基因模块和影像模块不保留。例如，使用相关性p=0.05为阈值，仅保留正相关p<0.05的或负相关<0.05的基因模块和影像模块。
步骤S170，对筛选出的基因模块中的基因进行通路富集，得到与影像模块相关的基因通路。
通路富集的方法包括但不限于使用Metascape网站、Cytosacpe软件、ClusterProfiler R包等，通路富集得到的通路可以用来表征影像模块，即对影像模块在生物学层面上进行解释。例如某影像模块与某基因模块相关，且基因模块与代谢有关，那么可以认为影像模块在生物学上的意义与代谢有强相关的联系。
步骤S180，计算基因通路的GSVA分数，进一步筛选与影像模块相关的基因通路，在生物学意义层面对影像模块进行解释。
GSVA（Gene Set Variation Analysis，基因集变异分析）将基因表达矩阵转换成通路富集分数矩阵，通路富集分数矩阵中的分数代表了某通路在某病例上的激活程度。挑选出步骤S170中通路富集中各通路的GSVA分数，使用这些GSVA分数再和步骤S160筛选后得到的影像模块做相关性分析，计算相关性的函数包括但不限于皮尔逊相关系数、斯皮尔曼相关系数等。
根据基因通路和影像模块之间的相关性，可以筛选出与若干个强相关的基因通路和影像模块，对于相关性不大的基因通路和影像模块不保留。例如使用相关性p=0.05为阈值，仅保留正相关p<0.05的或负相关<0.05的基因通路和影像模块。使用最终保留的通路对影像模块进行解释，这样影像模块就拥有了生物学上的可解释性。
需说明的是，在实际应用中，可根据数据量、执行速度和预测精确度等方面的要求设置适当的相关性阈值，并且与步骤S160中的相关性阈值不必须相同。
步骤S190，使用最终保留的影像特征进行生存预测。
在删去不相关的影像模块之后，相当于在生物学层面对影像特征进行了特征选择，最终影像模块中的影像特征就是所需要的具有生物学意义的影像特征，这些具有生物学意义的影像特征能够使神经网络的可解释性和泛化能力都变得更强。在找到与基因强相关的影像模块之后，对神经网络进行重新训练，在实际应用中，可以采用多种训练方案。
例如，在重训练时，保持全连接层之前的网络的权重不变，在全连接层使用筛选后的影像模块中的影像特征代替原有的影像特征，然后对全连接层的权重进行重新训练。
又如，使用影像模块的特征值代替原本的影像特征进行重训练。具体地，保持全连接层之前的网络的权重不变，对筛选后的影像模块做降维处理，得到每个影像模块的特征值，然后使用影像模块的特征值代替原有的影像特征，对全连接层的权重进行重新训练。这里使用的降维方法包括但不限于PCA（Principal Component Analysis，主成分分析）、LDA（Linear Discriminant Analysis，线性判别分析）等。这种方法使用到的特征值数量比原有的影像特征数量显著降低，从而增强了泛化能力。
经过上述步骤训练和重训练获得的神经网络模型可用于针对个体病例进行预测。例如，对于输入的病例图像数据，网络最终输出得到一个风险值，这个风险值可以当作是一个预测因子。进一步地，结合每个病例的生存数据，构建回归模型（如运用Lasso Cox回归建立预测模型）来对患者进行生存预测。
