CN115602311A - Pancreatic cancer auxiliary inspection tool based on collagen fiber multivariate parameter analysis - Google Patents

Pancreatic cancer auxiliary inspection tool based on collagen fiber multivariate parameter analysis Download PDF

Info

Publication number
CN115602311A
CN115602311A CN202210849955.9A CN202210849955A CN115602311A CN 115602311 A CN115602311 A CN 115602311A CN 202210849955 A CN202210849955 A CN 202210849955A CN 115602311 A CN115602311 A CN 115602311A
Authority
CN
China
Prior art keywords
collagen fibers
multivariate
pancreatic
pancreatic cancer
collagen
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.)
Pending
Application number
CN202210849955.9A
Other languages
Chinese (zh)
Inventor
刘智毅
钱书豪
孟佳
丁志华
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Zhejiang University ZJU
Jiaxing Research Institute of Zhejiang University
Original Assignee
Zhejiang University ZJU
Jiaxing Research Institute of Zhejiang University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Zhejiang University ZJU, Jiaxing Research Institute of Zhejiang University filed Critical Zhejiang University ZJU
Priority to CN202210849955.9A priority Critical patent/CN115602311A/en
Publication of CN115602311A publication Critical patent/CN115602311A/en
Pending legal-status Critical Current

Links

Images

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
    • G16H50/20—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for computer-aided diagnosis, e.g. based on medical expert systems
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/0002—Inspection of images, e.g. flaw detection
    • G06T7/0012—Biomedical image inspection
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
    • G06V10/765—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects using rules for classification or partitioning the feature space
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
    • G06V10/80—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level
    • G06V10/806—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level of extracted features
    • 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
    • G16H30/20—ICT specially adapted for the handling or processing of medical images for handling medical images, e.g. DICOM, HL7 or PACS
    • 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
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10056—Microscopic image
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10068—Endoscopic image
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20081—Training; Learning
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20212—Image combination
    • G06T2207/20221—Image fusion; Image merging
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30004—Biomedical image processing
    • G06T2207/30096—Tumor; Lesion
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V2201/00—Indexing scheme relating to image or video recognition or understanding
    • G06V2201/03—Recognition of patterns in medical or anatomical images
    • G06V2201/032—Recognition of patterns in medical or anatomical images of protuberances, polyps nodules, etc.

Landscapes

  • Engineering & Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • Medical Informatics (AREA)
  • General Health & Medical Sciences (AREA)
  • Public Health (AREA)
  • Theoretical Computer Science (AREA)
  • Databases & Information Systems (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • General Physics & Mathematics (AREA)
  • Biomedical Technology (AREA)
  • Epidemiology (AREA)
  • Primary Health Care (AREA)
  • Physics & Mathematics (AREA)
  • Pathology (AREA)
  • Artificial Intelligence (AREA)
  • Computing Systems (AREA)
  • Evolutionary Computation (AREA)
  • Software Systems (AREA)
  • Radiology & Medical Imaging (AREA)
  • Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
  • Data Mining & Analysis (AREA)
  • Multimedia (AREA)
  • Quality & Reliability (AREA)
  • Investigating Or Analysing Biological Materials (AREA)

Abstract

The invention discloses a pancreatic cancer diagnosis tool based on collagen fiber multivariate fusion analysis, which is used for quantitative analysis of collagen fiber structural characteristics of pancreatic tissues and diagnosis of cancer pancreatic tissues, can realize fully-automatic auxiliary diagnosis of normal and cancer pancreatic tissues, improves sensitivity and specificity and simultaneously gives consideration to excellent reliability; through comprehensive analysis of various structural parameters, the method greatly improves the information complementarity of pancreatic tissue collagen fiber analysis, and further improves the accuracy of cancer diagnosis; the invention also utilizes the combination of the quantization method of the pixel level resolution and the pseudo-color coding technology to visually reflect the structural characteristics of the collagen fibers, thereby enhancing the readability of information; the method realizes the quantitative analysis of the three-dimensional image, greatly enhances the application range and potential of the pancreatic cancer diagnosis technology based on collagen fiber multivariate fusion analysis compared with the traditional two-dimensional analysis, and has profound significance for the full-automatic auxiliary diagnosis of pancreatic cancer.

