Quantitative analysis method for white spot infection degree of large yellow croaker's spleen based on machine learning

Abstract

The application discloses a method for quantitatively analyzing the infection degree of white spots in the spleen of Pseudosciaena crocea based on machine learning, and relates to a threshold segmentation technology of machine learning. 1) Data acquisition: dissect the Pseudosciaena crocea infected with internal visceral white spot disease, use a high-definition camera to shoot the spleen, and acquire pictures of the white spots in the spleen to be quantitatively analyzed; 2) creation of an interactive segmentation GUI interface: used for reading the pictures of the white spots in the spleen of the Pseudosciaena crocea and automatically converting the pictures into binary images; 3) running of a main program: the main program is written by a python code and is used for the GUI interface to display the selected pictures on the interactive segmentation GUI interface, so that the pictures are provided for contrast segmentation in the next step; 4) picture processing: selecting the pictures, drawing a frame to select a region to be segmented, adjusting a segmentation threshold to acquire a relatively accurate segmentation result, and realizing visual interactive segmentation of the white spots in the spleen; and 5) quantitative analysis: realizing the quantitative analysis of the infection degree of the white spots in the spleen of the Pseudosciaena crocea by the proportion of the white spots in the spleen.

Description

TECHNICAL FIELD

[0001] The present application relates to threshold segmentation technology of machine learning, and in particular to a machine learning-based quantification analysis method for white spot infection degree of large yellow croaker spleen. BACKGROUND

[0002] Aquaculture is the fastest growing food production sector in the world. With the increasing demand for protein and nutrient products, it is imperative to increase aquaculture production.

[0003] Large yellow croaker is an important marine fish in the southeast coastal areas of China. With the increasing of the breeding density, the lack of effective prevention measures leads to the widespread prevalence of various fish diseases, among which the internal organ white spot disease is the main seasonal disease, which brings great economic losses to the large yellow croaker culture industry and has become a bottleneck restricting the sustainable development of freshwater aquaculture. This disease mainly occurs in spring and autumn every year, belongs to cold water bacterial disease, and can occur in juvenile and adult fish. When the disease occurs, the liver, spleen and kidney can be observed to have obvious white nodules with a diameter of 1-2mm. In recent years, genetic breeding work has been widely concerned, and quantification of infection degree is crucial for the breeding work of internal organ white spot disease.

[0004] In order to improve the specific disease resistance of fish and better understand the mechanism of the disease, researchers need an accurate and objective method to measure the subtle differences of pathological phenotype. The severity of internal organ white spot disease is usually estimated by the human eye on site to a certain level or percentage. However, visual estimation shows that there are human errors and differences between the repeated evaluation of the severity of internal organ white spot disease by the same observer and the evaluation of the severity of internal organ white spot disease by different observers. Therefore, it is particularly important to develop an objective, standard and rapid quantification analysis process for spleen white spot infection degree. SUMMARY

[0005] The present application aims to solve the problem that the quantification analysis of large yellow croaker spleen white spot is entirely based on subjective estimation and there is no objective standard evaluation process, and provides a machine learning-based quantification analysis method for large yellow croaker spleen white spot infection degree, which realizes objective, standard and rapid quantification analysis of spleen white spot by accurate interactive segmentation of spleen and white spot area based on threshold segmentation method of machine vision.

[0006] The present application relates to threshold segmentation technology of machine learning, which realizes the segmentation of spleen and white spot through interactive semi-automatic visual segmentation; then the areas (pixel points) of white spot and spleen are calculated respectively, and the proportions of them are further calculated, so as to objectively measure the infection degree of large yellow croaker spleen white spot; the present application realizes standard, rapid and visual analysis of large yellow croaker spleen white spot infection degree.

[0007] The main processes of this invention include:

[0008] Data acquisition: Obtain images of white spots on the spleen to be quantified and analyzed;

[0009] The data acquisition steps first involve dissecting the large yellow croaker infected with visceral white spot disease, placing the dissected spleen on a background plate of pure color, and then using a high-definition camera to photograph the spleen.

[0010] The main program runs to open the visual interactive segmentation interface;

[0011] The main program is written in Python. When the main program runs, it calls libraries such as OpenCV to open a visual interactive segmentation interface and displays the selected image on the interactive interface for further comparison and segmentation.

[0012] Furthermore, in order to reduce the influence of surrounding irrelevant pixels on the segmentation results, the interactive segmentation interface allows users to select the white dot region of the spleen as the final region to be segmented by clicking and dragging a rectangular box with the left mouse button.

