CN114529541A

CN114529541A - Methods, electronic devices, media, and systems for predicting cell expression in pancreatic cancer microenvironment

Info

Publication number: CN114529541A
Application number: CN202210309777.0A
Authority: CN
Inventors: 卢明智; 边云; 邵成伟; 刘芳; 方旭; 李晶; 王铁功
Original assignee: First Affiliated Hospital of Naval Military Medical University of PLA
Current assignee: First Affiliated Hospital of Naval Military Medical University of PLA
Priority date: 2022-03-28
Filing date: 2022-03-28
Publication date: 2022-05-24

Abstract

The application provides a method, electronic device, medium and system for predicting cell expression in a pancreatic cancer microenvironment, the method comprising obtaining a CT image of a pancreas and labeling a tumor region; extracting the image omics characteristics of the tumor area to obtain an omics parameter data set, carrying out characteristic screening on the tumor omics parameter data set, and selecting target characteristics; obtaining a cell immunohistochemical staining section corresponding to the CT image, and grouping the cell immunohistochemical staining section, wherein the grouping result comprises a high expression group and a low expression group; training an XGboost classifier by using the target features and the grouping result to obtain a prediction model; outputting the cell expression according to the model. The method can non-invasively and efficiently predict cell expression in a pancreatic cancer microenvironment.

Description

Methods, electronic devices, media, and systems for predicting cell expression in pancreatic cancer microenvironment

Technical Field

The invention relates to the field of medical image processing, in particular to a method, electronic equipment, medium and system for predicting cell expression in a pancreatic cancer microenvironment.

Background

Pancreatic cancer is a common malignant tumor, and the main means for treating cancer include surgery, radiotherapy, chemotherapy and immunotherapy. Patients undergo radical surgery in the early stages of pancreatic cancer, but most cases develop relapses and metastases. In addition, pancreatic cancer is less sensitive to radiation and chemotherapy.

In the microenvironment of pancreatic cancer, it is a complex system of homeostasis consisting of tumor cells, infiltrating lymphocytes, fibroblasts, many cytokines and catalytic factors. Different cellular expressions in the pancreatic cancer microenvironment affect the patient's disease progression, prognosis, survival and efficacy. In the related art, the evaluation of cell expression in the microenvironment of pancreatic cancer is based on post-operative pathological specimens, which is invasive and time consuming.

Disclosure of Invention

The application provides a method, electronic equipment, medium and system for predicting cell expression in a pancreatic cancer microenvironment, which can non-invasively and efficiently predict cell expression in a pancreatic cancer microenvironment.

A first aspect of the application discloses a method of predicting cell expression in a pancreatic cancer microenvironment, the method comprising, obtaining a CT image of a pancreas and labeling a tumor region; extracting the image omics characteristics of the tumor area to obtain an omics parameter data set, carrying out characteristic screening on the tumor omics parameter data set, and selecting target characteristics; obtaining a cell immunohistochemical staining section corresponding to the CT image, and grouping the cell immunohistochemical staining section, wherein the grouping result comprises a high expression group and a low expression group; training an XGboost classifier by using the target features and the grouping result to obtain a prediction model; outputting the cell expression according to the model.

In one possible implementation of the first aspect, performing feature screening on the tumor mass study parameter dataset includes selecting the target feature using analysis of variance and Spearman correlation.

In one possible implementation of the first aspect, the feature screening the lumomics parameter dataset further comprises selecting the target feature using Lasso regression analysis.

In one possible implementation of the first aspect above, the cells are tumor infiltrating lymphocytes, including CD4+ T, CD8+ T and CD +20B cells.

In one possible implementation of the first aspect described above, the cell is Fibroblast Activation Protein (FAP).

In one possible implementation of the first aspect, the grouping the immunohistochemical stained sections comprises scoring the tumor-infiltrating lymphocytes, the scoring using survival as a predictor variable, and positive expression of the CD4+ T, CD8+ T, and CD +20B cells as a percentage of intratumor as an independent variable, establishing a COX regression model, and calculating the tumor-infiltrating lymphocyte score according to the COX regression model; and dividing the cellular immunohistochemical staining section into the high expression group or the low expression group according to the tumor infiltrating lymphocyte score by using X-tile software and taking survival as a result.

In one possible implementation of the first aspect, grouping the cytoimmunohistochemical stained sections comprises using X-tile software to classify the cytoimmunohistochemical stained sections into the high expression group or the low expression group according to the percentage of FAP in the tumor as a result of survival.

In one possible implementation of the first aspect, the CT image includes an arterial phase image and a portal phase image, and 1409 cine features are extracted from both the arterial phase image and the portal phase image to form the omic parameter dataset.

In a possible implementation of the first aspect, the method further includes dividing the CT image into a training set and a validation set, where the training set is used for constructing the prediction model, and the validation set is used for validating the prediction model.

A second aspect of the application discloses an electronic device comprising a memory storing computer executable instructions and a processor; the instructions, when executed by the processor, cause the apparatus to carry out a method of predicting cell expression in a pancreatic cancer microenvironment according to the first aspect of the application.

A third aspect of the present application discloses a computer readable medium storing one or more programs, the one or more programs being executable by one or more processors to implement the method of predicting cell expression in a pancreatic cancer microenvironment of the first aspect of the present application.