Description

Pancreatic cancer auxiliary inspection tool based on collagen fiber multivariate parameter analysis
Technical Field
The invention belongs to the technical field of image analysis and disease auxiliary inspection of biological tissues, and particularly relates to an auxiliary pancreatic cancer inspection tool based on collagen fiber multivariate parameter analysis.
Background
Pancreatic cancer is one of the common malignancies of the digestive tract, with morbidity at position 13 of the malignancy worldwide, while mortality at position 8 of tumor-associated deaths. The five-year survival rate after diagnosis is only about 10%, and about 75% of patients die one year after diagnosis and are one of the worst-case malignant tumors. The severe prognosis of pancreatic cancer is largely due to late diagnosis, and most patients with pancreatic cancer lose the chance of surgical treatment when they are diagnosed. The innovation of an auxiliary pancreatic cancer examination tool is a key for improving pancreatic cancer prognosis, at present, the auxiliary pancreatic cancer examination mainly depends on pathological examination, and procedures of the auxiliary pancreatic cancer examination comprise links such as formalin fixation, paraffin embedding, staining and sectioning. However, the pathological examination has its own drawbacks, firstly, the whole process takes 1-3 days, which requires a lot of manpower and time; secondly, because the tissues need to be sliced, the pathology examination cannot observe the three-dimensional structure of the pancreatic tissues under the original volume condition; thirdly, the long preparation process may introduce artifacts, which affect the accuracy of the examination; fourth, since the entire pathological examination process is performed manually by a pathologist, it will inevitably introduce human factors affecting the reliability of the examination.
Quantitative analysis of structural characteristics of extracellular matrix (mainly comprising collagen fibers) is a brand new idea for auxiliary examination of pancreatic cancer. Extracellular matrix is a complex network composed of various macromolecules, has unique biomechanical and chemical mechanisms, and plays a key role in many life processes. Collagen fibers are a major component of the extracellular matrix and are continually undergoing self-remodeling under the influence of cell-matrix interactions such as cell migration, division and differentiation. Many studies have shown that the changes in the mechanical and biochemical states of collagen fibers are related to tumor evolution, and in addition, collagen fibers are the most important protein component of the pancreatic interstitial tissue, which makes the structural characteristics of collagen fibers a potential biomarker for pancreatic cancer examination.
However, the current method for quantifying the collagen fiber structure and the associated auxiliary examination tools for pancreatic cancer are very insufficient and have certain limitations. Firstly, these techniques can only quantify the single structural characteristics of collagen fibers, such as density, direction and arrangement order degree, but lack the technique of comprehensively analyzing the multiple structural parameters of collagen fibers; secondly, most of the existing technologies are limited to two-dimensional analysis of collagen fibers, and few technologies can quantify the three-dimensional structure of the collagen fibers, so that with the development of imaging technologies, the three-dimensional structure of the collagen fibers can be clearly revealed, and a quantitative analysis tool needs to be expanded towards three dimensions. Third, techniques for analyzing the single structural features of collagen fibers and corresponding disease detection tools have limited detection capabilities for lesions, often with low accuracy. Fourth, the conventional method for quantifying collagen fibers usually can only obtain an overall numerical result for a whole image, and cannot reflect local tissue abnormalities. In summary, there is a need for a high-accuracy pixel-level resolution analysis tool for performing multidimensional quantification on a three-dimensional structure of collagen fibers to perform high-accuracy multivariate comprehensive analysis on extracellular matrices of two-dimensional and three-dimensional pancreatic tissues, so as to realize early detection and auxiliary inspection of pancreatic cancer micro-lesions.
Disclosure of Invention
The invention provides an auxiliary pancreatic cancer examination tool based on collagen fiber multivariate fusion analysis based on the defects of the prior art. The method can carry out multi-dimensional quantitative analysis on the structural characteristics of two-dimensional/three-dimensional collagen fibers to obtain multiple parameters of the collagen fibers such as signal intensity, direction variance, space curvature and the like, and the parameters are utilized to carry out auxiliary inspection on the disease condition of pancreatic tissues. The method can process and analyze two-dimensional and three-dimensional collagen fiber images obtained by various optical microscopy technologies, and can realize real-time monitoring and evaluation of the health state of pancreatic tissues by combining the endoscopic imaging technology. In conclusion, the method has the advantages of rapidness, simplicity, accuracy, strong applicability and the like.
The invention is realized by adopting the following technical scheme:
the invention discloses a pancreatic cancer auxiliary inspection tool based on collagen fiber multivariate parameter analysis, which is characterized in that the inspection tool is used for fusion analysis of multivariate structural characteristics of pancreatic tissue collagen fibers and establishment of an analysis model of the pancreatic tissue collagen fibers for characterization and calculation, and the construction of the inspection tool comprises the following steps:
1) Imaging the collagen fibers of the extracellular matrix of the pancreatic tissue to be examined by a second harmonic technology and a two-photon fluorescence technology to obtain an optical image with micron-scale resolution; the second harmonic technology and the two-photon fluorescence technology provide micron-scale high resolution and three-dimensional imaging capability, and compared with the traditional technology, the tissue structure change caused by cancer evolution can be better revealed;
2) Extracting structural features of collagen fibers in the optical image;
3) Calculating the multivariate structure parameters of the signal intensity, the spatial orientation, the spatial curvature and the direction variance characterization parameters of the structural characteristics of the collagen fibers respectively to obtain a numerical result;
