CN116994770A - Immune crowd determination method and system based on multidimensional analysis - Google Patents
Immune crowd determination method and system based on multidimensional analysis Download PDFInfo
- Publication number
- CN116994770A CN116994770A CN202311257835.0A CN202311257835A CN116994770A CN 116994770 A CN116994770 A CN 116994770A CN 202311257835 A CN202311257835 A CN 202311257835A CN 116994770 A CN116994770 A CN 116994770A
- Authority
- CN
- China
- Prior art keywords
- image
- target
- target disease
- drug
- resistant
- Prior art date
- Legal status (The legal status 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 status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 40
- 238000004458 analytical method Methods 0.000 title claims description 30
- 238000011282 treatment Methods 0.000 claims abstract description 117
- 239000003814 drug Substances 0.000 claims abstract description 104
- 229940079593 drug Drugs 0.000 claims abstract description 103
- 230000000694 effects Effects 0.000 claims abstract description 35
- 238000009169 immunotherapy Methods 0.000 claims abstract description 33
- 208000002154 non-small cell lung carcinoma Diseases 0.000 claims abstract description 27
- 238000012549 training Methods 0.000 claims abstract description 15
- 201000010099 disease Diseases 0.000 claims description 204
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 claims description 204
- 230000008859 change Effects 0.000 claims description 83
- 239000013610 patient sample Substances 0.000 claims description 45
- 238000001514 detection method Methods 0.000 claims description 31
- 102000052116 epidermal growth factor receptor activity proteins Human genes 0.000 claims description 22
- 108700015053 epidermal growth factor receptor activity proteins Proteins 0.000 claims description 22
- 239000003112 inhibitor Substances 0.000 claims description 22
- 229940121647 egfr inhibitor Drugs 0.000 claims description 21
- YOHYSYJDKVYCJI-UHFFFAOYSA-N n-[3-[[6-[3-(trifluoromethyl)anilino]pyrimidin-4-yl]amino]phenyl]cyclopropanecarboxamide Chemical group FC(F)(F)C1=CC=CC(NC=2N=CN=C(NC=3C=C(NC(=O)C4CC4)C=CC=3)C=2)=C1 YOHYSYJDKVYCJI-UHFFFAOYSA-N 0.000 claims description 21
- 238000012216 screening Methods 0.000 claims description 21
- 206010059866 Drug resistance Diseases 0.000 claims description 20
- 238000004364 calculation method Methods 0.000 claims description 15
- 108090000623 proteins and genes Proteins 0.000 claims description 15
- 206010064571 Gene mutation Diseases 0.000 claims description 9
- 238000013136 deep learning model Methods 0.000 claims description 9
- 230000001024 immunotherapeutic effect Effects 0.000 claims description 7
- 238000000611 regression analysis Methods 0.000 claims description 7
- 238000011176 pooling Methods 0.000 claims 1
- 238000004393 prognosis Methods 0.000 abstract description 33
- 238000013528 artificial neural network Methods 0.000 abstract description 22
- 238000006243 chemical reaction Methods 0.000 abstract description 5
- 238000004141 dimensional analysis Methods 0.000 abstract description 5
- 208000029729 tumor suppressor gene on chromosome 11 Diseases 0.000 description 23
- 206010028980 Neoplasm Diseases 0.000 description 19
- 230000004083 survival effect Effects 0.000 description 12
- 238000011160 research Methods 0.000 description 8
- 206010058467 Lung neoplasm malignant Diseases 0.000 description 7
- 201000005202 lung cancer Diseases 0.000 description 7
- 208000020816 lung neoplasm Diseases 0.000 description 7
- 229940076838 Immune checkpoint inhibitor Drugs 0.000 description 5
- 108091008026 Inhibitory immune checkpoint proteins Proteins 0.000 description 5
- 102000037984 Inhibitory immune checkpoint proteins Human genes 0.000 description 5
- 230000008901 benefit Effects 0.000 description 5
- 238000004422 calculation algorithm Methods 0.000 description 5
- 238000010586 diagram Methods 0.000 description 5
- 238000000605 extraction Methods 0.000 description 5
- 239000012274 immune-checkpoint protein inhibitor Substances 0.000 description 5
- 230000004048 modification Effects 0.000 description 5
- 238000012986 modification Methods 0.000 description 5
- 230000035772 mutation Effects 0.000 description 5
- 210000001519 tissue Anatomy 0.000 description 5
- 229940121358 tyrosine kinase inhibitor Drugs 0.000 description 5
- 238000010276 construction Methods 0.000 description 4
- 238000013135 deep learning Methods 0.000 description 4
- 230000001225 therapeutic effect Effects 0.000 description 4
- 206010027476 Metastases Diseases 0.000 description 3
- 238000003745 diagnosis Methods 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 210000004072 lung Anatomy 0.000 description 3
- 239000011159 matrix material Substances 0.000 description 3
- 230000009401 metastasis Effects 0.000 description 3
- 230000001575 pathological effect Effects 0.000 description 3
- 230000007170 pathology Effects 0.000 description 3
- 230000004044 response Effects 0.000 description 3
- 239000005483 tyrosine kinase inhibitor Substances 0.000 description 3
- 150000004917 tyrosine kinase inhibitor derivatives Chemical class 0.000 description 3
- 238000012795 verification Methods 0.000 description 3
- 238000011122 anti-angiogenic therapy Methods 0.000 description 2
- 210000000038 chest Anatomy 0.000 description 2
- 238000002591 computed tomography Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000002489 hematologic effect Effects 0.000 description 2
- 230000002757 inflammatory effect Effects 0.000 description 2
- 230000003993 interaction Effects 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 230000007246 mechanism Effects 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 230000000391 smoking effect Effects 0.000 description 2
- 238000002626 targeted therapy Methods 0.000 description 2
- 208000010507 Adenocarcinoma of Lung Diseases 0.000 description 1
- 230000007067 DNA methylation Effects 0.000 description 1
- -1 EGFR amino acid Chemical class 0.000 description 1
- 101000851181 Homo sapiens Epidermal growth factor receptor Proteins 0.000 description 1
- 108091000080 Phosphotransferase Proteins 0.000 description 1
- 238000012952 Resampling Methods 0.000 description 1
- 210000001015 abdomen Anatomy 0.000 description 1
- 230000004913 activation Effects 0.000 description 1
- 230000000259 anti-tumor effect Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 238000012098 association analyses Methods 0.000 description 1
- 230000004791 biological behavior Effects 0.000 description 1
- 210000000988 bone and bone Anatomy 0.000 description 1
- 210000004556 brain Anatomy 0.000 description 1
- 201000011510 cancer Diseases 0.000 description 1
- 238000002512 chemotherapy Methods 0.000 description 1
- 238000009096 combination chemotherapy Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000013170 computed tomography imaging Methods 0.000 description 1
- 238000004590 computer program Methods 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 238000002790 cross-validation Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000010201 enrichment analysis Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 210000002950 fibroblast Anatomy 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 230000004907 flux Effects 0.000 description 1
- 238000009499 grossing Methods 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 210000002865 immune cell Anatomy 0.000 description 1
- 230000036039 immunity Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000011337 individualized treatment Methods 0.000 description 1
- 230000008595 infiltration Effects 0.000 description 1
- 238000001764 infiltration Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 239000003446 ligand Substances 0.000 description 1
- 210000004185 liver Anatomy 0.000 description 1
- 201000005249 lung adenocarcinoma Diseases 0.000 description 1
- 239000003550 marker Substances 0.000 description 1
- 238000010197 meta-analysis Methods 0.000 description 1
- 238000012164 methylation sequencing Methods 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000035764 nutrition Effects 0.000 description 1
- 235000016709 nutrition Nutrition 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 102000020233 phosphotransferase Human genes 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 230000002035 prolonged effect Effects 0.000 description 1
- 238000011002 quantification Methods 0.000 description 1
- 230000003439 radiotherapeutic effect Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 206010041823 squamous cell carcinoma Diseases 0.000 description 1
- 238000011272 standard treatment Methods 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000002560 therapeutic procedure Methods 0.000 description 1
- 210000000779 thoracic wall Anatomy 0.000 description 1
- 210000001364 upper extremity Anatomy 0.000 description 1
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/70—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for mining of medical data, e.g. analysing previous cases of other patients
-
- 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
- G06N3/045—Combinations of networks
-
- 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
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H30/00—ICT specially adapted for the handling or processing of medical images
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/30—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for calculating health indices; for individual health risk assessment
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/50—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for simulation or modelling of medical disorders
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H70/00—ICT specially adapted for the handling or processing of medical references
- G16H70/60—ICT specially adapted for the handling or processing of medical references relating to pathologies
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Public Health (AREA)
- Medical Informatics (AREA)
- General Health & Medical Sciences (AREA)
- Data Mining & Analysis (AREA)
- Biomedical Technology (AREA)
- Epidemiology (AREA)
- Primary Health Care (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Pathology (AREA)
- Databases & Information Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computing Systems (AREA)
- Artificial Intelligence (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Evolutionary Computation (AREA)
- Molecular Biology (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Radiology & Medical Imaging (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Image Analysis (AREA)
- Investigating Or Analysing Biological Materials (AREA)
Abstract
The invention belongs to the field of intelligent medical treatment, in particular to a multi-dimensional analysis-based immune crowd determination method and system, which are used for searching CT image histology characteristics related to curative effect and prognosis of a GFR-TKI drug-resistant NSCLC patient receiving immune treatment, and constructing and verifying a multi-dimensional prognosis model based on the image histology characteristics; training a deep neural network prediction system based on image histology characteristics, which is used for predicting the curative effect and prognosis of immunotherapy, verifying the external applicability and generalization capability of the deep neural network in a prospective data set, and determining the value of the deep neural network in the curative effect and prognosis prediction of a GFR-TKI drug-resistant NSCLC patient, so as to realize the conversion from theory to clinical practice application; the invention discusses the potential molecular basis of model curative effect and prognosis prediction, and clarifies the molecular biological characteristics reflected by the phenotype characteristics of the deep neural network prediction system.