A fourth aspect of the present application discloses a system for predicting cell expression in a pancreatic cancer microenvironment, the system comprising, an acquisition module for acquiring CT images of the pancreas and labeling tumor regions; the extraction module is used for extracting the image omics characteristics of the tumor region to obtain an omics parameter data set, and performing characteristic screening on the tumor omics parameter data set to select target characteristics; the grouping module is used for acquiring a cell immunohistochemical staining section corresponding to the CT image and grouping the cell immunohistochemical staining section, wherein the grouping result comprises a high expression group and a low expression group; the training module is used for training an XGboost classifier by using the target characteristics and the grouping result to obtain a prediction model; a prediction module to output the cell expression according to the model.

A fifth aspect of the application discloses a computer program product which, when executed by a processor, implements the method of predicting cell expression in a pancreatic cancer microenvironment of the first aspect of the application.

According to the method and the device for predicting the cell expression in the pancreatic cancer microenvironment, the cell immunohistochemical staining sections corresponding to the CT images are grouped according to high expression and low expression, the CT images are subjected to image omics characteristic extraction to obtain target characteristics, the XGboost classifier is trained according to the grouping results and the target characteristics to form a prediction model, the trained model can be used for predicting the cell expression in the microenvironment corresponding to the pancreatic cancer CT images, and patients effective for subsequent treatment can be screened efficiently.

Drawings

FIG. 1 is a schematic flow chart of a method of predicting tumor-infiltrating lymphocyte expression in pancreatic cancer, according to one embodiment of the present application;

FIG. 2 is a schematic diagram of a method 200 for predicting tumor-infiltrating immune cell expression, in accordance with an embodiment of the present application;

FIG. 3a is a graph of survival-based patient cohort and tumor-infiltrating lymphocyte cutoff using X-tile software in an embodiment of the present application;

FIG. 3b is a graph showing the difference in survival between the high and low tumor infiltrating lymphocyte score groups according to one embodiment of the present application;

figure 4 bar graph of omics feature weights in a pancreatic cancer tumor-infiltrating lymphocyte expression prediction model in one embodiment of the present application;

FIG. 5 is a schematic representation of ROC curves for a model for predicting pancreatic cancer tumor-infiltrating lymphocyte expression in a training set and a validation set, in accordance with an embodiment of the present application;

FIG. 6 is a schematic illustration of the prediction of pancreatic cancer tumor infiltrating lymphocyte expression and corresponding imaging of an embodiment of the present application;

figure 7 bar graph of omics feature weights in a pancreatic cancer fibroblast activation protein prediction model in one embodiment of the present application;

FIG. 8 is a schematic representation of the prediction of pancreatic cancer fibroblast activation protein expression and corresponding imaging of one embodiment of the present application;

FIG. 9 is a schematic diagram of a system 900 for predicting cell expression in a pancreatic cancer microenvironment, according to one embodiment of the present application;

fig. 10 is a block diagram of an electronic device 1000 of an embodiment of the present application.

Detailed Description

The present application is further described with reference to the following detailed description and the accompanying drawings. It is to be understood that the illustrative embodiments of the present disclosure include, but are not limited to, methods, electronic devices, media and systems for predicting cell expression in a pancreatic cancer microenvironment, and that the specific embodiments described herein are intended to be illustrative only and are not intended to be limiting. In addition, for convenience of description, only a part of structures or processes related to the present application, not all of them, is illustrated in the drawings.

The following description of the embodiments of the present application is provided by way of specific examples, and other advantages and effects of the present application will be readily apparent to those skilled in the art from the disclosure herein. While the description of the present application will be described in conjunction with the preferred embodiments, it is not intended to limit the features of the present invention to that embodiment. Rather, the invention has been described in connection with embodiments for the purpose of covering alternatives and modifications as may be extended based on the claims of the present application. In the following description, numerous specific details are included to provide a thorough understanding of the present application. The present application may be practiced without these particulars. Moreover, some of the specific details have been omitted from the description in order to avoid obscuring or obscuring the focus of the present application. It should be noted that the embodiments and features of the embodiments in the present application may be combined with each other without conflict.

Further, various operations will be described as multiple discrete operations, in a manner that is most helpful in understanding the illustrative embodiments; however, the order of description should not be construed as to imply that these operations are necessarily order dependent. In particular, these operations need not be performed in the order of presentation.

The terms "comprising," "having," and "including" are synonymous, unless the context dictates otherwise. The phrase "A/B" means "A or B". The phrase "A and/or B" means "(A and B) or (A or B)".

As used herein, the term "module" may refer to, be, or include: an Application Specific Integrated Circuit (ASIC), an electronic circuit, a processor (shared, dedicated, or group) and/or memory that executes one or more software or firmware programs, a combinational logic circuit, and/or other suitable components that provide the described functionality.