4) Marking the position of the pancreatic tissue subjected to second harmonic/two-photon fluorescence imaging, then performing hematoxylin-eosin staining on the pancreatic tissue for pathological examination, and then labeling whether the marked area contains a cancer focus according to the result of the pathological examination;
5) Constructing a support vector machine model based on the numerical result of the multi-element structure parameter calculation obtained in the step 3) and the label information obtained in the step 4);
6) In the support vector machine model, each independent sample obtains a vector according to the multivariate parameter value of the independent sample
Figure BDA0003754325140000031
The multivariate parameter comprises signal intensity, direction variance and space curvature, including
Figure BDA0003754325140000032
Wherein s is i ,v i ,w i Respectively obtaining the signal intensity, the direction variance and the space curvature value which are obtained by the calculation in the step 3);
the equation of the decision plane obtained according to the multivariate parameters is as follows:
Figure BDA0003754325140000033
wherein
Figure BDA0003754325140000034
B is the intercept of the decision plane, and the judgment condition of whether the pancreatic cancer tissue contains the cancer focus is as follows:
Figure BDA0003754325140000035
y i indicates tag information, y i =1 indicates that pancreatic cancer foci are not contained in tissue, and y i = -1 indicates that pancreatic cancer foci are contained in the tissue;
7) Training is carried out through a large number of multivariate parameters and label information to obtain k and b, and the method is specifically described as the following problem of solving the condition maximum value:
Figure BDA0003754325140000036
wherein s.t. indicates simultaneous satisfaction, m is the total number of samples;
8) Training to obtain decision surface normal vector
Figure BDA0003754325140000037
Intercept b;
further obtaining an auxiliary pancreatic cancer examination tool based on collagen fiber multivariate parameter analysis. The multivariate parameter information and the label information (whether tissues contain the pancreatic cancer focus) are used for training a support vector machine model together to obtain a decision plane which takes the multivariate parameters of the collagen fiber as features, so that the examination of the pancreatic cancer focus is carried out, the multivariate parameters including signal intensity, space curvature, direction variance and the like are comprehensively considered for constructing an auxiliary examination model of the pancreatic cancer, the information complementarity of different collagen fiber structural features is ensured, and the examination of the pancreatic cancer focus has higher accuracy compared with a single parameter examination model.
As a further improvement, in step 3) of the present invention, the signal intensity quantifies the content of collagen fibers in the biological tissue; the direction of the collagen fibers in the image is quantified by spatial orientation, and the value is 0-180 degrees; the space curvature quantifies the bending degree of the collagen fibers, the value is 0-1, and the larger the value is, the more bent collagen fibers are represented; the direction variance quantifies the arrangement of the collagen fibers, the value interval is 0-1, the value of 0 represents the completely parallel arrangement of the collagen fibers, and the value of 1 represents the completely unordered arrangement of the collagen fibers. The above parameters quantitatively reflect the structural characteristics of the collagen fibers, and the influence of human subjective factors on cancer examination is avoided.
As a further improvement, the signal intensity parameter of the present invention is obtained by calculating an average value of collagen fiber signals in the image; the spatial orientation parameter is pixel level resolution, and is quantized according to the change of the signal intensity of the collagen fiber in the neighborhood where the pixel is located along different directions; the spatial curvature parameter is pixel-level resolution, and is quantified by analyzing the change of the spatial orientation of collagen fibers in the neighborhood of the pixel; the direction variance parameter is pixel-level resolution, and is quantified by calculating the variance of the spatial orientation of collagen fibers in the neighborhood of the pixel.
As a further improvement, the normal vector of the decision surface obtained in the step 8) is trained
Figure BDA0003754325140000041
Compared with the prior art, the invention has the following beneficial effects:
1. compared with the traditional pathology examination, the invention saves a large amount of labor and time cost, simultaneously avoids the problem possibly caused by a long-time preparation process, and is hopeful to realize real-time monitoring and evaluation of the health state of the pancreatic tissue by combining the endoscopic technology. Has the advantages of rapidness, conciseness, accuracy, strong applicability and the like.
2. Compared with the traditional pathology examination and collagen fiber analysis tool, the method can analyze two-dimensional and three-dimensional images, and can only analyze two-dimensional images compared with the background technology, the method provided by the invention can more completely analyze the structural information of the collagen fibers provided by three-dimensional imaging, the application scene is very wide, the application range and the potential of the pancreatic cancer auxiliary examination technology based on collagen fiber multivariate fusion analysis are greatly enhanced, the method can realize more comprehensive and complete understanding of biological tissues, and has profound significance for the research and analysis of full-automatic auxiliary examination, life activities and disease evolution processes of pancreatic cancer.
3. Compared with pathological examination, the method can obtain quantitative analysis results, so that the health state of the pancreatic tissues is objectively evaluated according to numerical results, and subjective factors caused by human factors are avoided.
4. Compared with the traditional collagen fiber analysis tool which can only analyze a single structural parameter, the invention can carry out structure quantification of different dimensions on collagen fibers, trains a support vector machine model by taking a plurality of parameters as characteristics to realize the examination of pancreatic cancer focuses, provides collagen fiber morphological information which is mutually complemented by the plurality of parameters, and further completely analyzes the collagen fibers and pancreatic tissues, greatly improves the information complementarity of pancreatic tissue collagen fiber analysis, and further improves the accuracy of cancer examination compared with the traditional technology;