相应地，本发明还提供一种基于影像基因组学的生存预测系统，用于实现上述方法的一个方面或多个方面。例如，该系统包括：数据采集单元，其用于获取多个肿瘤患者的影像数据及各患者的生存期数据和RNA数据，建立数据集合；肿瘤区域提取单元，其用于从影像数据中分割出各患者的肿瘤区域，该肿瘤区域包括真实肿瘤区域和疑似肿瘤区域；影像模块提取单元，其用于将各患者的影像数据输入神经网络，以提取影像特征并进行聚类，得到多个影像模块；基因模块提取单元，其用于利用RNA数据获取各患者的基因模块；第一筛选单元，其用于根据基因模块与影像模块之间的相关性进行筛选，选出满足第一相关性条件的多个强相关的基因模块和影像模块；基因通路分析单元，其用于对所筛选出的基因模块中的基因进行通路富集，得到与影像模块相关的基因通路；第二筛选单元，其用于计算基因通路的基因集变异分析分数，保留与影像模块的相关性满足第二相关性条件的强相关的基因通路，该基因通路用于对影像模块在生物学层面上进行解释；生存预测单元，其用于使用最终保留的影像特征进行生存预测。
综上所述，本发明所提出的基于影像基因组学的技术方案，首次结合了深度学习技术和生物信息技术，解决了目前基于影像的深度学习生存预测方法缺乏生物学可解释性以及缺乏泛化能力的问题。可应用于肿瘤的生存预测，有利于医生更好地理解生存预测方面的强项与不足，并明确生存预测方法的知识边界，了解其在何种情况下有效，从而恰当地信任和使用生存预测方法，用于临床指导。
本发明可以是系统、方法和/或计算机程序产品。计算机程序产品可以包括计算机可读存储介质，其上载有用于使处理器实现本发明的各个方面的计算机可读程序指令。
计算机可读存储介质可以是可以保持和存储由指令执行设备使用的指令的有形设备。计算机可读存储介质例如可以是――但不限于――电存储设备、磁存储设备、光存储设备、电磁存储设备、半导体存储设备或者上述的任意合适的组合。计算机可读存储介质的更具体的例子（非穷举的列表）包括：便携式计算机盘、硬盘、随机存取存储器（RAM）、只读存储器（ROM）、可擦式可编程只读存储器（EPROM或闪存）、静态随机存取存储器（SRAM）、便携式压缩盘只读存储器（CD-ROM）、数字多功能盘（DVD）、记忆棒、软盘、机械编码设备、例如其上存储有指令的打孔卡或凹槽内凸起结构、以及上述的任意合适的组合。这里所使用的计算机可读存储介质不被解释为瞬时信号本身，诸如无线电波或者其他自由传播的电磁波、通过波导或其他传输媒介传播的电磁波（例如，通过光纤电缆的光脉冲）、或者通过电线传输的电信号。
这里所描述的计算机可读程序指令可以从计算机可读存储介质下载到各个计算/处理设备，或者通过网络、例如因特网、局域网、广域网和/或无线网下载到外部计算机或外部存储设备。网络可以包括铜传输电缆、光纤传输、无线传输、路由器、防火墙、交换机、网关计算机和/或边缘服务器。每个计算/处理设备中的网络适配卡或者网络接口从网络接收计算机可读程序指令，并转发该计算机可读程序指令，以供存储在各个计算/处理设备中的计算机可读存储介质中。