In some cases, the disclosed embodiments may be implemented in hardware, firmware, software, or any combination thereof. The disclosed embodiments may also be implemented as instructions carried by or stored on one or more transitory or non-transitory machine-readable (e.g., computer-readable) storage media, which may be read and executed by one or more processors. For example, the instructions may be distributed via a network or other computer readable medium. Thus, a machine-readable medium may include any mechanism for storing or transmitting information in a form readable by a machine (e.g., a computer), without limitation, a floppy diskette, optical disk, read-only memory (CD-ROM), magneto-optical disk, read-only memory (ROM), Random Access Memory (RAM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), magnetic or optical card, flash memory, or a tangible machine-readable memory for transmitting information over the Internet via electrical, optical, acoustical or other forms of propagated signals (e.g., carrier waves, infrared signals, digital signals, etc.). Thus, a machine-readable medium includes any type of machine-readable medium suitable for storing or transmitting electronic instructions or information in a form readable by a machine (e.g., a computer).

In the drawings, some features of the structures or methods are shown in a particular arrangement and/or order. However, it is to be understood that such specific arrangement and/or ordering may not be required. In some embodiments, these features may be arranged in a manner and/or order different from that shown in the illustrative figures. Additionally, the inclusion of structural or methodical features in a particular figure is not meant to imply that such features are required in all embodiments, and in some embodiments, these features may not be included or may be combined with other features.

It is to be understood that, although the terms first, second, etc. may be used herein to describe various elements or data, these elements or data should not be limited by these terms. These terms are used merely to distinguish one feature from another. For example, a first feature may be termed a second feature, and, similarly, a second feature may be termed a first feature, without departing from the scope of example embodiments.

It should be noted that in this specification, like reference numerals and letters refer to like items in the following drawings, and thus, once an item is defined in one drawing, it need not be further defined and explained in subsequent drawings.

To make the objects, technical solutions and advantages of the present application more clear, embodiments of the present application will be described in further detail below with reference to the accompanying drawings.

Pancreatic cancer is one of the most common malignant tumors of the digestive system, and is usually treated by means of surgery, radiotherapy, chemotherapy and the like. Of these, although approximately 20% of patients can undergo radical surgery, relapse and metastasis have been found in most cases. In addition, pancreatic cancer is less sensitive to chemotherapy and radiation therapy. Therefore, pancreatic cancer has a 5-year survival rate of less than 8%. The various components of the pancreatic cancer microenvironment, which have a great relevance to the disease progression, prognosis, survival and treatment of patients, are a breakthrough in finding new strategies for treatment. Pancreatic cancer microenvironments include tumor cells, infiltrating lymphocytes, fibroblasts, many cytokines and catalytic factors. The expression of pancreatic cancer microenvironment infiltrating lymphocytes represents the immune state of a patient and is obviously related to the curative effect of immunotherapy of the patient, and the fibroblast activation protein FAP can be used as a new therapeutic target of pancreatic cancer.

Therefore, the method can evaluate the cell expression in the microenvironment of the pancreatic cancer and guide the treatment of the pancreatic cancer, can realize accurate medical treatment and improve the curative effect, and can reduce the waste of medical resources.

The assessment of cellular expression in the microenvironment in the related art typically relies on post-operative pathology examination, i.e., based on post-operative pathology specimens. This method is invasive and not all patients can undergo surgical treatment. The preoperative evaluation can be performed by biopsy, but the evaluation effect of the method is poor due to the factors that the method can obtain a very small amount of tissues, so that the evaluation is inaccurate, puncture complications can be caused, the sample acquisition is influenced by the skill of an operator, and the like.

In order to solve the above problems, an embodiment of the present application provides a method for assessing the microenvironment by preoperatively and non-invasively predicting the cell expression in the microenvironment of pancreatic cancer, so as to efficiently screen out patients effective for subsequent treatment.

Referring to fig. 1, the method 100 includes:

s110, a CT image of the pancreas is acquired and the tumor region is marked.

In the imaging technology, the size, shape, edge and density of the pancreatic mass can be observed by utilizing CT, and whether space-occupying lesion and the blood vessel condition of surrounding tissues exist or not can be checked. For patients diagnosed with pancreatic cancer, surgical resection may be performed and CT images of the pancreas taken preoperatively. In some embodiments, pancreatic cancer can be diagnosed by post-operative pathological findings.

In some embodiments, the acquired CT image is a CT enhanced image. In an enhanced CT scan, after a contrast agent is generally intravenously injected into a blood vessel, the contrast agent flows to various organs and lesions along with the flow of blood. Contrast imaging consists of three phases, the arterial phase, the portal phase, and the delayed phase. The artery phase refers to the arterial vessel filling development phase and can show the characteristics of the contrast agent after flowing into the artery; the portal period refers to the venous vessel filling development period and can show the characteristics of blood passing through the portal vein after the arterial period; the delay phase is a characteristic exhibited after the arterial and portal phases. The arterial phase time after the injection of the contrast agent ranges from 20 to 25 seconds, the portal phase time ranges from 60 to 70 seconds, and the delay phase time ranges from 110 to 130 seconds.

And S120, extracting the image omics characteristics of the tumor region to obtain an omics parameter data set, and performing characteristic screening on the tumor omics parameter data set to select target characteristics.

The image omics features comprise original features and filtering-like features. The filter-like features are based on class 6 filters, and include Laplacian of Gaussian (Laplacian of Gaussian), Wavelet analysis (Wavelet), Square (Square), Square Root (Square Root), log (logarihm) and Exponential (Exponential) features. The specific features included in each class of feature value are First Order statistics (First Order), Shape features (Shape), and texture features.