5. compared with the traditional collagen fiber analysis tool, only one integral numerical result can be obtained for one image, the space orientation, space curvature and direction variance parameters used by the method are quantitative information of pixel resolution, and the high-resolution numerical information can reflect local abnormality of tissues and is beneficial to detection of tiny focuses.
Drawings
FIG. 1 is a flow chart of the construction of an auxiliary pancreatic cancer examination tool based on collagen fiber multivariate parameter analysis;
fig. 2 is a schematic diagram of the optical imaging result of the pancreatic cancer auxiliary examination tool based on collagen fiber multivariate parameter analysis and the output result of multivariate parameter calculation: FIGS. 2 (a) and 2 (b) are second harmonic/two-photon fluorescence representative images of cancerous and normal pancreatic tissue; FIGS. 2 (c) - (f) are second harmonic signal intensity images, spatial orientation pseudo color code patterns, direction variance pseudo color code patterns and spatial curvature pseudo color code patterns of collagen fibers of cancer pancreatic tissues calculated by collagen fiber multivariate parameters; FIGS. 2 (g) - (j) are second harmonic signal intensity images, spatial orientation pseudo color code patterns, direction variance pseudo color code patterns and spatial curvature pseudo color code patterns of collagen fibers of normal pancreatic tissues calculated by using the multivariate parameters of the collagen fibers; FIG. 2 (k) is a histogram comparing the results of image signal intensity, direction variance, and spatial curvature of cancer and normal tissues;
FIG. 3 is a schematic diagram showing the comparison between the output result of the pancreatic cancer auxiliary inspection tool and the multi-parameter and single-parameter inspection tools: FIG. 3 (a) is the examination result outputted by the auxiliary examination tool for pancreatic cancer, which comprises an examination result three-dimensional scattergram constructed by using three parameters of direction variance, spatial curvature and signal intensity as coordinate axes, and the evaluation result of whether a cancer focus exists in the tissue, wherein the accuracy rate of the result is up to 96.2%; FIG. 3 (b) is a receiver operating characteristic curve (ROC) for a support vector machine model under auxiliary examination, with an area under the curve (AUC) of 0.965; FIG. 3 (c) is a two-dimensional scattergram of inspection results constructed with coordinate axes using different parameters; FIG. 3 (d) shows the results of pancreatic cancer examination using single parameters of signal intensity, direction variance and spatial curvature alone, which are less accurate than the results of examination using a plurality of parameters of the three.
Detailed Description
The invention discloses a pancreatic cancer auxiliary inspection tool based on collagen fiber multivariate fusion analysis, and fig. 1 is a flow chart for constructing a pancreatic cancer auxiliary inspection tool based on collagen fiber multivariate parameter analysis, which comprises the following steps:
optical imaging: collagen fibers in the extracellular matrix of pancreatic tissue are imaged. Imaging tools include second harmonic imaging, two-photon fluorescence imaging, and the like.
And (3) calculating the multivariate parameters of the collagen fibers: performing multivariate quantitative analysis on the collagen fiber image of the pancreatic tissue obtained in the step 1), and comprehensively calculating parameters such as signal intensity, spatial orientation, spatial curvature, direction variance and the like of the collagen fiber. The signal intensity parameter is obtained by calculating the average value of collagen fiber signals in the image, so that the content of collagen fibers in biological tissues is quantified; the spatial orientation quantifies the direction of pixel-level resolution of the collagen fibers in the image, and the value is 0-180 degrees; the spatial curvature can quantify the bending degree of the pixel level resolution of the collagen fibers through the change of the spatial orientation of the collagen fibers, and obtain a normalized quantity value, wherein the more the value is, the more bent collagen fibers are represented; the directional variance quantifies the arrangement of collagen fibers with pixel-level resolution by calculating the variance of the spatial orientation of collagen fiber pixels in a neighborhood, the value interval is 0-1, the value of 0 represents the completely parallel collagen fiber arrangement, and the value of 1 represents the completely unordered collagen fiber arrangement. Besides the numerical calculation result, the pseudo-color coding image can be output according to the quantization result of the pixel level resolution, so that the collagen fiber structure information can be displayed more clearly.
Marking the position of the pancreatic tissue subjected to second harmonic/two-photon fluorescence imaging, then performing hematoxylin-eosin staining on the pancreatic tissue for pathological examination, and then labeling whether the marked area contains a cancer focus or not according to the pathological examination result;
constructing a support vector machine model based on the obtained numerical result of the multi-element structure parameter calculation and the obtained label information; and (3) checking whether a potential cancer focus exists in the tissue according to three parameters of signal intensity, direction variance and space curvature obtained by calculating the multivariate parameters of the collagen fibers. The model ensures the information complementarity provided by different structural characteristics of the collagen fibers, improves the precision compared with single parameter analysis, and can realize the pancreatic cancer auxiliary examination application with high accuracy.
In the support vector machine model, each independent sample obtains a vector according to the multivariate parameter value of the independent sample
Figure BDA0003754325140000061
The multivariate parameter comprises signal intensity, direction variance and space curvature, including
Figure BDA0003754325140000062
Wherein s is i ,v i ,w i Respectively calculating the signal intensity, the direction variance and the space curvature value obtained in the step;