用于执行本发明操作的计算机程序指令可以是汇编指令、指令集架构（ISA）指令、机器指令、机器相关指令、微代码、固件指令、状态设置数据、或者以一种或多种编程语言的任意组合编写的源代码或目标代码，所述编程语言包括面向对象的编程语言—诸如Smalltalk、C++、Python等，以及常规的过程式编程语言—诸如“C”语言或类似的编程语言。计算机可读程序指令可以完全地在用户计算机上执行、部分地在用户计算机上执行、作为一个独立的软件包执行、部分在用户计算机上部分在远程计算机上执行、或者完全在远程计算机或服务器上执行。在涉及远程计算机的情形中，远程计算机可以通过任意种类的网络—包括局域网(LAN)或广域网(WAN)—连接到用户计算机，或者，可以连接到外部计算机（例如利用因特网服务提供商来通过因特网连接）。在一些实施例中，通过利用计算机可读程序指令的状态信息来个性化定制电子电路，例如可编程逻辑电路、现场可编程门阵列（FPGA）或可编程逻辑阵列（PLA），该电子电路可以执行计算机可读程序指令，从而实现本发明的各个方面。
这里参照根据本发明实施例的方法、装置（系统）和计算机程序产品的流程图和/或框图描述了本发明的各个方面。应当理解，流程图和/或框图的每个方框以及流程图和/或框图中各方框的组合，都可以由计算机可读程序指令实现。
这些计算机可读程序指令可以提供给通用计算机、专用计算机或其它可编程数据处理装置的处理器，从而生产出一种机器，使得这些指令在通过计算机或其它可编程数据处理装置的处理器执行时，产生了实现流程图和/或框图中的一个或多个方框中规定的功能/动作的装置。也可以把这些计算机可读程序指令存储在计算机可读存储介质中，这些指令使得计算机、可编程数据处理装置和/或其他设备以特定方式工作，从而，存储有指令的计算机可读介质则包括一个制造品，其包括实现流程图和/或框图中的一个或多个方框中规定的功能/动作的各个方面的指令。
也可以把计算机可读程序指令加载到计算机、其它可编程数据处理装置、或其它设备上，使得在计算机、其它可编程数据处理装置或其它设备上执行一系列操作步骤，以产生计算机实现的过程，从而使得在计算机、其它可编程数据处理装置、或其它设备上执行的指令实现流程图和/或框图中的一个或多个方框中规定的功能/动作。
附图中的流程图和框图显示了根据本发明的多个实施例的系统、方法和计算机程序产品的可能实现的体系架构、功能和操作。在这点上，流程图或框图中的每个方框可以代表一个模块、程序段或指令的一部分，所述模块、程序段或指令的一部分包含一个或多个用于实现规定的逻辑功能的可执行指令。在有些作为替换的实现中，方框中所标注的功能也可以以不同于附图中所标注的顺序发生。例如，两个连续的方框实际上可以基本并行地执行，它们有时也可以按相反的顺序执行，这依所涉及的功能而定。也要注意的是，框图和/或流程图中的每个方框、以及框图和/或流程图中的方框的组合，可以用执行规定的功能或动作的专用的基于硬件的系统来实现，或者可以用专用硬件与计算机指令的组合来实现。对于本领域技术人员来说公知的是，通过硬件方式实现、通过软件方式实现以及通过软件和硬件结合的方式实现都是等价的。
以上已经描述了本发明的各实施例，上述说明是示例性的，并非穷尽性的，并且也不限于所披露的各实施例。在不偏离所说明的各实施例的范围和精神的情况下，对于本技术领域的普通技术人员来说许多修改和变更都是显而易见的。本文中所用术语的选择，旨在最好地解释各实施例的原理、实际应用或对市场中的技术改进，或者使本技术领域的其它普通技术人员能理解本文披露的各实施例。本发明的范围由所附权利要求来限定。