S130, obtaining a cell immunohistochemical staining section corresponding to the CT image, and grouping the cell immunohistochemical staining section, wherein the grouping result comprises a high expression group and a low expression group.

As described above, for a patient diagnosed with pancreatic cancer and having undergone surgical resection, CT images of the pancreas before surgery are acquired of the patient, and corresponding immunohistochemically stained sections of cells after surgery are acquired, each immunohistochemically stained section being converted into a digital pathology image by a scanner. After delineation of tumor boundaries, some cells in the sections can be quantitatively analyzed using a digital pathology analysis platform. And grouping corresponding cytoimmunohistochemical staining sections according to the quantification result of the cells. The grouping results can be divided into high expression and low expression groups.

S140, training the XGboost classifier by using the target features and the grouping result to obtain a prediction model.

The xgboost (extreme Gradient Boosting) is an algorithm based on GBDT (Gradient Boosting Decision Tree). One application of XGBoost is to perform data feature mining analysis.

And inputting the screened target characteristics into an XGboost classifier for training and screening to obtain the most relevant omics characteristics to form a prediction model.

S150, outputting the cell expression according to the prediction model.

By utilizing the established prediction model, the pancreas CT image can be input into the model to directly obtain the cell expression in the microenvironment corresponding to the pancreas CT image, and the model outputs the high-expression or low-expression result corresponding to the CT image. In some embodiments, the result output by the model is a probability of high expression or a probability of low expression.

The method of predicting cell expression in a pancreatic cancer microenvironment in fig. 1 is for immunohistochemically stained sections of cells corresponding to CT images, grouped according to high and low expression; and performing image omics feature extraction on the CT image to obtain target features, and training an XGboost classifier according to the grouping result and the target features to obtain the image omics features most relevant to a certain component in the microenvironment to form a prediction model. Cell expression in the corresponding microenvironment in the pancreatic CT images can be predicted using the trained model. The method of fig. 1 utilizes iconography to predict the microenvironment, i.e., the microenvironment is assessed by a preoperative and noninvasive method for predicting cell expression in the pancreatic cancer microenvironment, and patients effective for subsequent treatment can be efficiently screened out.

In some embodiments, the CT images include an arterial phase image and a portal phase image, from which 1409 cine features are extracted for each patient, constituting an omic parameter dataset.

In some embodiments, performing feature screening on the lumomics parameter dataset comprises selecting target features using analysis of variance and Spearman correlation. In some embodiments, feature screening for the lumpectomics parameter dataset further comprises selecting the target feature using Lasso (least absolute shrinkage and selection operator) regression analysis.

In some embodiments, the method of fig. 1 can be applied to predict expression of tumor infiltrating lymphocytes in a pancreatic cancer microenvironment. Tumor infiltrating lymphocytes including CD4+ T, CD8+ T and CD +20B cells. Compared with the prediction of single tumor infiltrating lymphocyte, the prediction accuracy can be improved by considering three cells, namely CD4+ T, CD8+ T and CD + 20B. In some examples, grouping the immunohistochemical staining sections includes scoring tumor-infiltrating lymphocytes, the scoring using survival as a predictor variable, positive expression of CD4+ T, CD8+ T, and CD +20B cells as a percentage of intratumor as an independent variable, establishing a COX regression model, and calculating the tumor-infiltrating lymphocyte score from the COX regression model. And then, using X-tile software to survive as a result, and dividing the cell immunohistochemical staining section into a high expression group or a low expression group according to the tumor infiltrating lymphocyte score.

The COX regression Model, also called COX Proportional-risks Model, is a semi-parameter regression Model, simultaneously studies the relationship between a plurality of risk factors and the occurrence condition and the occurrence time of event outcome, simultaneously evaluates the influence of several factors on survival, and can overcome the defect of single factor limitation in simple survival analysis.

In some embodiments, the method of fig. 1 can be applied to predict expression of the fibroblast activation protein FAP in a pancreatic cancer microenvironment. In some examples, grouping the cytoimmunohistochemical stained sections includes using X-tile software to conclude survival and classifying the cytoimmunohistochemical stained sections into a high expression group or a low expression group according to FAP percentage within the tumor.

In some embodiments, a training set and a validation set may be employed to verify the accuracy of the constructed model. For example, for pancreatic enhancement images from different patients, the images may be divided into two groups, a training set and a validation set, by their acquisition time. The training set is used for building the model, and the verification set is used for verifying the model.

To better illustrate the method 100 in fig. 1, the present application provides two examples i and ii.

Example i

Cancer immunotherapy currently has significant efficacy in treating many malignancies, however only a fraction of patients with advanced solid tumors respond to immunotherapy. The tumor-infiltrating immune cells play an important role in cancer immunotherapy, and subsequent therapy can be efficiently guided by predicting the expression of the tumor-infiltrating immune cells in a pancreatic cancer microenvironment through the iconography. Example i provides a method for predicting tumor-infiltrating immune cell expression in the immune environment of pancreatic cancer using imagemics. FIG. 2 shows a schematic diagram of a method 200 for predicting tumor-infiltrating immune cell expression in example i.

S210, a plurality of pancreas CT images are obtained, and the tumor area is marked.

In example i, 183 CT image samples of the pancreas were collected. These samples were all taken preoperatively, and these patients were all pathologically diagnosed as pancreatic cancer postoperatively.

