k (.) is the kernel density function (non-negative, integral 1, conforming to probability density properties, and mean 0); h >0 is a smoothing parameter, called bandwidth. The kernel density estimation is that the data + bandwidth of each data point is taken as the parameter of the kernel function through the kernel function to obtain n kernel functions, the estimation function of the kernel density is formed through linear superposition, and the kernel density probability density function is obtained after normalization.

In short, when the probability distribution of a certain event is known, if a certain number appears during observation, the probability density of the number is considered to be high, the probability density of the number closer to the certain number is also relatively high, and the probability density of the number farther away from the certain number is relatively low. Based on this consideration, K can be used to fit the far, small, near and large probability density we imagined for the first number in the observation. The probability density distribution functions fitted to each observation are averaged. If some number is important, a weighted average can be taken.

Based on FUJI software and an R language platform, the applicant establishes a set of methods for quantifying glioblastoma multiforme invasiveness by using a Kernel Density Estimation (KDE) algorithm. The invention utilizes the nuclear density estimation algorithm and visualization function encapsulated by the R language ggplot2 package and the MASS package to deduce the nuclear density estimation diagram of the nuclear distribution. The invention can help to effectively judge the invasion condition of the tumor from pathological sections in scientific research work, count the tumor nuclear density of each area by utilizing the nuclear density, roughly judge the invasion trend of the tumor from a visual map, and further reflect the invasion capacity of the tumor from the sections.

Compared with the traditional method for judging the invasion front and quantifying the invasion by mechanically repeating film reading, the method has the advantages that the computer algorithm is adopted, so that the efficiency is high, the time is fast, the accuracy is high, and a large amount of repetitive work such as a large number of film reading and counting is reduced; compared with nuclear magnetic resonance image judgment, the method has high accuracy, overcomes the defect that tumor cells cannot be found microscopically by nuclear magnetic resonance technology, has higher resolution than nuclear magnetic resonance, and has more accurate interpretation on tumor cells in an affected area.

Drawings

In order to make the object, technical scheme and beneficial effect of the invention more clear, the invention provides the following drawings for explanation:

FIG. 1 is a flow chart of the present invention;

FIG. 2 is a schematic view of a process for quantifying an immunodeficient mouse graft tumor by a cell nucleus density method, wherein A is HE for displaying the immunodeficient mouse graft tumor, B is a binarized image for generalizing each cell nucleus into a binarized point, and C is a generated two-dimensional cell nucleus density estimation image;

fig. 3 is a flow chart of quantifying the invasion front of glioblastoma cell by the cell nucleus density method, where a is an immunofluorescence original picture, blue signals represent cell nuclei, B is a generated two-dimensional nuclear density estimation picture, and different nuclear densities are represented by rainbow color block gradients.

Detailed Description

Preferred embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

Example 1:

as shown in fig. 1, a visualization method for quantifying glioma invasiveness based on a nuclear density function specifically includes the following steps:

(1) converting each cell nucleus in the glioma tissue section into a binary image;

the specific method of the step (1) is as follows:

(1-1) acquiring a glioma tissue section image by using image acquisition equipment, and importing the image into ImageJ software; (A in FIG. 2)

And (1-2) opening the picture, running the Macro code of FUJI for batch processing, selecting a color threshold value, and selecting a Convert to Mask column to obtain a binary image.

In the step (1-1), the glioma tissue section image is an HE or immunofluorescence image.

The specific process of the step (1-2) is as follows:

(1-2-1) running FUJI software, and opening an HE image or a confocal scanning image; open ("…/…/…");

(2) Performing nuclear density estimation of cell nucleus to obtain nuclear density information map of cell

The specific method of the step (2) is as follows:

(2-1) removing impurity points from the binary image and processing the binary image by a watershed method to obtain positioning information of cell nucleuses; (B in FIG. 2)

(C in FIG. 2)

In the step (2-1), if the picture is an HE stained picture, the hematoxylin staining depth can affect the screening effect, and the color threshold value needs to be adjusted repeatedly to an optimal value; if necessary, the cell nucleus can be adjusted in blocks until all cell nuclei are selected; in the case of fluorescent pictures, only pictures of blue channels were extracted using Photoshop.

The flow of the step (2-1) is as follows:

In the step (2-2), the kernel density information map may be set as a two-dimensional statistical histogram or a two-dimensional kernel density histogram, where the two-dimensional statistical histogram may be set as a square or a hexagon in a visualization type, and the two-dimensional kernel density histogram may be set as a grid or a polygon in a visualization type. Two-dimensional kernel density histogram method referring to the figures of this example, in addition, if one wants to obtain a simple two-dimensional statistical histogram to reflect the kernel density of a region. The picture region can be divided into many small square lattices, the number of the lattices is realized by setting the bin value, the cell nucleus frequency in the lattices is counted by using the getbin 2d () of ggplot2, and the visualization step is executed.