Claims (10)

  1. 一种基于影像基因组学的生存预测方法，包括以下步骤：
    步骤S1，获取多个肿瘤患者的影像数据及各患者的生存期数据和RNA数据，建立数据集合；
    步骤S2，从影像数据中分割出各患者的肿瘤区域，该肿瘤区域包括真实肿瘤区域和疑似肿瘤区域；
    步骤S3，将各患者的影像数据输入神经网络，以提取影像特征并进行聚类，得到多个影像模块；
    步骤S4，利用RNA数据获取各患者的基因模块；
    步骤S5，根据基因模块与影像模块之间的相关性进行筛选，选出满足第一相关性条件的多个强相关的基因模块和影像模块；
    步骤S6，对所筛选出的基因模块中的基因进行通路富集，得到与影像模块相关的基因通路；
    步骤S7，计算基因通路的基因集变异分析分数，保留与影像模块的相关性满足第二相关性条件的强相关的基因通路，该基因通路用于对影像模块在生物学层面上进行解释；
    步骤S8，使用最终保留的影像特征进行生存预测。
  2. 根据权利要求1所述的方法，其中，在步骤S5中，所述根据基因模块与影像模块之间的相关性进行筛选，选出满足相关性阈值的多个强相关的基因模块和影像模块包括：
    对影像模块作降维处理，得到每个影像模块的特征值；
    通过降维得到的特征值计算基因模块与影像模块之间的相关性；
    根据基因模块和影像模块之间的相关性，筛选出多个强相关的基因模块和影像模块。
  3. 根据权利要求2所述的方法，其中，采用主成分分析或线性判别分析对影像模块作降维处理，根据皮尔逊相关系数或斯皮尔曼相关系数计算基因模块与影像模块之间的相关性。
  4. 根据权利要求1所述的方法，其中，在步骤S7中，所述计算基因通路的基因集变异分析分数，保留与影像模块相关的基因通路包括：
    通过基因集变异分析将基因表达矩阵转换成通路富集分数矩阵，通路富集分数矩阵中的分数代表了相关通路在对应病例上的激活程度；
    挑选出骤S6中得到的通路富集中各通路的基因集变异分析分数，使用这些基因集变异分析分数与步骤S5筛选后得到的影像模块做相关性分析；
    根据基因通路和影像模块之间的相关性，筛选出满足设定的相关性阈值的多个强相关的基因通路和影像模块。
  5. 根据权利要求1所述的方法，其中，在步骤S8中，所述使用最终保留的影像特征进行生存预测包括：
    在筛选出与基因通路具有强相关的影像模块之后，对所述神经网络进行重训练，该神经网络的最后一个全连接层的输入作为从影像数据中提取到的影像特征，在重训练时，保持全连接层之前的网络权重不变，在全连接层使用所筛选出的影像模块中的影像特征代替原有的影像特征，然后对全连接层的权重进行重新训练。
  6. 根据权利要求1所述的方法，其中，在步骤S8中，所述使用最终保留的影像特征进行生存预测包括：
    在筛选出与基因通路具有强相关的影像模块之后，对所述神经网络进行重训练，该神经网络的最后一个全连接层的输入作为从影像数据中提取到的影像特征，在重训练时，保持全连接层之前的神经网络的权重不变，对所筛选后的影像模块做降维处理，得到每个影像模块的特征值，然后使用影像模块的特征值代替原有的影像特征，对全连接层的权重进行重新训练。
  7. 根据权利要求1所述的方法，其中，步骤S1还包括：对数据集合中的图像数据进行图像标准化，以将不同量级的数据转化为统一量度。
  8. 根据权利要求5或6所述的方法，还包括：
    对于待预测的病例的影像数据，将其输入经重训练的所述神经网络，输出得到一个风险因子；
    根据所述风险因子结合该病例的生存数据，构建回归模型进行生存预测。
  9. 一种基于影像基因组学的生存预测系统，包括：
    数据采集单元：用于获取多个肿瘤患者的影像数据及各患者的生存期数据和RNA数据，建立数据集合；
    肿瘤区域提取单元：用于从影像数据中分割出各患者的肿瘤区域，该肿瘤区域包括真实肿瘤区域和疑似肿瘤区域；
    影像模块提取单元：用于将各患者的影像数据输入神经网络，以提取影像特征并进行聚类，得到多个影像模块；
    基因模块提取单元：用于利用RNA数据获取各患者的基因模块；
    第一筛选单元：用于根据基因模块与影像模块之间的相关性进行筛选，选出满足第一相关性条件的多个强相关的基因模块和影像模块；
    基因通路分析单元：用于对所筛选出的基因模块中的基因进行通路富集，得到与影像模块相关的基因通路；
    第二筛选单元：用于计算基因通路的基因集变异分析分数，保留与影像模块的相关性满足第二相关性条件的强相关的基因通路，该基因通路用于对影像模块在生物学层面上进行解释；
    生存预测单元：用于使用最终保留的影像特征进行生存预测。
  10. 一种计算机可读存储介质，其上存储有计算机程序，其中，该程序被处理器执行时实现根据权利要求1至8中任一项所述方法的步骤。
PCT/CN2021/137310 2021-03-11 2021-12-12 一种基于影像基因组学的生存预测方法和系统 WO2022188490A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202110264535.X 2021-03-11
CN202110264535.XA CN112907555B (zh) 2021-03-11 2021-03-11 一种基于影像基因组学的生存预测方法和系统