S220, obtaining a cell immunohistochemical staining section corresponding to the CT image, establishing a tumor infiltrating lymphocyte score, and grouping the cell immunohistochemical staining section according to the lymphocyte infiltrating lymphocyte score.

CD4+ T, CD8+ T, CD +20B cells are the most prominent immune cells that make up the pancreatic cancer microenvironment. The scores were established in example i for all three immune cells. In other embodiments, a score may also be established for a single immune cell in a pancreatic cancer microenvironment. And quantifying the CD4+ T, CD8+ T and CD +20B lymphocytes by pathology software, establishing a COX regression model by taking survival of a patient as an outcome according to the area proportion of the CD4+ T, CD8+ T and CD +20B lymphocytes, and calculating the score of the tumor infiltrating lymphocytes according to the model.

Table 1 shows an example of calculating the score of the lymphoid tumor infiltrating lymphocytes using the regression formula in example i.

TABLE 1 COX regression analysis of tumor infiltrating lymphocytes

Tumor infiltrating lymphocyte score of 0.0024 × CD4+0.0269 × CD8+0.0139 × CD20

Fig. 3a shows TILs (tumor infiltrating lymphocyte score) distribution for patients corresponding to CT image samples of 183 pancreases in example i. An optimal TILs cutoff (0.73) can be determined using X-tile software, and the patients can be classified into TILs high and low groups based on the cutoff. FIG. 3b shows a schematic of the survival difference between the TILs high and low groups. FIG. 3b shows that there is a significant difference in survival between the TILs high and low groups, with P < 0.0001.

And S230, extracting the image omics characteristics of the tumor area to obtain an omics parameter data set, and performing characteristic screening on the tumor omics parameter data set to select target characteristics.

And (3) performing arterial phase and portal phase image omics feature extraction on each tumor region, wherein the characteristic extraction comprises first-order statistics, shape features, gray level co-occurrence matrix (GLCM) features, Gray Level Dependency Matrix (GLDM) features, gray level travel length matrix (GLRLM) features, gray level scale region matrix (GLSZM) features and neighborhood gray level region difference matrix (NGTDM) features, each feature comprises various descriptions and statistical values, 1409 group characteristics are extracted from each patient in each phase to form an omic parameter data set.

And reducing dimensions of the omics features extracted from the pancreatic tumor by variance analysis and Spearman correlation analysis to screen the features, and excluding image omics features which have no obvious difference between groups or no obvious association with TILs score expression. From these, 25 arterial phase and 12 portal venous phase imaging omics features were selected.

S240, training the XGboost classifier to obtain a prediction model.

And training and further screening by using an XGboost classifier to obtain 13 arterial-phase omics characteristics and 7 portal-phase omics characteristics, and forming a prediction model by using 20 total histological characteristics.

Fig. 4 lists a bar chart of the above 20 mathematical feature weights. In a _ B _ C in the expression form of each feature in fig. 3, a represents a category to which the feature belongs. In the Wavelet filtering, the Wavelet filtering may be composed of a combination of high-pass (H) and low-pass (L) filtering of each dimension, including Wavelet-LLH, Wavelet-LHL, Wavelet-LHH, Wavelet-HLL, Wavelet-HLH, Wavelet-HHL, Wavelet-LLL, and Wavelet-HHH. First order statistics firstorder, including Minimum of gray values in Minimum-ROI, Median gray intensity within Mean-ROI, Maximum gray intensity within Maximum-ROI, Skewness, Interquartile Range, Median, Total Energy, 10percent, 90percent, Root Mean Squared, etc.; glcm (gray-Level Co-occurrrence Matrix) is a gray Level Co-occurrence Matrix, which represents the number of times a certain shape of pixel pair appears in a gray Level image in the whole image, including features ClusterShade, Imc1, Imc2, Correlation, etc. Glszm (Gray-level Size Zone Matrix) is a Gray scale Zone Matrix, and stores the Size and number information of connected domains of all Gray scales in an image, and the Size and number of 2D/3D connected domains in the image are measured.

S250, outputting the cell expression according to the prediction model.

The CT image may be input to a prediction model, and the model outputs a high-expression or low-expression result corresponding to the CT image, and in some embodiments, the output result is a high-expression probability or a low-expression probability.

In some embodiments, a training set and a validation set may be employed to verify the accuracy of the constructed model. For example, in example i, the images may be divided into two groups, i.e., a training set (136) and a validation set (47), according to their acquisition times. The training set is used for building the model, and the verification set is used for verifying the model. The number of TILs high groups and the number of TILs low groups in the training set are 76 and 60, respectively. The number of TILs high groups and the number of TILs low groups in the validation set are 24 and 23, respectively.

Fig. 5 shows ROC curves (receiver operating characteristics curve) on the training set and the validation set using the established prediction model.

The ROC curve is also called sensitivity curve (sensitivity curve). The same sensitivity is reflected by each point on the curve, which are both responses to the same signal stimulus. The area under the ROC curve (AUC) refers to the area enclosed by the ROC curve and the x-axis (1, 0) - (1, 1). Generally, the diagnostic test proves to be of diagnostic value as long as the AUC is greater than 0.5. Meanwhile, the closer the AUC is to 1, the closer to the (0, 1) point, the better the authenticity of the diagnostic test is demonstrated. Fig. 5 shows that the AUC of the model on the training set and the validation set is 0.93 and 0.79, respectively, which indicates that the prediction effect of the prediction model is better.