the realization of the pancreatic cancer auxiliary examination utilizes a pre-trained mathematical model of a support vector machine with collagen fiber multivariate parameters as characteristics, in order to train the model, a large number of samples from different pancreatic tissues are imaged and then multivariate parameters are calculated, and then multivariate parameter information and label information (whether the tissues contain the focus of pancreatic cancer) are used for training the support vector machine model together, and the detailed steps are as follows:
an independent sample is set to obtain a vector according to the multivariate parameter values (signal strength, direction variance and space curvature) of the sample as
Figure BDA0003754325140000071
Is provided with
Figure BDA0003754325140000072
Wherein s is i ,v i ,w i The signal intensity, direction variance and space curvature value obtained by calculating the sample respectively have the equation of a decision plane
Figure BDA0003754325140000073
Wherein
Figure BDA0003754325140000074
B is the intercept of the decision plane. The conditions for judging whether the pancreatic cancer tissue contains a cancer lesion are as follows:
Figure BDA0003754325140000075
y i indicates tag information, y i =1 indicates that the tissue does not contain pancreatic cancer lesions, and y i =1 indicates that the tissue contains a pancreatic cancer lesion. In order to obtain k and b, a large amount of multivariate parameters and label information are required to be used for training, which is specifically described as the problem of solving the conditional maximum as follows:
Figure BDA0003754325140000076
where s.t. indicates simultaneous satisfaction and m is the total number of samples. Training and testing a large number of samples to finally obtain a decision surface normal vector
Figure BDA0003754325140000077
Intercept b =0.12887999.
It should be noted that although the collagen fiber multivariate parameter calculation of the present invention involves the calculation of four parameters of signal intensity, spatial orientation, direction variance and spatial curvature, the spatial orientation parameter is not used in the mathematical model used in the pancreatic cancer auxiliary examination because the value of the spatial orientation changes due to the change of the reference coordinate system; however, since the calculation of the two parameters of the direction variance and the spatial curvature is performed based on the spatial orientation, the values of the two parameters contain morphological information provided by the spatial orientation, and the values of the two parameters are not affected by the spatial coordinate system.
For a pancreatic tissue which needs to be subjected to pancreatic cancer examination, firstly, two-dimensional/three-dimensional imaging is carried out on collagen fibers in the tissue, then, all structural parameters of the tissue are calculated, parameter values are input into an auxiliary pancreatic cancer examination tool, and according to judgment conditions:
Figure BDA0003754325140000081
it is possible to obtain an examination result of the presence of a potential cancer lesion in the tissue, where y i =1 indicates that pancreatic cancer foci are not contained in tissue, and y i =1 indicates that the tissue contains a pancreatic cancer lesion.
The following detailed description of the embodiments of the present invention will be made with reference to the drawings and specific examples, but the present invention is not limited thereto.
An image analysis method for extracting space curvature characteristics of a fibrous structure based on space orientation change in a neighborhood comprises the following steps:
1) The collagen fibers in the extracellular matrix of the pancreatic tissue are first optically imaged. Imaging tools include second harmonic imaging, two-photon fluorescence imaging, and the like. The procedure is shown in FIG. 1, and second harmonic-two-photon fluorescence images of representative cancer and normal pancreatic tissues are shown in FIGS. 2 (a) and (b).
2) Then, the obtained collagen fiber image of the pancreatic tissue is subjected to calculation of multivariate parameter calculation, and parameters such as signal intensity, spatial orientation, spatial curvature, direction variance and the like of the collagen fiber are calculated respectively, and the flow is shown in fig. 1. The signal intensity parameter is obtained by calculating the average value of collagen fiber signals in the image, so that the content of collagen fibers in biological tissues is quantified; the spatial orientation quantifies the direction of pixel-level resolution of the collagen fibers in the image, and the value is 0-180 degrees; the spatial curvature can quantify the bending degree of the pixel level resolution of the collagen fibers through the change of the spatial orientation of the collagen fibers, and obtain a normalized quantity value, wherein the more the value is, the more bent collagen fibers are represented; the directional variance quantifies the arrangement of collagen fibers with pixel-level resolution by calculating the variance of the spatial orientation of collagen fiber pixels in a neighborhood, the value interval is 0-1, the value of 0 represents the completely parallel collagen fiber arrangement, the value of 1 represents the completely unordered collagen fiber arrangement, and the numerical result of the multivariate parameters is shown in fig. 2 (k). In addition to the multivariate parameter results, a pseudo-color code pattern generated from the numerical results of the different parameters can be output to more clearly assess the structural features of the pancreatic tissue, a representative pseudo-color code pattern being shown in fig. 2 (c-j).
3) Then the obtained multivariate parameters are imported into an auxiliary pancreatic cancer examination tool, and vectors are formed according to numerical results of the multivariate parameters of the collagen fibers
Figure BDA0003754325140000082
Wherein s is i ,v i ,w i And respectively calculating the obtained signal intensity, direction variance and space curvature value. Then, the pancreatic tissue was examined for the presence of a cancer lesion using the following criteria:
Figure BDA0003754325140000091
wherein
Figure BDA0003754325140000092
b＝0.12887999，y i =1 indicates that the tissue does not contain pancreatic cancer lesions, and y i = -1 indicates that the tissue contains pancreatic cancer lesions. In addition to outputting the examination results of normal or cancer, when a plurality of pancreatic tissue images are inputted at a time, the model can also output a scatter diagram so as to better analyze remodeling of collagen fibers caused by pancreatic cancerThe situation is as follows. The flow is shown in fig. 1, and the output result and the corresponding scatter diagram are shown in fig. 3 (a).