Description
Technical Field
The invention belongs to the field of intelligent medical treatment, and particularly relates to an immune crowd determination method and system based on multidimensional analysis.
Background
Lung cancer is the malignant tumor with highest morbidity and mortality worldwide, and the number of new cases per year exceeds 200 ten thousand. The early lung cancer is asymptomatic, more than 70% of patients are already in the late stage at the time of diagnosis, and the survival rate of 5 years is less than 20%. About 85% of lung cancer patients are Non-small cell lung cancer (NSCLC), with lung adenocarcinoma and squamous carcinoma being the most common pathological types. In the immunotherapeutic era, tyrosine Kinase Inhibitor (TKI) targeted therapy is the first line of treatment for NSCLC patients with driving gene mutations. However, most patients develop drug resistance within 12 months, and the treatment methods after drug resistance are limited. Recent studies have shown that Immune Checkpoint Inhibitor (ICI) treatment has become one of the standard therapeutic approaches for advanced NSCLC, whereas clinical studies have shown that EGFR mutated lung cancer rarely benefits from Immune Checkpoint Inhibitor (ICI) treatment. Recent studies have shown that ICI treatment has shown superior results in EGFR-sensitive mutant non-small cell lung cancer. However, it is not clear at present that the population benefiting from EGFR mutant lung cancer patients can be provided with an accurate prediction platform, and accurate prediction of the response of patients to targeted therapy before treatment is helpful for making a more reasonable individualized treatment scheme, so that the method has important clinical guiding significance.
Recent researches find that the image histology researches can extract a large amount of image features from image images in a high throughput manner, convert image information into digital information, wherein a large amount of intensity, structure, edge and texture features can be automatically extracted and quantified, and the size, shape and texture features and the like of tumor tissues are characterized, and the previous researches have proved the potential of the image histology researches in diagnosis stage, prognosis and curative effect evaluation of NSCLC patients. By characterizing internal heterogeneity, the image histology features can indirectly reflect the state of the tumor tissue microenvironment, while the microenvironment states such as immune cell infiltration, fibroblast interaction and the like in the tumor tissue have been proved to be related to the curative effect of immunotherapy, so the image histology features are potential markers for predicting the curative effect of the tumor immunotherapy. Based on the complexity and heterogeneity of tumor and the complex mechanism of EGFR mutation, single image histology features only represent the internal features of tumor tissues, while multidimensional source data can comprehensively evaluate the biological behaviors of tumor, the nutrition, the immune state and the like of organism, and provide more abundant feature information. The past research results also show that the prognosis model constructed by integrating the clinical pathology and the image histology features is obviously superior to a single parameter model, which suggests that the multidimensional data source can further improve the efficiency of the prognosis model. The multidimensional integrated model constructed based on the image histology features is expected to more accurately predict the curative effect of EGFR-TKI patient immunotherapy.
The most widely used survival model at present is a Cox proportional hazards model, which relies on linear proportional assumptions and cannot reflect interactions between covariates; however, in real world studies, nonlinear correlations between variables are more common, and thus the CPH model is not entirely applicable to clinical medical studies. With the development of artificial intelligence technology, the deep learning algorithm can process a high-dimensional and complex nonlinear function through automatic learning, so that the dilemma is well solved. The deep neural network is a network model of a deep learning algorithm, which comprises an input layer, a plurality of hidden layers and an output layer, wherein the connection between the input layer and the hidden layers is established through layer-by-layer learning, and the intermediate plurality of hidden layers convert the network from linear to nonlinear, so that big data with complex nonlinear characteristics can be processed: and finally, comprehensively weighting and connecting the operation results of the hidden layer with the output layer, thereby achieving the purpose of model construction and providing technical support for the construction of the multidimensional model under the big data background. Deep neural networks have begun to be applied in clinical medicine related studies such as tumor diagnosis, molecular typing, etc., but in view of the complexity of survival analysis outcome variables, research of deep neural networks in survival analysis and prognosis model construction has remained relatively limited.
In the prior art, only the therapeutic efficacy of EGFR mutant patients is focused on, and no treatment after EGFR-TKI treatment resistance is focused on. Therefore, the invention aims to extract mass image quantitative characteristics of lung cancer focus CT based on chest CT images of patient treatment by utilizing an image histology technology, screen out image histology labels highly related to the treatment effect of EGFR-TKI post-drug-resistant immune patients by combining a deep neural network system, fuse clinical indexes, and combine patient survival prognosis information to construct a platform for directly predicting the treatment effect of EGFR-TKI post-drug-resistant immune patients so as to screen out targeted treatment dominant population, guide clinical decisions, and has great scientific research significance and clinical application value.
Disclosure of Invention
According to a first aspect of the present invention, the present invention claims a method for determining immune population based on multidimensional analysis, comprising:
acquiring a historical patient of a target disease, extracting treatment change data and physiological attribute data of the historical patient, and constructing an image histology reference database;
inputting a target disease inhibitor into a training patient sample with a target disease to obtain a target disease drug-resistant patient sample and analyzing the drug-resistant image characteristics of the target disease drug-resistant patient sample;
According to the drug-resistant image characteristics of the target disease drug-resistant patient sample and the image histology reference database, performing data matching, dividing the target disease drug-resistant patient sample into an effective group and an ineffective group, and setting corresponding immune crowd determination weights;
and obtaining image detection characteristics of a patient with the target disease to be determined, carrying out similarity calculation on the image detection characteristics and drug-resistant image characteristics, and predicting immune treatment dominant population of the target disease based on a multidimensional deep learning model of image histology.