Figure 6 shows an example of predicting the expression of pancreatic cancer tumor-infiltrating lymphocytes using the model established in example i. FIG. 6A shows a cross-sectional arterial phase pancreas enhancement image of patient M in which a low density lump image (box line marked) with one margin of the pancreatic head is visible. Using the predictive model in example i, a high probability of 21% expression of CD4, CD8 and CD20 was obtained for the patient's pancreatic cancer tumor infiltrating lymphocytes. FIGS. 6B-D show staining of immunohistochemical sections with D4, CD8, and CD20 in patient M, with a small amount of tumor-infiltrating lymphocyte infiltration in the visual field (X20). FIG. 6E shows a cross-sectional arterial phase pancreatic enhancement image of patient N in which a low density lump image (box line marked) with one margin of the pancreatic head is visible. Using the predictive model in example i, the patient's pancreatic cancer tumor infiltrating lymphocytes CD4, CD8 and CD20 were found to have a high expression probability of 91.91%. FIGS. 6F-H show staining of immunohistochemical sections of patient N with D4, CD8, and CD20, showing massive tumor-infiltrating lymphocyte infiltration in the visual field (X20). The example of figure 6 demonstrates that the prediction model in example i can predict well the expression of tumor infiltrating lymphocytes in the microenvironment using imaging omics.

Example ii

In the tumor microenvironment, Fibroblast Activation Protein (FAP) can be selectively expressed in cancer-associated fibers, FAP is selectively expressed in more than 90% of epithelial cancers, but is expressed at a very low level in healthy tissues, and FAP expression represents the level of cancer-associated fibers. FAP can directly promote proliferation and migration of fibroblasts and other types of cells, including tumor cells, endothelial cells, and immune cells, leading to escape of tumor invasiveness, extracellular matrix degradation, tumor angiogenesis, and immune monitoring. Therefore, FAP can be used as a target point for tumor treatment. Therefore, finding a new therapeutic target becomes a new strategy for improving the curative effect and prognosis of pancreatic cancer patients.

Example ii provides a method 600 for predicting FAP expression in a pancreatic cancer setting using imagery omics.

S610-S650 in method 600 is substantially the same as S210-S250 in method 200, and only the differences between the two are discussed below. In S610, 152 CT image samples of the pancreas are collected. These samples were all taken preoperatively, and these patients were all pathologically diagnosed as pancreatic cancer postoperatively.

In S620, post-operative FAP immunohistochemically stained sections of all patients are obtained, their expression quantified, and pancreatic cancer patients are then grouped according to expression.

In S630, portal phase imaging omics feature extraction is performed on each tumor region to form an omics parameter dataset. And reducing dimensions of the omics features extracted from the pancreatic tumor by variance analysis, Spearman correlation analysis and LASSO regression to screen the features, and excluding the imaging omics features which have no obvious difference between groups or have no obvious association with score expression. From these, 17 portal phase imaging omics features were selected.

In S640, the XGBoost classifier is used for training and further screening to obtain 13 portal phasics features, and the bar graph of these feature weights is shown in fig. 7.

At S650, the cell expression is output according to the model. In some embodiments, a training set and a validation set may be employed to verify the accuracy of the constructed model. For example, in example ii, the images may be divided into two groups, i.e., a training set (94) and a validation set (58), according to their acquisition times. The training set is used for building the model, and the verification set is used for verifying the model. The number of TILs high groups and the number of TILs low groups in the training set are 36 and 58, respectively. The number of high groups and the number of low groups in the validation set were 25 and 33, respectively.

According to the ROC curves of the established prediction model on the training set and the verification set, the AUC of the model on the training set and the AUC of the model on the verification set are respectively 0.97 and 0.75, and the prediction effect of the prediction model is better.

Fig. 8 shows an example of predicting FAP expression using the model established in example ii. Figure 8A shows a cross-sectional portal pancreatic enhancement image of patient X with a low density lump shadow (white arrow) with an unclear border visible at the tail of the pancreas. Using the predictive model in example ii, the patient's FAP under-expression probability of 73.66% was obtained. Fig. 8B shows a quantification picture of FAP-stained immunohistochemical sections of patients with low expression of FAP (x 1) visible in the tumor area. Figure 8C shows a cross-sectional portal pancreatic enhancement image of patient Y in which a low density lump shadow (at white arrows) of the posterior segment of the pancreas lacking clear borders is visible. Using the predictive model in example ii, the high expression probability for FAP for this patient was found to be 96.83%. Fig. 8D shows FAP-stained immunohistochemical section quantification pictures of patient Y with FAP high expression (x 1) visible in the tumor region. The example of fig. 8 demonstrates that the prediction model in example ii can predict FAP expression well in a microenvironment using imagery omics.

Referring now to fig. 9, fig. 9 illustrates a schematic structural diagram of a system for predicting cell expression in a pancreatic cancer microenvironment, according to one embodiment of the present application. A system 900 for predicting cell expression in a pancreatic cancer microenvironment comprises the following modules.