Publications (1)

Publication Number Publication Date
WO2022188490A1 true WO2022188490A1 (zh) 2022-09-15

Family

ID=76104826

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2021/137310 WO2022188490A1 (zh) 2021-03-11 2021-12-12 一种基于影像基因组学的生存预测方法和系统

Country Status (2)

Country Link
CN (1) CN112907555B (zh)
WO (1) WO2022188490A1 (zh)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117116339A (zh) * 2023-10-24 2023-11-24 中日友好医院(中日友好临床医学研究所) 基于wgcna识别影像组学生物学特征的方法及装置
CN117334246A (zh) * 2023-09-28 2024-01-02 之江实验室 一种基于计算的药物重定位的方法、装置及存储介质

Families Citing this family (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112907555B (zh) * 2021-03-11 2023-01-17 中国科学院深圳先进技术研究院 一种基于影像基因组学的生存预测方法和系统
CN114822691A (zh) * 2022-04-13 2022-07-29 上海人工智能创新中心 基于图卷积神经网络的临床事件预测装置
CN114999569B (zh) * 2022-08-03 2022-12-20 北京汉博信息技术有限公司 一种针对病灶基质的分型方法、装置及计算机可读介质
CN116385441B (zh) * 2023-06-05 2023-09-05 中国科学院深圳先进技术研究院 一种基于mri对少突胶质细胞瘤进行风险分层的方法及系统
CN116403076B (zh) * 2023-06-06 2023-08-22 中国科学院深圳先进技术研究院 一种基于dti序列对gbm患者进行风险分层的方法及系统
CN116705296A (zh) * 2023-06-06 2023-09-05 中国科学院深圳先进技术研究院 一种基于常规mri序列对gbm患者进行风险分层的方法及系统

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106815481A (zh) * 2017-01-19 2017-06-09 中国科学院深圳先进技术研究院 一种基于影像组学的生存期预测方法及装置
CN107169497A (zh) * 2017-04-14 2017-09-15 中国科学院苏州生物医学工程技术研究所 一种基于基因影像学的肿瘤影像标记物提取方法
CN110097921A (zh) * 2019-05-30 2019-08-06 复旦大学 基于影像组学的胶质瘤内基因异质性可视化定量方法和系统
US20190370969A1 (en) * 2018-05-30 2019-12-05 Siemens Healthcare Gmbh Methods for generating synthetic training data and for training deep learning algorithms for tumor lesion characterization, method and system for tumor lesion characterization, computer program and electronically readable storage medium
CN112907555A (zh) * 2021-03-11 2021-06-04 中国科学院深圳先进技术研究院 一种基于影像基因组学的生存预测方法和系统

Family Cites Families (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105046110A (zh) * 2015-06-28 2015-11-11 中国科学院合肥物质科学研究院 一种克服信号失真的肿瘤细胞异常通路识别方法
CN108897984A (zh) * 2018-05-07 2018-11-27 上海理工大学 基于ct影像组学特征与肺癌基因表达间相关性分析方法
JP7100336B2 (ja) * 2019-05-16 2022-07-13 ペイジ．エーアイ インコーポレイテッド デジタル病理学のために、画像を処理し、処理された画像を分類するためのシステムおよび方法
CN110390996A (zh) * 2019-08-18 2019-10-29 段艺 一种肝细胞癌可变剪接事件预后模型及其构建方法和应用
KR102170297B1 (ko) * 2019-12-16 2020-10-26 주식회사 루닛 조직병리체학 데이터의 해석 정보를 제공하는 방법 및 시스템
CN112330596A (zh) * 2020-10-14 2021-02-05 达州市中心医院 一种基于机器学习算法的用于预测crc转移的影像组学（ra）模型