Further, the step of obtaining the historic patient of the target disease, extracting treatment change data and physiological attribute data of the historic patient, and constructing an image histology reference database specifically comprises the following steps:
acquiring historical patients of target diseases, extracting historical patients of the target diseases with similar treatment paths from the acquired historical patients of the target diseases, and taking the historical patients of the target diseases with similar treatment paths as an initial cluster;
querying treatment change data corresponding to historical patients of the target disease in the initial cluster;
acquiring a preset treatment change rate, and acquiring a historical patient of the target disease, in which the treatment change data is within the preset treatment change rate, from the initial cluster as a historical patient of the target disease;
Extracting historical patients of the target diseases from the initial clusters to obtain target clusters, and establishing an image histology reference database according to the target clusters;
acquiring treatment change data corresponding to historical patients of the target disease in the target cluster in the image histology reference database;
calculating the average dressing change frequency and the physiological attribute change value of the treatment change data corresponding to the historical patient of the target disease;
obtaining immune reference values of the target clusters in the image histology reference database according to the average dressing change frequency and the physiological attribute change value;
obtaining a target immune reference value, comparing the immune reference value with the target immune reference value to obtain whether the target cluster is credible or not, and outputting a judgment result of whether the target cluster is credible or not.
Further, the step of inputting the target disease inhibitor into a training patient sample with the target disease to obtain a target disease drug-resistant patient sample and analyzing the drug-resistant image characteristics of the target disease drug-resistant patient sample specifically includes:
the disease of interest is EGFR mutated NSCLC;
the target disease inhibitor is an EGFR-TKI inhibitor;
And the characteristic content of the drug-resistant image characteristic is the same as the data content of the image histology reference database.
Further, the data matching is performed according to the drug-resistant image characteristics of the target disease drug-resistant patient sample and the image histology reference database, the target disease drug-resistant patient sample is divided into an effective group and an ineffective group, and corresponding immune crowd determination weights are set, and the method specifically comprises the following steps:
dividing focus ROI (region of interest) areas of the drug-resistant image features to obtain areas to be extracted;
extracting effective features from the region to be extracted, and screening labels;
matching the extracted effective features and the screened labels with treatment change data and physiological attribute data of the historical patients in the image histology reference database;
dividing the corresponding matched target disease drug-resistant patients into an effective group and an ineffective group according to the treatment efficacy of the historical patients, and setting the immune crowd determination weights of the effective group and the ineffective group.
Further, the obtaining the image detection feature of the patient with the target disease to be determined, performing similarity calculation on the image detection feature and the drug-resistant image feature, and predicting the dominant population of immunotherapy of the target disease based on the multidimensional deep learning model of image histology specifically includes:
Performing similarity calculation on the image detection features and the drug-resistant image features to obtain a reference target disease drug-resistant patient with similarity meeting a preset threshold;
fusing the image detection characteristics to construct an individuation prediction model, and constructing a gene state and immunotherapy curative effect prediction model by fusing image histology tags, gene mutation and clinical information by using a multivariate Cox regression analysis method and Nomogram;
inputting the reference target disease drug-resistant patient into the gene state and immunotherapy curative effect prediction model to obtain the determining weight of the immune crowd to be determined of the target disease patient to be determined;
and selecting the patient with the target disease to be determined, the weight of which is greater than a preset weight value, as the immunotherapeutic dominant population of the target disease.
According to a second aspect of the present invention, the present invention claims an immune population determining system based on multidimensional analysis, comprising:
the database building module is used for obtaining a historical patient of a target disease, extracting treatment change data and physiological attribute data of the historical patient and building an image histology reference database;
the drug resistance analysis module is used for inputting a target disease inhibitor to a training patient sample with a target disease to obtain a target disease drug resistance patient sample and analyzing the drug resistance image characteristics of the target disease drug resistance patient sample;
The weight determining module is used for carrying out data matching on the drug-resistant image characteristics of the target disease drug-resistant patient sample and the image histology reference database, dividing the target disease drug-resistant patient sample into an effective group and an ineffective group, and setting corresponding immune crowd determining weights;
the crowd determination module is used for obtaining image detection characteristics of a patient with a target disease to be determined, carrying out similarity calculation on the image detection characteristics and drug-resistant image characteristics, and predicting immune treatment dominant crowd of the target disease based on a multidimensional deep learning model of image histology.
Further, the database creation module specifically includes:
acquiring historical patients of target diseases, extracting historical patients of the target diseases with similar treatment paths from the acquired historical patients of the target diseases, and taking the historical patients of the target diseases with similar treatment paths as an initial cluster;
querying treatment change data corresponding to historical patients of the target disease in the initial cluster;
acquiring a preset treatment change rate, and acquiring a historical patient of the target disease, in which the treatment change data is within the preset treatment change rate, from the initial cluster as a historical patient of the target disease;
Extracting historical patients of the target diseases from the initial clusters to obtain target clusters, and establishing an image histology reference database according to the target clusters;
acquiring treatment change data corresponding to historical patients of the target disease in the target cluster in the image histology reference database;
calculating the average dressing change frequency and the physiological attribute change value of the treatment change data corresponding to the historical patient of the target disease;
obtaining immune reference values of the target clusters in the image histology reference database according to the average dressing change frequency and the physiological attribute change value;
obtaining a target immune reference value, comparing the immune reference value with the target immune reference value to obtain whether the target cluster is credible or not, and outputting a judgment result of whether the target cluster is credible or not.
Further, the drug resistance analysis module specifically includes:
the disease of interest is EGFR mutated NSCLC;
the target disease inhibitor is an EGFR-TKI inhibitor;
and the characteristic content of the drug-resistant image characteristic is the same as the data content of the image histology reference database.
Further, the weight determining module specifically includes:
Dividing focus ROI (region of interest) areas of the drug-resistant image features to obtain areas to be extracted;
extracting effective features from the region to be extracted, and screening labels;
matching the extracted effective features and the screened labels with treatment change data and physiological attribute data of the historical patients in the image histology reference database;
dividing the corresponding matched target disease drug-resistant patients into an effective group and an ineffective group according to the treatment efficacy of the historical patients, and setting the immune crowd determination weights of the effective group and the ineffective group.
Further, the crowd determination module specifically includes:
performing similarity calculation on the image detection features and the drug-resistant image features to obtain a reference target disease drug-resistant patient with similarity meeting a preset threshold;
fusing the image detection characteristics to construct an individuation prediction model, and constructing a gene state and immunotherapy curative effect prediction model by fusing image histology tags, gene mutation and clinical information by using a multivariate Cox regression analysis method and Nomogram;
inputting the reference target disease drug-resistant patient into the gene state and immunotherapy curative effect prediction model to obtain the determining weight of the immune crowd to be determined of the target disease patient to be determined;
And selecting the patient with the target disease to be determined, the weight of which is greater than a preset weight value, as the immunotherapeutic dominant population of the target disease.
The invention requests to protect an immune crowd determining method and system based on multidimensional analysis, searches CT image histology characteristics related to curative effect and prognosis of a GFR-TKI drug-resistant NSCLC patient receiving immune treatment, and constructs and verifies a multidimensional prognosis model based on the image histology characteristics; training a deep neural network prediction system based on image histology characteristics, which is used for predicting the curative effect and prognosis of immunotherapy, verifying the external applicability and generalization capability of the deep neural network in a prospective data set, and determining the value of the deep neural network in the curative effect and prognosis prediction of a GFR-TKI drug-resistant NSCLC patient, so as to realize the conversion from theory to clinical practice application; the invention discusses the potential molecular basis of model curative effect and prognosis prediction, and clarifies the molecular biological characteristics reflected by the phenotype characteristics of the deep neural network prediction system.
Drawings
FIG. 1 is a workflow diagram of a multi-dimensional analysis-based immune population determination method as claimed in the present invention;
FIG. 2 is a second workflow diagram of a multi-dimensional analysis-based immune population determination method as claimed in the present invention;
FIG. 3 is a third workflow diagram of a multi-dimensional analysis-based immune population determination method in accordance with the claimed invention;
FIG. 4 is a fourth operational flow diagram of a multi-dimensional analysis-based immune population determination method in accordance with the present invention;
FIG. 5 is a block diagram of an immune population determination system based on multidimensional analysis as claimed in the present invention.