[0013] Interactive visual segmentation of white spots on the spleen.

[0014] The interactive visualization and segmentation of spleen white spots will be displayed in the left half of the interactive interface after the selected area is selected. A progress bar at the top of the interface represents the segmentation threshold. The segmentation threshold can be adjusted to control the segmentation results of the spleen and white spots. By comparing the original image in the left half with the threshold progress bar at the top, the segmentation result will be displayed in the right half of the interactive segmentation interface.

[0015] Furthermore, when the white dots on the left and right sides of the interactive segmentation interface are the same or nearly the same, the number of pixels in the spleen is calculated, which is the spleen area. The number of pixels in the white dots is the white dot area. The area ratio of the white dot area to the spleen area is the area ratio. This ratio can measure the severity of rhubarb disease. According to the box plot results, a spleen infection threshold (Ratio) of approximately 0.02 is correlated with mortality.

[0016] In order to be more intuitive and convenient, the application develops an interactive segmentation GUI interface. Through the interface, the photographed large yellow croaker spleen white dot picture can be read and automatically converted into a binary image. The uppermost part of the GUI interface is a threshold progress bar and a current threshold value (threshold) display, which can be adjusted by dragging the progress bar to adjust the segmentation threshold. The binary image is displayed in the lower left corner of the GUI, and the segmentation result of the spleen and white dot under the current segmentation threshold is displayed in the lower right corner. The text part between the threshold progress bar and the image display shows the white dot area (White_dot_Area), the spleen area (Spleen_Area) and the ratio of the white dot to the spleen area (Ratio). The application realizes accurate interactive segmentation of the spleen and white dot area by a threshold segmentation method based on machine vision.

[0017] The application realizes the objective standard quantitative analysis of the large yellow croaker spleen white dot infection degree. Through the white dot spleen ratio, the large yellow croaker spleen white dot infection degree can be objectively and accurately quantified. Previously, the large yellow croaker spleen white dot infection degree was based on human eye observation and subjective evaluation. According to the box plot result, there is a very significant difference (P<0.001) between the Ratio characteristics of the survival individuals and the death individuals. In addition, the result shows that the spleen infection threshold (Ratio) of about 0.02 is related to the mortality rate. BRIEF DESCRIPTION OF DRAWINGS

[0018] Figure 1 is a method flowchart of the application.

[0019] Figure 2 is a collected original picture.

[0020] Figure 3 is an interactive segmentation interface with a rectangular frame selected in the region to be segmented.

[0021] Figure 4 is an interactive segmentation threshold value of 0.

[0022] Figure 5 is an interactive segmentation threshold value of 102.

[0023] Figure 6 is a box plot of the white dot spleen ratio of the large yellow croaker survival individuals and death individuals. DETAILED DESCRIPTION

[0024] The following examples will further illustrate the application with reference to the accompanying drawings.

[0025] Example 1:

[0026] As Figure 1As shown, the machine learning-based quantification method for the white spot infection degree of the large yellow croaker spleen includes main program running, selecting pictures to be analyzed, opening the spleen white spot visualization interactive segmentation interface, frame selecting the region to be segmented, adjusting the segmentation threshold to obtain a relatively accurate segmentation result, and the program automatically quantifying the spleen area, white spot area, and the white spot infection degree of the large yellow croaker spleen (white spot proportion of the spleen) according to the segmentation result.

[0027] The spleen white spot pictures are obtained by first dissecting the large yellow croaker infected with the internal organ white spot disease, placing the dissected spleen on a color-pure background plate, and then using a high-definition camera to take pictures of the spleen. The results are as shown in Figure 2 .

[0028] The main program is written in python language and mainly functions to open the GUI interface, which is composed of the cv2.imread command for reading pictures and the cv2.imshow command for displaying pictures in opencv. In addition, in order to realize interactive segmentation, the main program also includes the mouse callback function cv2.setMouseCallback. The mouse callback function can monitor the mouse action, which is prepared for the subsequent frame selection of the region of interest.

[0029] Further, in order to reduce the interference of too many surrounding pixels, as shown in Figure 3 , the rectangular frame is used to select and crop the spleen white spot region (region of interest) to be segmented. Specifically, the mouse callback function of opencv can monitor the mouse behavior. When the left mouse button is clicked, the callback function obtains the position information of the mouse at this time, drags the mouse, and dynamically displays the rectangular frame. When the left mouse button is released, the rectangular frame selection is completed, and the callback function obtains the position information of the mouse at this time. The rectangle is the region of interest. Then the left upper corner coordinates and the right lower corner coordinates of the rectangle obtained by the callback function are used to automatically crop the original picture, and the region of interest is binarized, so as to reduce the interference of too many surrounding pixels.