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106815481A (zh) * 2017-01-19 2017-06-09 中国科学院深圳先进技术研究院 一种基于影像组学的生存期预测方法及装置
CN107169497A (zh) * 2017-04-14 2017-09-15 中国科学院苏州生物医学工程技术研究所 一种基于基因影像学的肿瘤影像标记物提取方法
US20190370969A1 (en) * 2018-05-30 2019-12-05 Siemens Healthcare Gmbh Methods for generating synthetic training data and for training deep learning algorithms for tumor lesion characterization, method and system for tumor lesion characterization, computer program and electronically readable storage medium
CN110097921A (zh) * 2019-05-30 2019-08-06 复旦大学 基于影像组学的胶质瘤内基因异质性可视化定量方法和系统
CN112907555A (zh) * 2021-03-11 2021-06-04 中国科学院深圳先进技术研究院 一种基于影像基因组学的生存预测方法和系统

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117334246A (zh) * 2023-09-28 2024-01-02 之江实验室 一种基于计算的药物重定位的方法、装置及存储介质
CN117116339A (zh) * 2023-10-24 2023-11-24 中日友好医院(中日友好临床医学研究所) 基于wgcna识别影像组学生物学特征的方法及装置

Also Published As

Publication number Publication date
CN112907555B (zh) 2023-01-17
CN112907555A (zh) 2021-06-04

Similar Documents

Publication Publication Date Title
WO2022188490A1 (zh) 一种基于影像基因组学的生存预测方法和系统
Shimizu et al. Artificial intelligence in oncology
Taylor et al. Automated detection of moderate and large pneumothorax on frontal chest X-rays using deep convolutional neural networks: A retrospective study
Dias et al. Artificial intelligence in clinical and genomic diagnostics
US11462325B2 (en) Multimodal machine learning based clinical predictor
CN108062753B (zh) 基于深度对抗学习的无监督域自适应脑肿瘤语义分割方法
Casiraghi et al. Explainable machine learning for early assessment of COVID-19 risk prediction in emergency departments
Shanmugam et al. Alzheimer’s disease classification using pre-trained deep networks
Che et al. Interpretable deep models for ICU outcome prediction
US20200185055A1 (en) Methods and Systems for Nucleic Acid Variant Detection and Analysis
Wang et al. Does non-COVID-19 lung lesion help? investigating transferability in COVID-19 CT image segmentation
CN114945951A (zh) 用于图像分类的不确定性引导的半监督神经网络训练
Baltazar et al. Artificial intelligence on COVID-19 pneumonia detection using chest xray images
Hamida et al. A novel COVID-19 diagnosis support system using the stacking approach and transfer learning technique on chest X-ray images
CN114841280B (zh) 一种复杂疾病的预测分类方法、系统、介质、设备及终端
Montazeri et al. Machine learning models for image-based diagnosis and prognosis of COVID-19: Systematic review
JP2022535727A (ja) 創薬のための医療エンティティ間の意味関係の自動的検証および強化
Ma et al. A multitask deep learning approach for pulmonary embolism detection and identification
Kumar Applications of machine learning in disease pre-screening
Badea et al. Identifying transcriptomic correlates of histology using deep learning
US11354591B2 (en) Identifying gene signatures and corresponding biological pathways based on an automatically curated genomic database
Han-Trong et al. An efficient method for diagnosing brain tumors based on MRI images using deep convolutional neural networks
Khanna et al. Polygenic Risk Score for Cardiovascular Diseases in Artificial Intelligence Paradigm: A Review
Kaur et al. Rectal Cancer Prediction and Performance Based on Intelligent Variational Autoencoders Machine Using Deep Learning on CDAS Dataset
Rashid et al. ADVANCING THE UNDERSTANDING OF CLINICAL SEPSIS USING GENE EXPRESSION–DRIVEN MACHINE LEARNING TO IMPROVE PATIENT OUTCOMES

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 21929952

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 21929952

Country of ref document: EP

Kind code of ref document: A1