Detailed Description
Human epidermal growth factor receptor (EGFR amino acid kinase (TKI) is a first-line standard treatment for treating patients with EGFR-sensitive mutant advanced non-small cell lung cancer (NSCLC) but most patients can develop drug resistance within 12 months, and the treatment method after drug resistance is limited.
Some clinical trials and meta analysis also suggest that single-drug immunotherapy has limited benefit in EGFR mutant populations and that efficacy is inferior to EGFR wild-type patients. However, IMpower150 study to explore the value of different immune combination regimens in EGFR mutant populations, a subset of EGFR mutations was designed, and the results of the study showed that the total survival time of the programmed death ligand inhibitor combination chemotherapy and anti-angiogenic therapy was significantly prolonged compared to patients treated with chemotherapy in combination with anti-angiogenic therapy.
The international multicenter research subverts the theory of immune immunity to EGFR mutation, increases confidence for application of immunotherapy in EGFR mutant people, and brings new treatment options for EGFR-TKI resistant patients, however, people who benefit from EGFR mutant lung cancer patients in immunotherapy are not clear at present. If an accurate prediction platform can be constructed, the response of patients to immunotherapy can be accurately predicted before treatment, which is helpful for making a more reasonable individual treatment scheme and has important clinical guiding significance.
Image histology characterizes tumor heterogeneity by high throughput extraction and quantification of image features, and can be used for prognosis studies of tumor patients. In view of tumor heterogeneity and a complex anti-tumor mechanism of EGFR-TKI drug resistance, a single-mode marker cannot accurately predict prognosis, and integration of multidimensional data based on image histology can comprehensively reflect organism and tumor heterogeneity, so that prognosis prediction efficiency is expected to be further improved. The embodiment is to predict the curative effect of EGFR-TKI drug-resistant immunotherapy based on a multidimensional integrated model constructed by image histology characteristics, and construct the first image histology-based deep neural network curative effect and survival prediction system of NSCLC patients by training the deep neural network for predicting the curative effect of EGFR-TKI drug-resistant immunotherapy, so as to realize the conversion of deep learning from theory to clinical practice application. The potential molecular basis of model efficacy prediction is discussed on the basis, and theoretical basis and technical support are provided for wide application of future deep learning in tumor patient efficacy prediction.
According to a first embodiment of the present invention, referring to fig. 1, the present invention claims an immune crowd determining method based on multidimensional analysis, which is characterized by comprising:
acquiring a historical patient of a target disease, extracting treatment change data and physiological attribute data of the historical patient, and constructing an image histology reference database;
inputting a target disease inhibitor into a training patient sample with a target disease to obtain a target disease drug-resistant patient sample and analyzing the drug-resistant image characteristics of the target disease drug-resistant patient sample;
according to the drug-resistant image characteristics of the target disease drug-resistant patient sample and the image histology reference database, performing data matching, dividing the target disease drug-resistant patient sample into an effective group and an ineffective group, and setting corresponding immune crowd determination weights;
and obtaining image detection characteristics of a patient with the target disease to be determined, carrying out similarity calculation on the image detection characteristics and drug-resistant image characteristics, and predicting immune treatment dominant population of the target disease based on a multidimensional deep learning model of image histology.
Further, referring to fig. 2, the step of obtaining the historic patient of the target disease, extracting treatment variation data and physiological attribute data of the historic patient, and constructing an image histology reference database specifically includes:
Acquiring historical patients of target diseases, extracting historical patients of the target diseases with similar treatment paths from the acquired historical patients of the target diseases, and taking the historical patients of the target diseases with similar treatment paths as an initial cluster;
querying treatment change data corresponding to historical patients of the target disease in the initial cluster;
acquiring a preset treatment change rate, and acquiring a historical patient of the target disease, in which the treatment change data is within the preset treatment change rate, from the initial cluster as a historical patient of the target disease;
extracting historical patients of the target diseases from the initial clusters to obtain target clusters, and establishing an image histology reference database according to the target clusters;
acquiring treatment change data corresponding to historical patients of the target disease in the target cluster in the image histology reference database;
calculating the average dressing change frequency and the physiological attribute change value of the treatment change data corresponding to the historical patient of the target disease;
obtaining immune reference values of the target clusters in the image histology reference database according to the average dressing change frequency and the physiological attribute change value;
Obtaining a target immune reference value, comparing the immune reference value with the target immune reference value to obtain whether the target cluster is credible or not, and outputting a judgment result of whether the target cluster is credible or not.
In this embodiment, after the obtaining the target immune reference value and comparing the immune reference value with the target immune reference value to obtain whether the target cluster is trusted, the method includes:
when the credibility of the target clusters is judged according to the immune reference values, covariance among the treatment change data corresponding to different target clusters is obtained;
acquiring variances of the treatment change data corresponding to different target clusters;
obtaining an associated reference value of the target cluster in the image histology reference database according to the covariance and the variance;
and obtaining a target association reference value, and comparing the association reference value with the target association reference value to obtain whether the target cluster is credible or not.
The obtaining the target association reference value, comparing the association reference value with the target association reference value to obtain whether the target cluster is credible or not, and then comprises the following steps:
when the credibility of the target clusters is judged according to the association reference values, target disease information corresponding to the target clusters is obtained;
Calculating the average dressing change frequency of the target diseases of the treatment change data corresponding to the target disease information in the target cluster;
obtaining a stability reference value of the target cluster in the image histology reference database according to the average dressing change frequency of the target disease;
and obtaining a target stability reference value, comparing the stability reference value with the target stability reference value, and judging whether the target cluster is credible or not.
After outputting the judging result of whether the target cluster is credible, the method comprises the following steps:
when the judging result is credible, receiving a prediction instruction for predicting treatment change data, wherein the prediction instruction carries data to be predicted;
inquiring the target cluster corresponding to the data to be predicted as a predicted target cluster;
acquiring target disease information corresponding to the data to be predicted, and acquiring first target disease treatment cost corresponding to the target disease information in the predicted target cluster;
when the treatment cost of the first target disease is lower than a preset value, adding a difficult-to-immunity tag to the predicted target cluster corresponding to the treatment cost of the first target disease;
and calculating the data to be predicted by adopting a first target disease treatment cost corresponding to the predicted target cluster without adding the difficult-to-immunity tag.
After outputting the judging result of whether the target cluster is credible, the method further comprises the following steps:
acquiring medical data to be judged and treatment change data to be judged corresponding to the medical data to be judged;
inquiring target clusters corresponding to the medical data to be judged;
extracting the target disease information contained in the medical data to be judged, and acquiring second target disease treatment cost corresponding to the target disease information in the target cluster;
calculating a difference value between the treatment change data to be judged and the treatment cost of the second target disease;
and outputting prompt information for prompting and monitoring the treatment change data to be judged when the difference exceeds the threshold value.
Specifically, in this example, all clinical pathology data included in the patient, including patient gender, age, smoking history, EGFR mutation status, anatomical typing, whether liver metastasis, bone metastasis, brain metastasis, were collected. The reinforced CT images of all the patients in the group within 1 month before treatment are collected, a Siemens 64-layer spiral CT scanner is adopted for CT scanning, the patients are in a supine position before scanning, chest and abdomen articles are removed, and the upper limbs of the patients are held up and scanned at the end of deep inhalation. Specific parameters for CT scan are as follows: tube voltage, 120kV; tube current, 250mAs; pitch, 1.062. The detector size was 0.625mm×64, the scanning layer thickness was 5mm, and the pitch was 5m. Enhanced CT images of arterial phase are extracted from the image archiving and communication system and stored as DICOM format images for feature extraction. The primary endpoint of this example was PFS and the secondary endpoint was 0S. Efficacy was assessed by imaging detection every 2 cycles according to RECIST1.1 solid tumor efficacy reference criteria, including complete/partial remission, stable, progression. PFS is defined as the time to begin receiving immunotherapy to the progression or death of the disease, and OS is defined as the time to begin receiving immunotherapy to the death of the patient. And integrating all data of the patient to construct an image histology database.