An acquisition module 910 for acquiring a CT image of a pancreas and marking a tumor region;

the extraction module 920 is configured to extract the image omics characteristics of the tumor region to obtain an omic parameter data set, perform characteristic screening on the tumor omic parameter data set, and select a target characteristic;

a grouping module 930, configured to obtain a cytoimmunohistochemical stained section corresponding to the CT image, and group the cytoimmunohistochemical stained section, where a grouping result includes a high expression group and a low expression group;

a training module 940 for training the XGBoost classifier using the target features and the grouping result to obtain a prediction model;

a prediction module 950 for outputting the cell expression according to the model.

The system for predicting cell expression in a pancreatic cancer microenvironment in fig. 9 is grouped by high and low expression for immunohistochemically stained sections of cells corresponding to CT images; and performing image omics feature extraction on the CT image to obtain target features, and training an XGboost classifier according to the grouping result and the target features to obtain the image omics features most relevant to a certain component in the microenvironment to form a prediction model. Cell expression in the corresponding microenvironment in the pancreatic CT images can be predicted using the trained model. The system of fig. 9 predicts microenvironment by using imagemics, i.e., microenvironment is assessed by preoperative and noninvasive methods for predicting cell expression in pancreatic cancer microenvironment, and patients effective for subsequent treatment can be efficiently screened out.

Referring now to FIG. 10, shown is a block diagram of an electronic device 1000 in accordance with one embodiment of the present application. The electronic device 1000 may include one or more processors 1002, system control logic 1008 coupled to at least one of the processors 1002, system memory 1004 coupled to the system control logic 1008, non-volatile memory (NVM)1006 coupled to the system control logic 1008, and a network interface 1010 coupled to the system control logic 1008.

The processor 1002 may include one or more single-core or multi-core processors. The processor 1002 may include any combination of general-purpose processors and dedicated processors (e.g., graphics processors, application processors, baseband processors, etc.). In embodiments herein, the processor 1002 may be configured to perform one or more embodiments in accordance with the various embodiments shown in fig. 1.

In some embodiments, system control logic 1008 may include any suitable interface controllers to provide any suitable interface to at least one of processors 1002 and/or any suitable device or component in communication with system control logic 1008.

In some embodiments, system control logic 1008 may include one or more memory controllers to provide an interface to system memory 1004. System memory 1004 may be used to load and store data and/or instructions. Memory 1004 of device 1000 may include any suitable volatile memory, such as suitable Dynamic Random Access Memory (DRAM), in some embodiments.

The NVM/memory 1006 may include one or more tangible, non-transitory computer-readable media for storing data and/or instructions. In some embodiments, the NVM/memory 1006 may include any suitable non-volatile memory, such as flash memory, and/or any suitable non-volatile storage device, such as at least one of a HDD (Hard Disk Drive), CD (Compact Disc) Drive, DVD (Digital Versatile Disc) Drive.

The NVM/memory 1006 can include a portion of a storage resource installed on a device of the device 1000, or it can be accessed by, but not necessarily a part of, the device. For example, the NVM/storage 1006 may be accessed over a network via the network interface 910.

In particular, the system memory 1004 and the NVM/storage 1006 may each include: a temporary copy and a permanent copy of the instructions 1020. The instructions 920 may include: instructions that, when executed by at least one of the processors 1002, cause the device 1000 to implement the method as shown in fig. 1. In some embodiments, the instructions 1020, hardware, firmware, and/or software components thereof may additionally/alternatively be disposed in the system control logic 1008, the network interface 1010, and/or the processor 1002.

Network interface 1010 may include a transceiver to provide a radio interface for device 1000 to communicate with any other suitable device (e.g., front-end module, antenna, etc.) over one or more networks. In some embodiments, the network interface 1010 may be integrated with other components of the device 1000. For example, the network interface 1010 may be integrated with at least one of the processor 1002, the system memory 1004, the NVM/storage 1006, and a firmware device (not shown) having instructions that, when executed by at least one of the processor 1002, the device 1000 implements one or more of the various embodiments shown in fig. 1.

The network interface 1010 may further include any suitable hardware and/or firmware to provide a multiple-input multiple-output radio interface. For example, network interface 1010 may be a network adapter, a wireless network adapter, a telephone modem, and/or a wireless modem.

In one embodiment, at least one of the processors 1002 may be packaged together with logic for one or more controllers of system control logic 1008 to form a System In Package (SiP). In one embodiment, at least one of the processors 1002 may be integrated on the same die with logic for one or more controllers of system control logic 1008 to form a system on a chip (SoC).

The apparatus 1000 may further comprise: input/output (I/O) devices 1012. I/O device 1012 may include a user interface to enable a user to interact with device 1000; the design of the peripheral component interface enables peripheral components to also interact with the device 1000.

In some embodiments, the user interface may include, but is not limited to, a display (e.g., a liquid crystal display, a touch screen display, etc.), a speaker, a microphone, one or more cameras (e.g., still image cameras and/or video cameras), a flashlight (e.g., a light emitting diode flash), and a keyboard.

In some embodiments, the peripheral component interfaces may include, but are not limited to, a non-volatile memory port, an audio jack, and a power interface.

The method embodiments of the present application may be implemented in software, magnetic, firmware, etc.