[0030] Spleen white dot visualization interactive segmentation, in order to use the threshold segmentation method in machine learning, the interactive segmentation process first needs to perform a grayscale operation on the picture. The selected binary grayscale image of the region to be segmented is displayed in the lower left part of the GUI interface, and the segmentation result is displayed in the lower right part of the GUI interface. There is a threshold segmentation progress bar on the top of the GUI interface, and the threshold on the left side of the progress bar displays the segmentation threshold. The specific operation of binaryzation is that the values greater than the segmentation threshold are reset to 255, and the values less than the segmentation threshold are reset to 0. The segmentation result of the spleen and white dot can be controlled by adjusting the segmentation threshold. By adjusting the threshold of the threshold progress bar on the top of the GUI interface, the segmentation result will be displayed in the lower right part of the GUI interface. The text part between the threshold segmentation progress bar and the display picture in the GUI interface is the quantitative result of the spleen, white dot and ratio, wherein White_dot_Area represents the white dot area, Spleen_Area represents the spleen area, and Ratio represents the ratio of the white dot to the spleen area.

[0031] Specifically, the white dot area White_dot_Area is calculated, first the connected domain of the segmented binary image is found, and the command used is cv2.findContours. These small connected domains are the white dot regions, then the areas of these small connected domains are calculated by the cv2.contourArea function, and finally the sum of these areas is White_dot_Area. The spleen area Spleen_Area is calculated by directly calculating the number of black pixels (pixel value is 0) in the segmentation result binary image and adding White_dot_Area. The infection degree Ratio = White_dot_Area / Spleen_Area. These three values are automatically calculated and displayed on the GUI interface during the segmentation process, as shown in Figure 4 and Figure 5 . Wherein Figure 4 is the segmentation result and quantitative analysis result when the segmentation threshold threshold is 0 (initial value). It can be seen that when threshold is 0, no part is segmented, White_dot_Area is 0.0, Spleen_Area is 0.0, and Ratio is Nan. Figure 5 and Figure 4 are the same picture with different segmentation thresholds, and the segmentation threshold threshold is 102 (optimal value), White_dot_Area is 16381.5, Spleen_Area is 89730.5, and Ratio is 0.18256. Figure 6 is the box plot of the white dot spleen ratio of the survival individuals and the dead individuals of Pseudosciaena crocea, the horizontal coordinate BS represents the binary trait, and the vertical coordinate Ratio is the white dot spleen ratio.Figure 6 The connection between the upper died and the survival is linked with 4 "*" representing the correlation, and the more "*" represents the stronger correlation. The results show that there is a significant difference in the Ratio characteristics between the surviving individuals and the dead individuals (P<0.001). It can be seen that the spleen infection threshold (Ratio) of about 0.02 is related to the mortality, Figure 5 The Ratio of the shown individual is greater than 0.02, which is a spleen white spot infection lethal individual.

[0032] The present application first provides a semi-automatic visualization and quantitative analysis method for the infection degree of the white spot of the large yellow croaker spleen. The main steps include data acquisition, main program running, selecting pictures, opening the spleen white spot visualization interactive segmentation interface, drawing a frame to select the region to be segmented, adjusting the threshold to obtain a relatively accurate segmentation result, automatically quantitatively analyzing the segmentation result of the spleen white spot, including the spleen area, the white spot area, and the proportion of the white spot to the spleen.

[0033] The technical innovation point of the present application is:

[0034] 1. Applying machine vision to the quantitative analysis of the infection degree of the white spot of the large yellow croaker viscera;

[0035] 2. Developing an interactive segmentation interface for the white spot of the large yellow croaker spleen, which can dynamically adjust the segmentation result;

[0036] 3. Realizing the quantitative analysis of the spleen area, the white spot area, and the proportion of the white spot to the spleen, and obtaining the spleen infection threshold (Ratio) of about 0.02 related to the mortality, which is more objective, standard, and accurate than the previous manual subjective observation of the infection degree of the white spot of the large yellow croaker spleen.

[0037] Finally, it should be pointed out that those skilled in the art can make improvements or changes according to the above description, and all these improvements and changes should be within the protection scope of the appended claims of the present application.

[0038] The above is an exemplary description of the present application in combination with the drawings. Obviously, the implementation of the present application is not limited by the above manner. Any improvement or change using the method concept and technical solution of the present application, or directly applying the concept and technical solution of the present application to other occasions without improvement, is within the protection scope of the present application.

