Search

CN-122023815-A - Intelligent interpretation method and system for weak positive samples of immune test

CN122023815ACN 122023815 ACN122023815 ACN 122023815ACN-122023815-A

Abstract

The invention relates to the technical field of image recognition, in particular to an intelligent interpretation method and system for a weak positive sample of immune test, comprising the following steps: and acquiring original image data of an immunochromatographic test strip to be detected, extracting an interested region image comprising a detection line region and a quality control line region according to the original image data, and constructing an image gray matrix based on the interested region image. In the invention, the traditional hard threshold binary judgment is converted into soft classification based on fuzzy logic, the gradual change characteristic of weak positive signals is described, and the misjudgment rate caused by signal blurring is reduced. A space constraint mechanism is introduced to filter isolated noise points by utilizing a pixel space topological relation and ensure the continuity of detection signals. And carrying out integral calculation on the detection line area according to the membership distribution matrix to obtain a target signal response integral value, and determining an interpretation result by comparing the target signal response integral value with a preset detection limit threshold value, so as to improve the detection rate of the low-concentration sample.

Inventors

  • CHEN YE
  • CHEN WEI
  • ZHANG JIAN
  • ZHAN FENG

Assignees

  • 常州市肿瘤医院(常州市第四人民医院)

Dates

Publication Date
20260512
Application Date
20260203

Claims (10)