Further, the step of inputting the target disease inhibitor into a training patient sample with the target disease to obtain a target disease drug-resistant patient sample and analyzing the drug-resistant image characteristics of the target disease drug-resistant patient sample specifically includes:
the disease of interest is EGFR mutated NSCLC;
the target disease inhibitor is an EGFR-TKI inhibitor;
and the characteristic content of the drug-resistant image characteristic is the same as the data content of the image histology reference database.
Wherein in this embodiment, the following operations are required:
(1) Screening the immune treatment efficacy and prognosis-related CT image histology characteristics of EGFR-TKI resistant NSCLC patients receiving immune treatment, and constructing and verifying image histology tags.
The method comprises the steps of (1) taking an EGFR-TKI drug-resistant NSCLC patient receiving immunotherapy into a tumor reinforced CT image before treatment, extracting massive image histology characteristics, reducing dimensions by applying minimum absolute value convergence, a selection operator algorithm and other method characteristics, and screening characteristics independent of progression-free survival; and calculating an image histology tag according to the characteristic coefficient, and verifying the efficacy of the image histology tag in an independent verification set.
(2) The value of the deep neural network in the immunotherapy curative effect and prognosis prediction of EGFR-TKI drug-resistant NSCLC patients is clarified, a prediction system of the deep neural network based on image histology characteristics is trained and verified, the conversion of the deep neural network from theory to practice is realized, the individuation accurate treatment is guided, and the potential molecular basis of the curative effect and prognosis prediction is discussed.
Collecting clinical pathology, hematologic inflammatory factors and image histology characteristics of EGFR-TKI resistant NSCLC patients receiving immunotherapy to construct Gao Weite collection, and establishing a low-dimensional characteristic subset after characteristic screening. And (3) training an optimal deep neural network by applying an activation function, initializing weights and adjusting various super parameters, constructing a deep surviving prediction system and an N-MTLR survival prediction system, evaluating the advantages of the deep neural network compared with CPH and machine learning prognosis models, and verifying the external applicability of the deep neural network in a prospective data set. Through genome DNA methylation sequencing difference analysis and functional enrichment analysis, a potential molecular biology foundation for deep neural network prognosis prediction is excavated.
(3) Screening CT image histology characteristics related to the immune treatment efficacy and prognosis of EGFR-TKI drug-resistant NSCLC patients, and constructing a multidimensional efficacy and prognosis model based on the image histology characteristics to guide individual accurate treatment.
Patients with EGFR-TKI resistant NSCLC receiving immunotherapy are divided into an effective group and an ineffective group according to the curative effect of the therapy, and clinical pathological features, hematological inflammatory factors and CT imaging histology features of the patients are collected to construct a high-dimensional feature subset. And establishing a low-dimensional feature subset after feature engineering screening. And constructing an integrated multidimensional prognosis model, and comparing the performances of the deep neural network, the CPH and the machine learning model.
Further, referring to fig. 3, the matching of data is performed according to the drug-resistant image characteristics of the target disease drug-resistant patient sample and the image histology reference database, the target disease drug-resistant patient sample is divided into an effective group and an ineffective group, and the corresponding immune crowd determination weight is set, which specifically includes:
dividing focus ROI (region of interest) areas of the drug-resistant image features to obtain areas to be extracted;
extracting effective features from the region to be extracted, and screening labels;
matching the extracted effective features and the screened labels with treatment change data and physiological attribute data of the historical patients in the image histology reference database;
dividing the corresponding matched target disease drug-resistant patients into an effective group and an ineffective group according to the treatment efficacy of the historical patients, and setting the immune crowd determination weights of the effective group and the ineffective group.
Wherein the drug resistant image in this embodiment is characterized as a CT image;
the CT image contains a large amount of digital information, but the conventional method is difficult to extract the image tag related to the immune therapeutic effect of EGFR-TKI resistant NSCLC patients, and the extraction of the image group tag related to the gene state and the high immune therapeutic prognosis is a challenging problem, and is one of the key technical problems to be solved by the invention. The invention aims to study an automatic focus segmentation method, a massive feature extraction method and a feature screening method, so as to realize the extraction of quantitative features and the construction of key image histology labels.
CT images can reflect the genes and the response to immunotherapy of NSCLC patients to a certain extent, but besides image information, clinical information such as demographic information, smoking history, family history and the like, gene mutation information (19-exon loss or 21-exon replacement, or other EGFR mutation types and the like) also proves to have a certain correlation with prognosis, and how to fuse the multisource information to construct an individualized prediction model is another key problem to be solved by the invention. The invention aims to utilize a multivariate Cox regression analysis method and Nomogram (Nomogram), fuse image histology tags, gene mutation and clinical information, construct a gene state and immunotherapy efficacy prediction model and realize individualized efficacy prediction.
Further, referring to fig. 4, the step of obtaining the image detection feature of the patient with the target disease to be determined, performing similarity calculation on the image detection feature and the drug-resistant image feature, and predicting the dominant population for immunotherapy of the target disease based on the multidimensional deep learning model of image histology specifically includes:
performing similarity calculation on the image detection features and the drug-resistant image features to obtain a reference target disease drug-resistant patient with similarity meeting a preset threshold;
Fusing the image detection characteristics to construct an individuation prediction model, and constructing a gene state and immunotherapy curative effect prediction model by fusing image histology tags, gene mutation and clinical information by using a multivariate Cox regression analysis method and Nomogram;
inputting the reference target disease drug-resistant patient into the gene state and immunotherapy curative effect prediction model to obtain the determining weight of the immune crowd to be determined of the target disease patient to be determined;
and selecting the patient with the target disease to be determined, the weight of which is greater than a preset weight value, as the immunotherapeutic dominant population of the target disease.
Wherein in this embodiment, it comprises:
the method comprises the steps of importing reinforced CT images before treatment, and using IBEX software under a MATLAB platform to delineate a region of interest (ROI) and extract massive image histology features from the region of interest. Two radiotherapeutic doctors respectively delineate the ROI, one doctor delineates the ROI twice, the ROI is a primary tumor focus in the lung, the lung boundary of the tumor is delineated under a lung window, and the boundary of tumor tissue and the chest wall is delineated under a mediastinal window. And (3) exporting the CT image and the tumor ROI region, performing image preprocessing by adopting a method of interpolation resampling, butterworth smoothing or LG filtering conversion, and then extracting the image histology characteristics in a high flux manner. The extracted image group is characterized by comprising 1041 main parameters, namely 9 main parameters, namely a gradient direction histogram, a two-dimensional gray level co-occurrence matrix, a three-dimensional gray level co-occurrence matrix (GLCM 3), a two-dimensional neighborhood gray level difference, a three-dimensional neighborhood gray level difference (NID 3), a gray level run-in matrix, intensity, an intensity histogram and shape. And calculating intra-group correlation coefficients according to the ROIs sketched by the same doctor twice, calculating inter-group correlation coefficients ICC according to the image histology characteristics extracted by the ROIs sketched by the two doctors, and screening parameters with ICC >0.75 as stable image histology characteristics to be incorporated into further characteristic screening.