The invention provides a pancreatic cancer auxiliary examination tool based on collagen fiber multivariate parameter analysis, which utilizes imaging technologies such as second harmonic and two-photon fluorescence to image collagen fibers of pancreatic tissues and carries out quantitative analysis on structural parameters of the collagen fibers such as spatial orientation, signal intensity, spatial curvature direction variance and the like on the basis of images. FIG. 2 is a schematic diagram of the optical imaging results of an auxiliary pancreatic cancer examination tool based on multivariate parameter analysis of collagen fibers and the output results of multivariate parameter calculation; FIGS. 2 (a) and 2 (b) are second harmonic/two-photon fluorescence representative images of cancer and normal pancreatic tissues; FIGS. 2 (c) - (f) are second harmonic signal intensity images, spatial orientation pseudo-color code patterns, direction variance pseudo-color code patterns and spatial curvature pseudo-color code patterns of collagen fibers of cancer pancreatic tissues calculated by using collagen fiber multivariate parameters; FIGS. 2 (g) - (j) are second harmonic signal intensity images, spatial orientation pseudo-color code patterns, direction variance pseudo-color code patterns and spatial curvature pseudo-color code patterns of collagen fibers of normal pancreatic tissues calculated by using collagen fiber multivariate parameters; FIG. 2 (k) is a histogram comparing the results of image signal intensity, direction variance, and spatial curvature of cancer and normal tissues.
Fig. 2 (a) and (b) show second harmonic/two-photon imaging images of representative cancer and normal pancreatic tissue, respectively. FIGS. 2 (c) (g) are second harmonic signal intensity plots of FIGS. 2 (a) (b), respectively, and FIGS. 2 (d) (e) (f) are pseudo-color-coded plots of spatial orientation, spatial curvature and direction variance, respectively, of collagen fibers in cancerous pancreatic tissue; FIG. 2 (h) (i) (j) are pseudo-colorcode plots of the spatial orientation, spatial curvature and direction variance, respectively, of collagen fibers in normal pancreatic tissue. The invention can analyze the collagen fiber with pixel level resolution, can enable a user to conveniently acquire the structural characteristic information of the collagen fiber by combining a pseudo-color coding technology, can clearly master the collagen fiber remodeling caused by cancer, and is favorable for detecting the micro focus with high-precision information provided by the pixel level resolution. FIG. 2 (k) (l) (m) shows the signal intensity, spatial curvature and directional variance of collagen fibers in normal and cancer pancreatic tissues. Compared with normal tissues, the signal intensity of collagen fibers of cancer pancreatic tissues is higher, the spatial curvature and the direction variance are lower, the rearrangement of the collagen fibers caused by the extracellular matrix abnormality influenced by the cancer progression is revealed, and the information complementarity brought by the multivariate fusion method provided by the invention is shown.
The auxiliary inspection tool for pancreatic cancer is a support vector machine model which takes collagen fiber multivariate parameters (including signal intensity, direction variance and space curvature) as characteristics, can perform auxiliary inspection on pancreatic tissue based on numerical information of the multivariate parameters to detect potential cancer focuses in the tissue, and fig. 3 is a schematic diagram showing comparison between an output result of the auxiliary inspection tool for pancreatic cancer and the multivariate parameters and single parameter inspection tool. FIG. 3 (a) is the examination result outputted by the auxiliary examination tool for pancreatic cancer, which comprises an examination result three-dimensional scattergram constructed by using three parameters of direction variance, spatial curvature and signal intensity as coordinate axes, and the evaluation result of whether a cancer focus exists in the tissue, wherein the accuracy rate of the result is up to 96.2%; FIG. 3 (b) is a receiver operating characteristic curve (ROC) for a support vector machine model under auxiliary examination, with an area under the curve (AUC) of 0.965; FIG. 3 (c) is a two-dimensional scattergram of inspection results constructed with coordinate axes using different parameters; FIG. 3 (d) shows the results of pancreatic cancer examination using single parameters of signal intensity, direction variance and spatial curvature alone, which are less accurate than the results of examination using a plurality of parameters of the three.
In the specific implementation process, if a single pancreatic tissue is examined, numerical results of multiple parameters of the tissue are input, and the auxiliary pancreatic cancer examination tool can directly output the result of whether a cancer focus exists in the tissue. When a plurality of sets of pancreatic tissue images are inputted at one time, a scatter diagram can be obtained by using different parameters as coordinate axes, as shown in fig. 3 (a) and (c), so as to show the remodeling condition of collagen fibers caused by cancer evolution. Fig. 3 (a) also shows the comparison of the examination result output by the auxiliary examination tool for pancreatic cancer with the actual situation, the accuracy is as high as 96.3%, the receiver operating characteristic curve (ROC) is shown in fig. 3 (b), and the area under the curve is as high as 0.965, which shows the high precision, high specificity and sensitivity of the auxiliary examination tool for pancreatic cancer proposed by the present invention. FIG. 3 (c) is a two-dimensional scattergram of inspection results constructed with coordinate axes using different parameters; fig. 3 (d) shows the pancreatic cancer examination result using single parameter of signal intensity, direction variance and spatial curvature alone, and also shows the comparison between the accuracy rates of the multi-parameter auxiliary examination and the single-parameter auxiliary examination, and in the case of performing auxiliary examination on the same series of pancreatic tissues, the accuracy rates of the signal intensity, the direction variance and the spatial curvature alone are respectively 88.4%,94.8% and 90.7%, and the three are all lower than the accuracy rate of the auxiliary examination using the multi-parameter comprehensively by 96.2%, which indicates that the information complementarity provided by different structural features of collagen fibers is fully utilized in the multi-parameter analysis, thereby improving the examination accuracy.
The above is only a preferred embodiment of the present invention, and is not intended to limit the present invention, and various modifications and changes will occur to those skilled in the art. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims (4)