  1. 1. The intelligent interpretation method for the weak positive sample of the immunity test is characterized by comprising the following steps of: acquiring original image data of an immunochromatographic test strip to be detected, extracting an interested region image comprising a detection line region and a quality control line region according to the original image data, and constructing an image gray matrix based on the interested region image; Calculating a local contrast enhancement value of each pixel point by utilizing an improved weber contrast operator according to the gray value of each pixel point in the image gray matrix and the statistical characteristics of a local neighborhood, and generating a weber contrast enhancement image based on the local contrast enhancement value; Constructing a feature vector set containing pixel intensity information and space neighborhood information of the Weber contrast enhancement image, and inputting the feature vector set into a preset space constraint fuzzy C-means clustering model; performing iterative operation on the feature vector set by using an improved objective function in the space constraint fuzzy C-means clustering model until the improved objective function converges to obtain a membership distribution matrix of each pixel point belonging to a positive signal class in the region-of-interest image; And carrying out integral calculation on the detection line area according to the membership distribution matrix to obtain a target signal response integral value, and determining a weak positive interpretation result corresponding to the original image data by comparing the target signal response integral value with a preset detection limit threshold value.
  2. 2. The intelligent interpretation method of the weak positive sample of the immune test according to claim 1, characterized in that according to the gray value of each pixel point in the gray matrix of the image and the statistical characteristics of the local neighborhood, the improved weber contrast operator is utilized to calculate the local contrast enhancement value of each pixel point, and the step of generating the weber contrast enhancement image based on the local contrast enhancement value specifically comprises: defining the position coordinates of the current pixel point in the image gray level matrix, and determining a local sliding window of the current pixel point according to the position coordinates; calculating a pixel gray average value and a pixel gray standard deviation in the local sliding window, and constructing a texture suppression weight factor by using the pixel gray standard deviation; substituting the gray value of the current pixel point, the pixel gray average value and the texture suppression weight factor into the improved Weber contrast operator, and calculating to obtain the local contrast enhancement value of the current pixel point; The calculation formula of the improved weber contrast operator is as follows: ; Wherein, the The representation being located at coordinates The local contrast enhancement value for the current pixel point at, Representing the original gray value of the current pixel point, Representing the pixel gray scale mean within the local sliding window, Representing the standard deviation of the pixel gray scale within the local sliding window, Representing a smoothing constant for preventing denominator from being zero, Representing a preset texture sensitivity adjustment coefficient for adjusting the action intensity of the texture suppression weight factor; And traversing all pixel points in the image gray matrix, and combining to generate the Weber contrast enhancement image according to each calculated local contrast enhancement value.
  3. 3. The intelligent interpretation method of the weak positive samples of the immune test according to claim 1, wherein the step of performing iterative operation on the feature vector set by using an improved objective function in the space constraint fuzzy C-means clustering model until the improved objective function converges to obtain a membership distribution matrix of each pixel point in the region of interest image belonging to a positive signal class specifically comprises: Initializing the number of clustering centers and fuzzy indexes, and randomly initializing a membership matrix; Introducing a spatial neighborhood penalty term to construct the improved objective function, wherein the spatial neighborhood penalty term is used for restraining adjacent pixel points to have similar membership, and adjusting the weight of the spatial neighborhood penalty term by utilizing a characteristic distance weighting factor; iteratively updating the cluster center matrix and the membership matrix by minimizing the improved objective function; Judging whether the change value of the improved objective function between two iterations is smaller than a preset convergence threshold value, if so, stopping the iteration and outputting the final membership distribution matrix; The formula of the improved objective function is as follows: ; Wherein, the A value representing the modified objective function, Representing the total number of pixels in the feature vector set, Representing the total number of cluster categories, Represent the first The pixel point belongs to The membership value of each cluster center, The ambiguity index is represented by a value of the ambiguity index, Represent the first The feature vector of each pixel point, Represent the first The feature vectors of the individual cluster centers, Represents the first The pixel point is connected with the first The euclidean distance between the centers of the clusters, Representing the weight coefficients of the spatial constraint, Represents the first A spatially neighboring set of pixels for a pixel point, Representing the first pixel in the set of spatially neighboring pixels A number of neighboring pixel points, Is defined as the first Feature vector of each pixel point and the first pixel point Euclidean distance between feature vectors of individual cluster centers, i.e , Represents a distance decay constant for adjusting the sensitivity of the spatial constraint as a function of feature distance.
  4. 4. The method for intelligently interpreting an immunometric weak positive sample according to claim 1, wherein in the step of obtaining raw image data of an immunochromatographic test strip to be detected and extracting an image of a region of interest including a detection line region and a quality control line region from the raw image data, the step of obtaining raw image data of the immunochromatographic test strip to be detected specifically includes: Controlling an image acquisition device to shoot the immunochromatographic test strip to be detected under the preset light source condition, and acquiring an initial shooting image comprising a test strip shell and an observation window; Identifying the edge profile of the observation window in the initial shooting image by utilizing an edge detection algorithm, and performing perspective transformation correction according to the edge profile of the observation window to obtain a corrected observation window image; converting the corrected observation window image into a single-channel gray scale image, and taking the single-channel gray scale image as the original image data; And intercepting the detection line area and the quality control line area according to the geometric dimension information of the original image data and the preset relative position proportion, and respectively defining the detection line area and the quality control line area as the region-of-interest image.
  5. 5. The intelligent interpretation method of the weak positive samples of the immune test according to claim 1, characterized in that the step of constructing a feature vector set containing pixel intensity information and spatial neighborhood information of the weber contrast enhanced image, and inputting the feature vector set into a preset spatial constraint fuzzy C-means clustering model specifically comprises: extracting, for each target pixel point in the weber contrast enhancement image, the local contrast enhancement value of the target pixel point itself as a first feature component; Calculating the average value of local contrast enhancement values of the target pixel points in a preset neighborhood range, and taking the average value as a second characteristic component; Calculating the variance of the local contrast enhancement value of the target pixel point in the preset neighborhood range, and taking the variance as a third characteristic component; Combining the first feature component, the second feature component and the third feature component to generate a three-dimensional feature vector of the target pixel point; and collecting the three-dimensional feature vectors of all the target pixel points to form the feature vector set, normalizing the feature vector set, and inputting the feature vector set to the space constraint fuzzy C-means clustering model.
  6. 6. The intelligent interpretation method of immune test weak positive samples as claimed in claim 3, characterized in that, in the step of iteratively updating the clustering center matrix and the membership matrix by minimizing the improvement objective function, the specific process of iterative updating includes: fixing the clustering center matrix, performing partial derivative solution on the membership matrix by using a Lagrange multiplier method, and updating each membership element in the membership matrix according to the current clustering center matrix and the feature vector set; fixing the updated membership matrix, performing bias guide on the improved objective function with respect to the clustering center, calculating a new clustering center position, and updating the clustering center matrix; And when updating the membership matrix, smoothly suppressing the membership of the noise point by using the space neighborhood punishment item, so that the membership of the isolated noise point belonging to the positive signal class is reduced.
  7. 7. The intelligent interpretation method of the weak positive samples of immune test according to claim 1, characterized in that the step of integrating and calculating the detection line area according to the membership distribution matrix to obtain a target signal response integral value, and determining the weak positive interpretation result corresponding to the original image data by comparing the target signal response integral value with a preset detection limit threshold value specifically comprises the following steps: extracting a submatrix corresponding to the detection line region from the membership distribution matrix; Performing binarization mask processing on the submatrix, and filtering out pixel points with membership values lower than a preset background noise threshold value to obtain an effective signal pixel set; Performing space accumulation integration on membership values in the effective signal pixel set, and calculating to obtain a target signal response integral value; acquiring the detection limit threshold value which is determined in advance through blank sample testing; judging whether the target signal response integral value is larger than the detection limit threshold value; If so, generating a weak positive interpretation result which indicates that the sample is positive; if not, generating the weak positive interpretation result which indicates that the sample is negative.
  8. 8. The intelligent interpretation method of immunometric weak positive samples as claimed in claim 1, characterized in that before calculating the local contrast enhancement value for each pixel point using the modified weber contrast operator, the method further comprises: Preprocessing the image gray matrix by using a Gaussian smoothing filter to remove high-frequency random noise, so as to obtain a smoothed image gray matrix; the method comprises the steps that a modified Weber contrast operator is utilized to calculate a local contrast enhancement value of each pixel point, and the local contrast enhancement value is specifically carried out based on the smoothed image gray matrix; the step of constructing the feature vector set including the pixel intensity information and the spatial neighborhood information of the weber contrast enhancement image specifically includes: Preliminary screening is carried out on the Weber contrast enhancement image by using a dual threshold hysteresis judgment algorithm, strong signal points and weak signal points communicated with the strong signal points are reserved, and a candidate signal mask is generated; and extracting features only for the pixel points in the coverage area of the candidate signal mask and constructing the feature vector set so as to reduce the calculated amount of clustering operation.
  9. 9. The method of claim 7, wherein after the obtaining the set of valid signal pixels and before spatially accumulating the membership values in the set of valid signal pixels, the method further comprises: Carrying out geometric analysis on a connected domain formed by the effective signal pixel sets; calculating an aspect ratio parameter and a compactness parameter of the connected domain; judging whether the length-width ratio parameter is positioned in a preset line characteristic interval or not, and judging whether the compactness parameter is larger than a preset shape rule degree threshold value or not; If the connected domain does not meet the requirement of the line characteristic interval or the requirement of the shape regularity threshold, the membership value of the pixel points in the connected domain is forcedly set to zero, and the connected domain is removed from the effective signal pixel set so as to eliminate the interference of nonspecific adsorption spots.
  10. 10. An intelligent interpretation system for an immunoassay weak positive sample, the system comprising: A memory for storing a computer program; A processor for executing the computer program stored in the memory to implement the method for intelligently interpreting an immunoassay weak positive sample according to any one of claims 1 to 9.