The screening-stable image histology features were Z-Score normalized to ensure comparability between data. And in the training set patients, adopting a minimum absolute value convergence and selection operator LASSO algorithm to carry out screening of image histology feature dimension reduction, curative effect and prognosis related features. LASSO regression is a statistical method capable of realizing continuous stability of feature screening and estimation at the same time and converting high-dimensional data into low-dimensional data by constructing a penalty function to compress some regression coefficients to be forcedly changed, thereby achieving the purposes of feature sparsification and screening. The stable image histology features are included in LASSO-Cox analysis and 10-fold cross validation is performed, and the optimal input value is determined according to the minimum value standard. And screening out the image histology parameters and the weights thereof with non-0 coefficients according to the model corresponding to the optimal input value. And then, performing association analysis on the prognosis related features obtained by screening the LASSO-Cox algorithm by calculating the Pearson related coefficient to remove redundancy, removing the related parameters (the related coefficient is more than 0.5), and finally screening out independent stable curative effect and prognosis related image histology features.
And calculating a determined value according to the weight coefficient of each parameter in LASSO-Cox regression by using the independent stable curative effect and prognosis related image histology characteristics obtained by screening, and finally obtaining an image histology tag (Radscore) of each patient. The predicted performance is referenced by calculating the Radscore predicted PFS C-index and plotting a subject work characteristic (ROC) to calculate the area under the curve (AUC). C-index is the ratio of the predicted survival and actual survival results to the same pair in all patient pairs in the data set. According to the image histology characteristics of the Radscore in the training set and the weight coefficients thereof, the Radscore of each patient in the verification set is calculated, the C-index and the AUC of the patient PFS of the Radscore prediction verification set are calculated, and the external applicability of the Radscore is defined. And Radscore is taken as an independent parameter, is taken into multi-factor Cox regression survival analysis, corrects the influence of clinical pathological characteristics on prognosis, and further verifies the independent prediction efficacy of Radscore.
According to a second embodiment of the present invention, referring to fig. 5, the present invention claims an immune crowd determining system based on multidimensional analysis, comprising:
the database building module is used for obtaining a historical patient of a target disease, extracting treatment change data and physiological attribute data of the historical patient and building an image histology reference database;
the drug resistance analysis module is used for inputting a target disease inhibitor to a training patient sample with a target disease to obtain a target disease drug resistance patient sample and analyzing the drug resistance image characteristics of the target disease drug resistance patient sample;
the weight determining module is used for carrying out data matching on the drug-resistant image characteristics of the target disease drug-resistant patient sample and the image histology reference database, dividing the target disease drug-resistant patient sample into an effective group and an ineffective group, and setting corresponding immune crowd determining weights;
the crowd determination module is used for obtaining image detection characteristics of a patient with a target disease to be determined, carrying out similarity calculation on the image detection characteristics and drug-resistant image characteristics, and predicting immune treatment dominant crowd of the target disease based on a multidimensional deep learning model of image histology.
Further, the database creation module specifically includes:
Acquiring historical patients of target diseases, extracting historical patients of the target diseases with similar treatment paths from the acquired historical patients of the target diseases, and taking the historical patients of the target diseases with similar treatment paths as an initial cluster;
querying treatment change data corresponding to historical patients of the target disease in the initial cluster;
acquiring a preset treatment change rate, and acquiring a historical patient of the target disease, in which the treatment change data is within the preset treatment change rate, from the initial cluster as a historical patient of the target disease;
extracting historical patients of the target diseases from the initial clusters to obtain target clusters, and establishing an image histology reference database according to the target clusters;
acquiring treatment change data corresponding to historical patients of the target disease in the target cluster in the image histology reference database;
calculating the average dressing change frequency and the physiological attribute change value of the treatment change data corresponding to the historical patient of the target disease;
obtaining immune reference values of the target clusters in the image histology reference database according to the average dressing change frequency and the physiological attribute change value;
Obtaining a target immune reference value, comparing the immune reference value with the target immune reference value to obtain whether the target cluster is credible or not, and outputting a judgment result of whether the target cluster is credible or not.
Further, the drug resistance analysis module specifically includes:
the disease of interest is EGFR mutated NSCLC;
the target disease inhibitor is an EGFR-TKI inhibitor;
and the characteristic content of the drug-resistant image characteristic is the same as the data content of the image histology reference database.
Further, the weight determining module specifically includes:
dividing focus ROI (region of interest) areas of the drug-resistant image features to obtain areas to be extracted;
extracting effective features from the region to be extracted, and screening labels;
matching the extracted effective features and the screened labels with treatment change data and physiological attribute data of the historical patients in the image histology reference database;
dividing the corresponding matched target disease drug-resistant patients into an effective group and an ineffective group according to the treatment efficacy of the historical patients, and setting the immune crowd determination weights of the effective group and the ineffective group.
Further, the crowd determination module specifically includes:
Performing similarity calculation on the image detection features and the drug-resistant image features to obtain a reference target disease drug-resistant patient with similarity meeting a preset threshold;
fusing the image detection characteristics to construct an individuation prediction model, and constructing a gene state and immunotherapy curative effect prediction model by fusing image histology tags, gene mutation and clinical information by using a multivariate Cox regression analysis method and Nomogram;
inputting the reference target disease drug-resistant patient into the gene state and immunotherapy curative effect prediction model to obtain the determining weight of the immune crowd to be determined of the target disease patient to be determined;
and selecting the patient with the target disease to be determined, the weight of which is greater than a preset weight value, as the immunotherapeutic dominant population of the target disease.
Those skilled in the art will appreciate that various modifications and improvements can be made to the disclosure. For example, the various devices or components described above may be implemented in hardware, or may be implemented in software, firmware, or a combination of some or all of the three.
A flowchart is used in this disclosure to describe the steps of a method according to an embodiment of the present disclosure. It should be understood that the steps that follow or before do not have to be performed in exact order. Rather, the various steps may be processed in reverse order or simultaneously. Also, other operations may be added to these processes.
Those of ordinary skill in the art will appreciate that all or a portion of the steps of the methods described above may be implemented by a computer program to instruct related hardware, and the program may be stored in a computer readable storage medium, such as a read only memory, a magnetic disk, or an optical disk. Alternatively, all or part of the steps of the above embodiments may be implemented using one or more integrated circuits. Accordingly, each module/unit in the above embodiment may be implemented in the form of hardware, or may be implemented in the form of a software functional module. The present disclosure is not limited to any specific form of combination of hardware and software.
Unless defined otherwise, all terms used herein have similar meanings as commonly understood by one of ordinary skill in the art to which this disclosure pertains. It will be further understood that terms, such as those defined in commonly used dictionaries, should be interpreted as having a meaning that is consistent with their meaning in the context of the relevant art and will not be interpreted in an idealized or overly formal sense unless expressly so defined herein.
The foregoing is illustrative of the present disclosure and is not to be construed as limiting thereof. Although a few exemplary embodiments of this disclosure have been described, those skilled in the art will readily appreciate that many modifications are possible in the exemplary embodiments without materially departing from the novel teachings and advantages of this disclosure. Accordingly, all such modifications are intended to be included within the scope of this disclosure as defined in the claims. It is to be understood that the foregoing is illustrative of the present disclosure and is not to be construed as limited to the specific embodiments disclosed, and that modifications to the disclosed embodiments, as well as other embodiments, are intended to be included within the scope of the appended claims. The disclosure is defined by the claims and their equivalents.
In the description of the present specification, reference to the terms "one embodiment," "some embodiments," "illustrative embodiments," "examples," "specific examples," or "some examples," etc., means that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the invention. In this specification, schematic representations of the above terms do not necessarily refer to similar embodiments or examples. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.
While embodiments of the present invention have been shown and described, it will be understood by those of ordinary skill in the art that: many changes, modifications, substitutions and variations may be made to the embodiments without departing from the spirit and principles of the invention, the scope of which is defined by the claims and their equivalents.