1. An auxiliary pancreatic cancer examination tool based on collagen fiber multivariate parametric analysis is characterized in that the examination tool is used for analyzing multivariate structural characteristics of pancreatic tissue collagen fibers for fusion and establishing an analysis model of the pancreatic tissue collagen fibers for characterization and calculation, and the construction of the examination tool comprises the following steps:
1) Imaging the collagen fibers of the extracellular matrix by pancreatic tissues to be inspected through a second harmonic technology and a two-photon fluorescence technology to obtain an optical image with micron-scale resolution;
2) Extracting structural features of collagen fibers in the optical image;
3) Calculating the multivariate structure parameters of the signal intensity, the space orientation, the space curvature and the direction variance characterization parameters of the structural characteristics of the collagen fibers respectively to obtain a numerical result;
4) Marking the position of the pancreatic tissue subjected to second harmonic/two-photon fluorescence imaging, then performing hematoxylin-eosin staining on the pancreatic tissue for pathological examination, and then labeling whether the marked area contains a cancer focus according to the result of the pathological examination;
5) Constructing a support vector machine model based on the numerical result of the multivariate structure parameter calculation obtained in the step 3) and the label information obtained in the step 4);
6) In the support vector machine model, each independent sample obtains a vector according to the multivariate parameter value thereof as
Figure FDA0003754325130000011
Said multivariate parameters include signal strength, direction variance and space curvature, including
Figure FDA0003754325130000012
Wherein s i ,v i ,w i Respectively obtaining the signal intensity, the direction variance and the space curvature value which are obtained by the calculation in the step 3);
the equation of the decision plane obtained according to the multivariate parameters is as follows:
Figure FDA0003754325130000013
wherein
Figure FDA0003754325130000014
B is the intercept of the decision plane, and the judgment condition of whether the pancreatic cancer tissue contains the cancer focus is as follows:
Figure FDA0003754325130000015
y i indicates tag information, y i =1 tissue not including pancreasCancer focus, and y i = -1 indicates pancreatic cancer foci are contained in the tissue;
7) Training is carried out through a large number of multivariate parameters and label information to obtain k and b, and the method is specifically described as the following problem of solving the condition maximum value:
Figure FDA0003754325130000021
wherein s.t. indicates simultaneous satisfaction, m is the total number of samples;
8) Training to obtain decision surface normal vector
Figure FDA0003754325130000022
An intercept b;
further obtaining an auxiliary pancreatic cancer examination tool based on collagen fiber multivariate parameter analysis.
2. The tool for auxiliary examination of pancreatic cancer based on multivariate parametric analysis of collagen fibers as defined in claim 1, wherein in said step 3), said signal intensity quantifies the content of collagen fibers in biological tissues; the direction of the collagen fibers in the image is quantified by spatial orientation, and the value is 0-180 degrees; the space curvature quantifies the bending degree of the collagen fibers, the value is 0-1, and the larger the value is, the more bent collagen fibers are represented; the direction variance quantifies the arrangement of the collagen fibers, the value interval is 0-1, the value of 0 represents the completely parallel arrangement of the collagen fibers, and the value of 1 represents the completely unordered arrangement of the collagen fibers.
3. The method of claim 2, wherein the signal intensity parameter is obtained by calculating an average value of collagen fiber signals in the image; the spatial orientation parameter is pixel level resolution, and is quantized according to the change of the signal intensity of the collagen fiber in the neighborhood where the pixel is located along different directions; the spatial curvature parameter is pixel-level resolution, and is quantified by analyzing the change of the spatial orientation of collagen fibers in the neighborhood of the pixel; the direction variance parameter is pixel-level resolution, and is quantified by calculating the variance of the spatial orientation of the collagen fibers in the neighborhood of the pixel.
4. The method for pancreatic cancer auxiliary examination based on collagen fiber multivariate parameter analysis as claimed in claim 1, 2 or 3, wherein the step 8) is trained to obtain a normal vector of decision plane
Figure FDA0003754325130000023
Intercept b =0.12887999.
CN202210849955.9A 2022-07-20 2022-07-20 Pancreatic cancer auxiliary inspection tool based on collagen fiber multivariate parameter analysis Pending CN115602311A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210849955.9A CN115602311A (en) 2022-07-20 2022-07-20 Pancreatic cancer auxiliary inspection tool based on collagen fiber multivariate parameter analysis