In the step (2-2), visualizing by using a stat _ dense _2d () function of an R language ggplot2 package, wherein the core density estimation of the cell nucleus can be carried out by combining a kde2d () function in a MASS package, and the visualization is carried out by using the result; the operation steps and the execution codes are mainly executed:

# 2-2-1 scatter plot function

ggplot(df,aes(x,y))；

+stat_density_2d_filled(geom＝"raster",aes(fill＝..density..),contour＝F)

+scale_fill_gradientn(colours＝colormap,limits＝c(2e-07,2e-06),

breaks＝seq(2e-07,2e-06,by＝1e-09),oob＝scales::squish)。

(2-2-4) setting theme parameters

+theme_classic()；

+theme(panel.background＝element_rect(fill＝"white",

colour ═ black ", size ═ 0.25), and # set the background of the panel

axis.line＝element_line(colour＝"black",size＝0.25),

# set coordinate axes color, size

axis.title＝element_text(size＝13,face＝"plain",color＝"black"),

# setting coordinate Axis heading text size, color

axis.text＝element_text(size＝12,face＝"plain",color＝"black"),

legend.position＝"right")

# sets coordinate axis scale text size, color.

(3) Judging the invasion trend of the tumor through a visualization result

The specific method of the step (3) is as follows: and judging the distribution density of the tumor cells in the area according to the shade of the color of the nuclear density information map, and roughly judging the invasion trend of the tumor.

Example 2:

another immunofluorescence image of human glioblastoma was taken and visualized as in example 1. The results show that this system reflects well the density of nuclei in the local area (fig. 3) and that the trend of tumor invasion can be accurately identified from fig. 3. The area with high nuclear density is the tumor core area at the upper right corner of the figure, and the tumor invasion area is at the left side, so that the color blocks with dark colors can be obviously found to cover the invasion front edge, which represents that the density of the area is relatively high.

Finally, it is noted that the above-mentioned preferred embodiments illustrate rather than limit the invention, and that, although the invention has been described in detail with reference to the above-mentioned preferred embodiments, it will be understood by those skilled in the art that various changes in form and detail may be made therein without departing from the scope of the invention as defined by the appended claims.

Claims

1. A visualization method for quantifying glioma invasiveness based on a nuclear density function is characterized by comprising the following specific steps:

(3) finally, judging the invasion trend of the tumor through a visualization result;

the specific method of the step (2) is as follows:

(2-2) introducing the nuclear localization information obtained in the step (2-1) into Rstudio, downloading ggplot2 and RColorBrewer packages, and performing mapping by using various nuclear density estimation algorithm-based parameters of ggplot2 to obtain a nuclear density information map of the cells;

the flow of the step (2-1) is as follows:

(2-1-1) removing impurities, a median filter, replacing the pixel with an average pixel value of 3 x 3 pixels in its neighborhood;

(2-1-2) performing an expanding operation, and then corroding to smooth the object and fill the vacancy;

(2-1-3) segmenting the connected cell nucleuses by a watershed method;

(2-1-4) analyzing the Particles, clicking Analyze and Analyze Particles, selecting record starts, and obtaining a result XM, wherein YM is a centroid coordinate, and File and Save as centroid coordinate results are put into an Excel table;

in the step (2-2), the kernel density information map is a two-dimensional statistical histogram or a two-dimensional kernel density histogram, wherein the two-dimensional statistical histogram is a square or a hexagon, and the two-dimensional kernel density histogram is a grid or a polygon;

# 2-2-1 scatter plot function

ggplot(df,aes(x,y))；

# (2-2-2) drawing a 2D density map function, wherein, the fact that the "raster" indicates that a classic grid map is generated, and the "polygon" can also be set to generate a polygon profile map; use. contour represents the contour line;

+stat_density_2d_filled(geom＝"raster",aes(fill＝..density..),contour＝F)

# (2-2-3) scale _ fill _ gradientn () setting to fill a color in a density map according to a corresponding density estimate; colormap sets a gradient color interval; limits sets the visual density value range; the break represents that the break parameter redefines the color bar range and divides the color range according to the break range; oob ═

scales, namely squish indicates that the color exceeding the threshold range is filled according to the limit value; these steps can be set to the same measure among different groups, and the comparison of the invasiveness among the groups is convenient;

+scale_fill_gradientn(colours＝colormap,limits＝c(2e-07,2e-06),

breaks＝seq(2e-07,2e-06,by＝1e-09),oob＝scales::squish)；

(2-2-4) setting theme parameters

+theme_classic()；

(2-2-5) setting the size of the background coordinate axis of the picture and the size of the characters

+theme(panel.background＝element_rect(fill＝"white",

colour ═ black ", size ═ 0.25), and # set the background of the panel

axis.line＝element_line(colour＝"black",size＝0.25),

# set coordinate axes color, size

axis.title＝element_text(size＝13,face＝"plain",color＝"black"),

# setting coordinate Axis heading text size, color

axis.text＝element_text(size＝12,face＝"plain",color＝"black"),

legend.position＝"right")

Setting the size and color of a coordinate axis scale text;

the specific method of the step (1) is as follows:

(1-2) opening a picture, operating Macro codes to perform batch processing, selecting a color threshold value, and selecting a Convert to Mask column to obtain a binary image;

in the step (1-1), the glioma tissue section image is an HE or immunofluorescence image;

the specific process of the step (1-2) is as follows:

(1-2-1) operating Fiji, and opening an HE image or a confocal scanning image;

(1-2-2) automatically selecting a color threshold; or a manual selection;

(1-2-3) setting a color interval range, and needing manual adjustment; because the hematoxylin staining time and the freshness of the dye can influence the color depth of the cell nucleus, the staining among different batches needs to adjust the threshold value so as to select all the cell nuclei in the picture as much as possible;

(1-2-4) after the cell nucleus area is selected, changing the image into a binary image, namely converting the image into a black and white plate, and converting the black and white plate into values of black foreground color and white background color so as to facilitate various subsequent operations of the binary image;