Program code may be applied to input instructions to perform the functions described herein and generate output information. The output information may be applied to one or more output devices in a known manner. For purposes of this application, a processing system includes any system having a processor such as, for example, a Digital Signal Processor (DSP), a microcontroller, an Application Specific Integrated Circuit (ASIC), or a microprocessor.

The program code may be implemented in a high level procedural or object oriented programming language to communicate with a processing system. The program code can also be implemented in assembly or machine language, if desired. Indeed, the mechanisms described herein are not limited in scope to any particular programming language. In any case, the language may be a compiled or interpreted language.

One or more aspects of at least one embodiment may be implemented by representative instructions stored on a computer-readable storage medium, which represent various logic in a processor, which when read by a machine causes the machine to fabricate logic to perform the techniques described herein. These representations, known as "IP cores" may be stored on a tangible computer-readable storage medium and provided to a number of customers or manufacturing facilities to load into the manufacturing machines that actually make the logic or processor.

One embodiment of the present application discloses a computer readable medium storing one or more programs executable by one or more processors to implement the method of the present application for predicting cell expression in a pancreatic cancer microenvironment.

One embodiment of the present application discloses a computer program product comprising a computer program that when executed by a processor implements the method of the present application for predicting cell expression in a pancreatic cancer microenvironment.

While the invention has been shown and described with reference to certain preferred embodiments thereof, it will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention.

Claims

1. A method of predicting cell expression in a pancreatic cancer microenvironment, comprising,

acquiring a CT image of pancreas and marking a tumor area;

extracting the image omics characteristics of the tumor area to obtain an omics parameter data set, carrying out characteristic screening on the tumor omics parameter data set, and selecting target characteristics;

obtaining a cell immunohistochemical staining section corresponding to the CT image, and grouping the cell immunohistochemical staining section, wherein the grouping result comprises a high expression group and a low expression group;

training an XGboost classifier by using the target features and the grouping result to obtain a prediction model;

outputting the cell expression according to the predictive model.

2. The method of predicting cell expression in a pancreatic cancer microenvironment of claim 1, wherein performing feature screening on the lumomics parameter dataset comprises selecting the target features using analysis of variance and Spearman correlation.

3. The method of predicting cell expression in a pancreatic cancer microenvironment of claim 2, wherein feature screening the lumomics parameter dataset further comprises selecting the target features using a Lasso regression analysis.

4. The method of claim 1, wherein the cells are tumor infiltrating lymphocytes comprising CD4+ T, CD8+ T, and CD +20B cells.

5. The method of predicting cell expression in a pancreatic cancer microenvironment of claim 1, wherein the cell is Fibroblast Activation Protein (FAP).

6. The method of predicting cell expression in a pancreatic cancer microenvironment of claim 4, wherein grouping the immunohistochemically stained sections comprises,

scoring said tumor infiltrating lymphocytes, said scoring taking survival as a predictor variable, taking the percentage of positive expression of said CD4+ T, CD8+ T, and CD +20B cells in the tumor as an independent variable, establishing a COX regression model, calculating said tumor infiltrating lymphocytes score according to said COX regression model;

and dividing the cellular immunohistochemical staining section into the high expression group or the low expression group according to the tumor infiltrating lymphocyte score by using X-tile software and taking survival as a result.

7. The method of predicting cell expression in a pancreatic cancer microenvironment of claim 5, wherein grouping the cytoimmunohistochemically stained sections comprises using X-tile software to conclude survival and classifying the cytoimmunohistochemically stained sections into the high expression panel or the low expression panel based on FAP intratumoral percentage.

8. The method of predicting cell expression in a pancreatic cancer microenvironment of claim 1, wherein the CT images comprise an arterial phase image and a portal phase image for which 1409 iconomic features are extracted, comprising the omic parameter dataset.

9. The method of predicting cell expression in a pancreatic cancer microenvironment of claim 1, further comprising dividing the CT images into a training set and a validation set, the training set being used to construct the predictive model, the validation set being used to validate the predictive model.

10. An electronic device, comprising a memory storing computer-executable instructions and a processor; the instructions, when executed by the processor, cause the apparatus to carry out the method of predicting cell expression in a pancreatic cancer microenvironment of any one of claims 1 to 9.

11. A computer readable medium storing one or more programs, the one or more programs being executable by one or more processors to perform the method of any of claims 1 to 9 for predicting cell expression in a pancreatic cancer microenvironment.

12. A system for predicting cell expression in a pancreatic cancer microenvironment, the system comprising:

the acquisition module is used for acquiring a CT image of pancreas and marking a tumor area;

the extraction module is used for extracting the image omics characteristics of the tumor region to obtain an omics parameter data set, and performing characteristic screening on the tumor omics parameter data set to select target characteristics;

the grouping module is used for acquiring a cell immunohistochemical staining section corresponding to the CT image and grouping the cell immunohistochemical staining section, wherein the grouping result comprises a high expression group and a low expression group;

the training module is used for training an XGboost classifier by using the target characteristics and the grouping result to obtain a prediction model;

a prediction module to output the cell expression according to the model.

13. A computer program product comprising a computer program which, when executed by a processor, implements a method of predicting cell expression in a pancreatic cancer microenvironment of any one of claims 1 to 9.