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210849955.9A CN115602311A (en) 2022-07-20 2022-07-20 Pancreatic cancer auxiliary inspection tool based on collagen fiber multivariate parameter analysis

Publications (1)

Publication Number Publication Date
CN115602311A true CN115602311A (en) 2023-01-13

Family

ID=84842274

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210849955.9A Pending CN115602311A (en) 2022-07-20 2022-07-20 Pancreatic cancer auxiliary inspection tool based on collagen fiber multivariate parameter analysis

Country Status (1)

Country Link
CN (1) CN115602311A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117530659A (en) * 2023-11-06 2024-02-09 浙江大学 Early cervical cancer transformation auxiliary diagnostic tool based on optical coherence tomography

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117530659A (en) * 2023-11-06 2024-02-09 浙江大学 Early cervical cancer transformation auxiliary diagnostic tool based on optical coherence tomography
CN117530659B (en) * 2023-11-06 2024-04-09 浙江大学 Early cervical cancer transformation auxiliary diagnostic tool based on optical coherence tomography

Similar Documents

Publication Publication Date Title
CN109670510B (en) Deep learning-based gastroscope biopsy pathological data screening system
US11561178B2 (en) Artificial fluorescent image systems and methods
US20190087954A1 (en) Pathology case review, analysis and prediction
KR20220119447A (en) Method for supporting pathology diagnosis using AI, and supporting device
US7760927B2 (en) Method and system for digital image based tissue independent simultaneous nucleus cytoplasm and membrane quantitation
EP1470411B1 (en) Method for quantitative video-microscopy and associated system and computer software program product
CN106462767B (en) Inspection device for processing and analyzing images
EP4070232A1 (en) Systems and methods for high throughput drug screening
US7865000B2 (en) Medical image processing apparatus for analyzing properties of living body based on image data thereof
US20130116511A1 (en) Generate Percentage of Positive Cells for Biomarkers by Normalizing and Autothresholding the Image Intensity Produced by Immunohistochemistry Technique
WO2012041333A1 (en) Automated imaging, detection and grading of objects in cytological samples
CN110991536A (en) Training method of early warning model of primary liver cancer
CN111656393A (en) Histological image analysis
CN115602311A (en) Pancreatic cancer auxiliary inspection tool based on collagen fiber multivariate parameter analysis
KR20110072098A (en) Method for normalizing image in digital pathology
Bengtsson Computerized cell image analysis: Past, present, and future
US9785848B2 (en) Automated staining and segmentation quality control
WO2022015728A1 (en) Cellular diagnostic and analysis methods
JP6246978B2 (en) Method for detecting and quantifying fibrosis
WO2023107844A1 (en) Label-free virtual immunohistochemical staining of tissue using deep learning
CN114596256B (en) Image analysis method for extracting space curvature characteristics of fibrous structure based on space orientation change in neighborhood
WO2022272014A1 (en) Computational techniques for three-dimensional reconstruction and multi-labeling of serially sectioned tissue
CN115272211A (en) Breast cancer inspection tool based on collagen fiber multivariate fusion analysis
CN117530659B (en) Early cervical cancer transformation auxiliary diagnostic tool based on optical coherence tomography
Ballerini et al. Fractal analysis of microscopic images of breast tissue

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