Claims (10)
1. An immune crowd determination method based on multidimensional analysis, comprising the steps of:
acquiring a historical patient of a target disease, extracting treatment change data and physiological attribute data of the historical patient, and constructing an image histology reference database;
Inputting a target disease inhibitor into a training patient sample with a target disease to obtain a target disease drug-resistant patient sample and analyzing the drug-resistant image characteristics of the target disease drug-resistant patient sample;
according to the drug-resistant image characteristics of the target disease drug-resistant patient sample and the image histology reference database, performing data matching, dividing the target disease drug-resistant patient sample into an effective group and an ineffective group, and setting corresponding immune crowd determination weights;
and obtaining image detection characteristics of a patient with the target disease to be determined, carrying out similarity calculation on the image detection characteristics and drug-resistant image characteristics, and predicting immune treatment dominant population of the target disease based on a multidimensional deep learning model of image histology.
2. The method for determining immune crowd based on multidimensional analysis according to claim 1, wherein the step of obtaining the historic patient of the target disease, extracting treatment change data and physiological attribute data of the historic patient, and constructing an image histology reference database comprises the following steps:
acquiring historical patients of target diseases, extracting historical patients of the target diseases with similar treatment paths from the acquired historical patients of the target diseases, and taking the historical patients of the target diseases with similar treatment paths as an initial cluster;
Querying treatment change data corresponding to historical patients of the target disease in the initial cluster;
acquiring a preset treatment change rate, and acquiring a historical patient of the target disease, in which the treatment change data is within the preset treatment change rate, from the initial cluster as a historical patient of the target disease;
extracting historical patients of the target diseases from the initial clusters to obtain target clusters, and establishing an image histology reference database according to the target clusters;
acquiring treatment change data corresponding to historical patients of the target disease in the target cluster in the image histology reference database;
calculating the average dressing change frequency and the physiological attribute change value of the treatment change data corresponding to the historical patient of the target disease;
obtaining immune reference values of the target clusters in the image histology reference database according to the average dressing change frequency and the physiological attribute change value;
obtaining a target immune reference value, comparing the immune reference value with the target immune reference value to obtain whether the target cluster is credible or not, and outputting a judgment result of whether the target cluster is credible or not.
3. The method for determining immune population based on multidimensional analysis according to claim 1, wherein the step of inputting a target disease inhibitor into a training patient sample having a target disease to obtain a target disease drug resistant patient sample and analyzing drug resistant image characteristics of the target disease drug resistant patient sample comprises the following steps:
The disease of interest is EGFR mutated NSCLC;
the target disease inhibitor is an EGFR-TKI inhibitor;
and the characteristic content of the drug-resistant image characteristic is the same as the data content of the image histology reference database.
4. The method for determining immune crowd based on multidimensional analysis according to claim 1, wherein the matching of the drug-resistant image characteristics of the target disease drug-resistant patient sample with the image histology reference database, dividing the target disease drug-resistant patient sample into an effective group and an ineffective group, and setting corresponding immune crowd determination weights, comprises:
dividing focus ROI (region of interest) areas of the drug-resistant image features to obtain areas to be extracted;
extracting effective features from the region to be extracted, and screening labels;
matching the extracted effective features and the screened labels with treatment change data and physiological attribute data of the historical patients in the image histology reference database;
dividing the corresponding matched target disease drug-resistant patients into an effective group and an ineffective group according to the treatment efficacy of the historical patients, and setting the immune crowd determination weights of the effective group and the ineffective group.
5. The method for determining immune crowd based on multidimensional analysis according to claim 1, wherein the steps of obtaining image detection characteristics of a patient with a target disease to be determined, performing similarity calculation on the image detection characteristics and drug-resistant image characteristics, and predicting immune treatment dominant crowd of the target disease based on a multidimensional deep learning model of image histology comprise the following steps:
performing similarity calculation on the image detection features and the drug-resistant image features to obtain a reference target disease drug-resistant patient with similarity meeting a preset threshold;
fusing the image detection characteristics to construct an individuation prediction model, and constructing a gene state and immunotherapy curative effect prediction model by fusing image histology tags, gene mutation and clinical information by using a multivariate Cox regression analysis method and Nomogram;
inputting the reference target disease drug-resistant patient into the gene state and immunotherapy curative effect prediction model to obtain the determining weight of the immune crowd to be determined of the target disease patient to be determined;
and selecting the patient with the target disease to be determined, the weight of which is greater than a preset weight value, as the immunotherapeutic dominant population of the target disease.
6. An immune crowd determination system based on multidimensional analysis, comprising:
the database building module is used for obtaining a historical patient of a target disease, extracting treatment change data and physiological attribute data of the historical patient and building an image histology reference database;
the drug resistance analysis module is used for inputting a target disease inhibitor to a training patient sample with a target disease to obtain a target disease drug resistance patient sample and analyzing the drug resistance image characteristics of the target disease drug resistance patient sample;
the weight determining module is used for carrying out data matching on the drug-resistant image characteristics of the target disease drug-resistant patient sample and the image histology reference database, dividing the target disease drug-resistant patient sample into an effective group and an ineffective group, and setting corresponding immune crowd determining weights;
the crowd determination module is used for obtaining image detection characteristics of a patient with a target disease to be determined, carrying out similarity calculation on the image detection characteristics and drug-resistant image characteristics, and predicting immune treatment dominant crowd of the target disease based on a multidimensional deep learning model of image histology.
7. The immune crowd determination system based on multidimensional analysis of claim 6, wherein the pooling module specifically comprises:
Acquiring historical patients of target diseases, extracting historical patients of the target diseases with similar treatment paths from the acquired historical patients of the target diseases, and taking the historical patients of the target diseases with similar treatment paths as an initial cluster;
querying treatment change data corresponding to historical patients of the target disease in the initial cluster;
acquiring a preset treatment change rate, and acquiring a historical patient of the target disease, in which the treatment change data is within the preset treatment change rate, from the initial cluster as a historical patient of the target disease;
extracting historical patients of the target diseases from the initial clusters to obtain target clusters, and establishing an image histology reference database according to the target clusters;
acquiring treatment change data corresponding to historical patients of the target disease in the target cluster in the image histology reference database;
calculating the average dressing change frequency and the physiological attribute change value of the treatment change data corresponding to the historical patient of the target disease;
obtaining immune reference values of the target clusters in the image histology reference database according to the average dressing change frequency and the physiological attribute change value;
Obtaining a target immune reference value, comparing the immune reference value with the target immune reference value to obtain whether the target cluster is credible or not, and outputting a judgment result of whether the target cluster is credible or not.
8. The immune crowd determination system based on multidimensional analysis of claim 7, wherein the drug resistance analysis module specifically comprises:
the disease of interest is EGFR mutated NSCLC;
the target disease inhibitor is an EGFR-TKI inhibitor;
and the characteristic content of the drug-resistant image characteristic is the same as the data content of the image histology reference database.
9. The immune crowd determination system based on multidimensional analysis of claim 6, wherein the weight determination module specifically comprises:
dividing focus ROI (region of interest) areas of the drug-resistant image features to obtain areas to be extracted;
extracting effective features from the region to be extracted, and screening labels;
matching the extracted effective features and the screened labels with treatment change data and physiological attribute data of the historical patients in the image histology reference database;
dividing the corresponding matched target disease drug-resistant patients into an effective group and an ineffective group according to the treatment efficacy of the historical patients, and setting the immune crowd determination weights of the effective group and the ineffective group.