Claims

1. A method for quantitatively analyzing the degree of white spot infection in the spleen of Pseudosciaena crocea based on machine learning, characterized in that The application relates to a white dot visualization interactive segmentation method for a fish spleen, and belongs to the technical field of image processing. 1) data acquisition: a large yellow croaker infected with visceral white dot disease is dissected, and the dissected spleen is placed on a color-pure background plate, a high-definition camera is used to shoot the spleen, and a spleen white dot picture to be quantitatively analyzed is acquired; The data acquisition is used for more clearly shooting the white dot, increasing a light source, and supplementing light; 2) creating an interactive segmentation GUI interface: the interface is used for reading the shot large yellow croaker spleen white dot picture, and automatically converting the picture into a binary image; The uppermost part of the GUI interface is a threshold progress bar and a current threshold value threshold display, the segmentation threshold value is adjusted by dragging the progress bar, the binary image is displayed in the lower left part of the GUI interface, and the segmentation result of the spleen and the white dot under the current segmentation threshold value is displayed in the lower right part of the GUI interface; the text part between the threshold progress bar and the image display is the white dot area White_dot_Area, the spleen area Spleen_Area and the white dot area ratio Ratio of the spleen area; 3) running a main program: the main program is written by a python code, is used for opening the interactive segmentation GUI interface created in step 2), and displays the selected picture on the interactive segmentation GUI interface, so that the picture is compared and segmented in the next step; The main program is written by using a python language, is used for opening the GUI interface, and is composed of a picture reading command cv2.imread in opencv and a picture display command cv2.imshow; in order to realize the interactive segmentation, the main program further comprises a mouse callback function cv2.setMouseCallback; the mouse callback function is used for monitoring the action of a mouse, and prepares for subsequent frame selection of a region of interest; 4) picture processing: selecting a picture, displaying the selected picture on the interactive segmentation GUI interface, frame selecting a region to be segmented, adjusting a segmentation threshold value to acquire a relatively accurate segmentation result, realizing spleen white dot visualization interactive segmentation; specifically, the spleen white dot visualization interactive segmentation interface is displayed in the left half part of the interactive interface after the region to be segmented is framed, and a progress bar is arranged in the upper part of the interactive interface, the progress bar represents a segmentation threshold value, the segmentation result of the spleen and the white dot is controlled by adjusting the segmentation threshold value; the original picture in the left half part is compared, the threshold value progress bar in the upper part is adjusted to adjust the threshold value, and the segmentation result is displayed in the right half part of the interactive segmentation interface. The picture processing is used for reducing the influence of surrounding irrelevant pixels on the segmentation result by dragging a rectangular frame to select the white spot area of the spleen as the final segmentation area; specifically, the rectangular frame is used to select and crop the white spot area of the spleen as the region of interest; the mouse callback function cv2.setMouseCallback is used to monitor the mouse behavior; when the left mouse button is clicked, the callback function obtains the position information of the mouse at this time, drags the mouse, and dynamically displays the rectangular frame; when the left mouse button is released, the rectangular frame selection is completed, and the callback function obtains the position information of the mouse at this time; the rectangular frame is the region of interest, and then the original image is automatically cropped using the left upper corner coordinates and the right lower corner coordinates of the rectangular frame obtained by the callback function, and the region of interest is binarized to reduce the interference of too many surrounding pixels; The spleen white spot visualization interactive segmentation uses a threshold segmentation method in machine learning, and the interactive segmentation process needs to perform a grayscale operation on the picture first; the grayscale image of the selected segmentation area after binarization is displayed in the lower left part of the GUI interface, and the segmentation result is displayed in the lower right part of the GUI interface; a threshold segmentation progress bar is arranged on the upper part of the GUI interface, and the threshold on the left side of the threshold segmentation progress bar displays the segmentation threshold. 5) Quantitative analysis: further quantitative analysis is performed on the segmentation result of the white spot of the spleen, and the infection degree of the white spot of the spleen of the large yellow croaker is quantified through the white spot of the spleen; When the left and right halves of the interactive segmentation interface are the same or close to each other, the pixel point number of the spleen is calculated, which is the area of the spleen; the pixel point number of the white spot is calculated, which is the area of the white spot; the area ratio of the white spot to the pixel point is the area ratio, which is used to measure the infection degree of the white spot disease of the large yellow croaker.

2. The method according to claim 1, wherein the method is characterized by In step 4), the specific operation of the binarization is that the values greater than the segmentation threshold are reset to 255, and the values less than the segmentation threshold are reset to 0.

Images