Description

Intelligent interpretation method and system for weak positive samples of immune test Technical Field The invention relates to the technical field of image recognition, in particular to an intelligent interpretation method and system for a weak positive sample in immune test. Background The image recognition technology is to acquire, process, analyze and understand image data by using a computer system, and enable a computer to automatically recognize a target object in a specific mode from a complex image background by simulating a biological vision mechanism or using a statistical principle. The prior art relies on global gray threshold segmentation or fixed edge detection algorithms when processing sample data in the field of image recognition. In actual operation, the immunochromatography test strip often has the phenomenon of uneven background material textures or inconsistent light source intensities during acquisition, the traditional global threshold algorithm processes all pixels based on unified standards, when the target signal intensity is close to background noise or lower than a global threshold, effective weak signals are easily filtered out as background noise to cause detection omission, otherwise, high-contrast mixed particles in the background are easily misjudged as positive signals. Therefore, improvements are needed. Disclosure of Invention The invention aims to solve the defects in the prior art, and provides an intelligent interpretation method and system for a weak positive sample in immune test. In order to achieve the purpose, the invention adopts the following technical scheme that the intelligent interpretation method for the weak positive sample of the immunity test comprises the following steps: acquiring original image data of an immunochromatographic test strip to be detected, extracting an interested region image comprising a detection line region and a quality control line region according to the original image data, and constructing an image gray matrix based on the interested region image; Calculating a local contrast enhancement value of each pixel point by utilizing an improved weber contrast operator according to the gray value of each pixel point in the image gray matrix and the statistical characteristics of a local neighborhood, and generating a weber contrast enhancement image based on the local contrast enhancement value; Constructing a feature vector set containing pixel intensity information and space neighborhood information of the Weber contrast enhancement image, and inputting the feature vector set into a preset space constraint fuzzy C-means clustering model; performing iterative operation on the feature vector set by using an improved objective function in the space constraint fuzzy C-means clustering model until the improved objective function converges to obtain a membership distribution matrix of each pixel point belonging to a positive signal class in the region-of-interest image; And carrying out integral calculation on the detection line area according to the membership distribution matrix to obtain a target signal response integral value, and determining a weak positive interpretation result corresponding to the original image data by comparing the target signal response integral value with a preset detection limit threshold value. Preferably, according to the gray value of each pixel point in the image gray matrix and the statistical characteristics of the local neighborhood, calculating the local contrast enhancement value of each pixel point by using an improved weber contrast operator, and generating a weber contrast enhancement image based on the local contrast enhancement value, which specifically includes: defining the position coordinates of the current pixel point in the image gray level matrix, and determining a local sliding window of the current pixel point according to the position coordinates; calculating a pixel gray average value and a pixel gray standard deviation in the local sliding window, and constructing a texture suppression weight factor by using the pixel gray standard deviation; substituting the gray value of the current pixel point, the pixel gray average value and the texture suppression weight factor into the improved Weber contrast operator, and calculating to obtain the local contrast enhancement value of the current pixel point; The calculation formula of the improved weber contrast operator is as follows: ; Wherein, the The representation being located at coordinatesThe local contrast enhancement value for the current pixel point at,Representing the original gray value of the current pixel point,Representing the pixel gray scale mean within the local sliding window,Representing the standard deviation of the pixel gray scale within the local sliding window,Representing a smoothing constant for preventing denominator from being zero,Representing a preset texture sensitivity adjustment coefficient for adjusting the ac