10. The immune crowd determination system based on multidimensional analysis of claim 7, wherein the crowd determination module specifically comprises:
performing similarity calculation on the image detection features and the drug-resistant image features to obtain a reference target disease drug-resistant patient with similarity meeting a preset threshold;
fusing the image detection characteristics to construct an individuation prediction model, and constructing a gene state and immunotherapy curative effect prediction model by fusing image histology tags, gene mutation and clinical information by using a multivariate Cox regression analysis method and Nomogram;
inputting the reference target disease drug-resistant patient into the gene state and immunotherapy curative effect prediction model to obtain the determining weight of the immune crowd to be determined of the target disease patient to be determined;
and selecting the patient with the target disease to be determined, the weight of which is greater than a preset weight value, as the immunotherapeutic dominant population of the target disease.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311257835.0A CN116994770B (en) | 2023-09-27 | 2023-09-27 | Immune crowd determination method and system based on multidimensional analysis |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311257835.0A CN116994770B (en) | 2023-09-27 | 2023-09-27 | Immune crowd determination method and system based on multidimensional analysis |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116994770A true CN116994770A (en) | 2023-11-03 |
| CN116994770B CN116994770B (en) | 2024-01-02 |
Family
ID=88527058
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311257835.0A Active CN116994770B (en) | 2023-09-27 | 2023-09-27 | Immune crowd determination method and system based on multidimensional analysis |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116994770B (en) |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113893342A (en) * | 2021-10-11 | 2022-01-07 | 蒋涛 | Application of pharmaceutical composition containing anti-PD-1 antibody in preparation of medicine for treating advanced non-small cell lung cancer |
| CN114462522A (en) * | 2022-01-26 | 2022-05-10 | 清华珠三角研究院 | Lung cancer life cycle prediction model training and prediction method, system, device and medium |
| WO2022170557A1 (en) * | 2021-02-10 | 2022-08-18 | Gnt Biotech & Medicals Corporation | A pharmaceutical combination and method for overcoming immune suppression or stimulating immune response against cancer |
| CN115274119A (en) * | 2022-09-30 | 2022-11-01 | 中国医学科学院北京协和医院 | Construction method of immunotherapy prediction model fusing multi-image mathematical characteristics |
| CN115440383A (en) * | 2022-09-30 | 2022-12-06 | 中国医学科学院北京协和医院 | System for predicting curative effect of PD-1/PD-L1 monoclonal antibody of advanced cancer patient |
| CN115497623A (en) * | 2022-09-27 | 2022-12-20 | 山东第一医科大学(山东省医学科学院) | Lung cancer prognosis prediction system based on image, pathology and gene multiomics |
| CN116230237A (en) * | 2023-05-06 | 2023-06-06 | 四川省医学科学院·四川省人民医院 | Lung cancer influence evaluation method and system based on ROI focus features |
-
2023
- 2023-09-27 CN CN202311257835.0A patent/CN116994770B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2022170557A1 (en) * | 2021-02-10 | 2022-08-18 | Gnt Biotech & Medicals Corporation | A pharmaceutical combination and method for overcoming immune suppression or stimulating immune response against cancer |
| CN113893342A (en) * | 2021-10-11 | 2022-01-07 | 蒋涛 | Application of pharmaceutical composition containing anti-PD-1 antibody in preparation of medicine for treating advanced non-small cell lung cancer |
| CN114462522A (en) * | 2022-01-26 | 2022-05-10 | 清华珠三角研究院 | Lung cancer life cycle prediction model training and prediction method, system, device and medium |
| CN115497623A (en) * | 2022-09-27 | 2022-12-20 | 山东第一医科大学(山东省医学科学院) | Lung cancer prognosis prediction system based on image, pathology and gene multiomics |
| CN115274119A (en) * | 2022-09-30 | 2022-11-01 | 中国医学科学院北京协和医院 | Construction method of immunotherapy prediction model fusing multi-image mathematical characteristics |
| CN115440383A (en) * | 2022-09-30 | 2022-12-06 | 中国医学科学院北京协和医院 | System for predicting curative effect of PD-1/PD-L1 monoclonal antibody of advanced cancer patient |
| CN116230237A (en) * | 2023-05-06 | 2023-06-06 | 四川省医学科学院·四川省人民医院 | Lung cancer influence evaluation method and system based on ROI focus features |
Non-Patent Citations (5)
| Title |
|---|
| BAI YI- FENG等: "Randomized Trial of First-Line Tyrosine Kinase Inhibitor With or Without Radiotherapy for Synchronous Oligometastatic EGFR-Mutated Non-Small Cell Lung Cancer", 《RANDOMIZED TRIAL OF FIRST-LINE TYROSINE KINASE INHIBITOR WITH OR WITHOUT RADIOTHERAPY FOR SYNCHRONOUS OLIGOMETASTATIC EGFR-MUTATED NON-SMALL CELL LUNG CANCER》, vol. 115, no. 06, pages 742 * |
| JIANGDIAN SONG等: "A new approach to predict progression-free survival in stage IV EGFR-mutant NSCLC patients with EGFR-TKI therapy", 《CLINICAL CANCER RESEARCH》, vol. 24, no. 15, pages 3583 - 3592 * |
| 宋江典: "基于定量化CT影像特征的非小细胞肺癌预后分析研究", 《中国博士学位论文全文数据库 医药卫生科技辑》, no. 06, pages 060 - 18 * |
| 白义凤等: "非小细胞肺癌组织LncRNA RP11-164P12.4表达及其临床意义", 《中华肿瘤防治杂志》, vol. 24, no. 16, pages 1148 - 1152 * |
| 赵倩: "临床参数联合影像组学预测非小细胞肺癌免疫治疗疗效的研究", 《中国优秀硕士学位论文全文数据库 医药卫生科技辑》, no. 12, pages 072 - 105 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116994770B (en) | 2024-01-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112768072B (en) | Cancer clinical index evaluation system constructed based on imaging omics qualitative algorithm | |
| EP2391887B1 (en) | Image-based risk score-a prognostic predictor of survival and outcome from digital histopathology | |
| Silva et al. | EGFR assessment in lung cancer CT images: analysis of local and holistic regions of interest using deep unsupervised transfer learning | |
| Wankhede et al. | Dynamic architecture based deep learning approach for glioblastoma brain tumor survival prediction | |
| GB2605473A (en) | Logistic model to determine 3D z-wise lesion connectivity | |
| Li et al. | A novel radiogenomics framework for genomic and image feature correlation using deep learning | |
| Meng et al. | Artificial intelligence-based radiomics in bone tumors: Technical advances and clinical application | |
| CN114565786A (en) | Tomography image classification device and method based on channel attention mechanism | |
| CN111027548A (en) | Brain glioma molecular marker nondestructive prediction method and system based on image omics | |
| US20210287801A1 (en) | Method for predicting disease state, therapeutic response, and outcomes by spatial biomarkers | |
| Lu et al. | Machine learning-based radiomics for prediction of epidermal growth factor receptor mutations in lung adenocarcinoma | |
| Pham et al. | Identifying an optimal machine learning generated image marker to predict survival of gastric cancer patients | |
| Christie et al. | Predicting recurrence risks in lung cancer patients using multimodal radiomics and random survival forests | |
| CN116230237B (en) | Lung cancer influence evaluation method and system based on ROI focus features | |
| Ma et al. | Automatic pulmonary ground‐glass opacity nodules detection and classification based on 3D neural network | |
| CN116994770B (en) | Immune crowd determination method and system based on multidimensional analysis | |
| CN116504406A (en) | Method and system for constructing lung cancer postoperative risk model based on image combination pathology | |
| Cheng et al. | Prediction of egfr mutation status in lung adenocarcinoma using multi-source feature representations | |
| Levenson et al. | Advancing precision medicine: algebraic topology and differential geometry in radiology and computational pathology | |
| CN115881312A (en) | Method and system for predicting prognosis of colorectal cancer in stage II, intelligent terminal and computer readable storage medium | |
| Pradhan | An early diagnosis of lung nodule using CT images based on hybrid machine learning techniques | |
| Gogi et al. | Review of Machine Learning Methods for the Survey on HCC Scenario and Prediction Strategy | |
| Sheng et al. | Modeling nodule growth via spatial transformation for follow-up prediction and diagnosis | |
| Li et al. | A 3D lung lesion variational autoencoder | |
| Mohamed Aarif et al. | Deep MammoNet: Early Diagnosis of Breast Cancer Using Multi-layer Hierarchical Features of Deep Transfer Learned Convolutional Neural Network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |