JPS59114667A - Region dividing method - Google Patents

Abstract

PURPOSE:To divide accurately a region of each component element by setting the threshold value which separates the background erythrocyte of nuclear cytoplasm of the component element leucocyte of a blood image in a 3-dimensional color space and in response to the distribution. CONSTITUTION:The minimum value min(g) of a picture (g) is obtained together with the threshold value TgBG corresponding to the fixed percentage value of the upper limit value of the picture. The value of density higher than the center value of histogram by half value width of the density histogram Hb<->BG of a picture (b) of the picture element existing in the region less than the threshold value TgBG of the picture (g) is set as the threshold value TbBG. Then the density histogram HrbBG is obtained for a picture (r-b) which exists in the region more than TgBG of the picture (g). Then threshold values TrbC and TrbN are set based on the histogram HrbBG. The density histrogram is obtained for pictures (r) and (g) which contain no erythrocyte RBC, and the threshold values TrN and TgN which divide the nucleous N and cytoplasm C of the leucocyte are obtained along with the threshold values TrC and TgC which divide the cytoplasm C and the background BG.

Description

DETAILED DESCRIPTION OF THE INVENTION Field of Application of the Invention The present invention relates to a method for dividing an image into regions using color, particularly for dividing a blood cell image into regions of white blood cell nuclei, cytoplasm, red blood cells, and background. The present invention relates to a suitable region dividing method.

"Prior art" The color separation wavelength for blood cell image input is r:56.
0-600nm, g: 510-5500m, b: 4
00 to 430 nm is suitable, but conventional region dividing methods (see Japanese Patent Application No. 185413/1987 for details) do not necessarily give satisfactory results for filter combinations in the above three wavelength ranges. .

For example, g': 550nm, b': 43Qn
If m is selected, even with the conventional area division method, the first
As shown in figure (a), nucleus N of white blood cell, cytoplasm C9 red blood cell RB
C.

It is possible to separate the background BG, but at g: 525 nm.

b: When the wavelength is 420 nm, as shown in FIG. 1 (tb), the distribution of the nucleus N and cytoplasm C of the white blood cell and the red blood cell RBC become close to each other, resulting in disadvantages such as difficulty in dividing the two. Also, when r: 590 nm, g', b' two-dimensional plane,
Although the distributions on the r and b two-dimensional planes are similar and the above-mentioned drawbacks are eliminated, there is a drawback that red blood cells RBC and background BG cannot necessarily be divided at a threshold value T7 corresponding to the threshold value T7.

``Object of the Invention'' The purpose of the present invention is to provide a region segmentation method that eliminates the above-mentioned drawbacks and divides regions more accurately than conventional region segmentation methods based on r,2 images that emphasize the color tone of the cytoplasm. It's about doing.

"Summary of the Invention" Regarding the contamination of red blood cells RBC into the background BG, in g or 5 images where the background and red blood cells are far apart from each other in the concentration distribution relationship in one image, T or T (Fig. 1(b)
)). T obtained from this threshold value and the difference image between the 1st image and the 5th image (Fig. 1(b)
)) By using the two threshold values b in combination, it is possible to divide the white blood cell nucleus N, the white blood cell cytoplasm C9, the background BG, and the red blood cell RBC.

On the other hand, r from which red blood cells RBC were removed in this way.

As shown in FIG. 2, the density histograms of the two images have different distribution states corresponding to background BG, cytoplasm C, and nucleus N depending on the color tone and density of the cytoplasm. For example, in the case of eosinophils, 2
In the density histogram of the image, the distribution of cytoplasm C is greatly different from the distribution of nucleus N, resulting in a bimodal histogram, but in the density histogram of one image, background BG, cytoplasm C, nucleus N are separated into three peaks, and the above A threshold for dividing the three is easily found. In this way, it is easier and more accurate to find the threshold for dividing the above three types from the density histogram of one image, and it is easier and more accurate to find the threshold for white blood cells, which allows for more accurate region segmentation. There are different types of white blood cells that can be divided into regions with high precision.

Therefore, based on the above two facts, we devised a new area division method as shown below.

“Embodiments of the invention” An embodiment of the present invention will be described below with reference to FIG.

Find the minimum value min(g) in 1 and 2 images, and find the threshold value T corresponding to a fixed ratio value (for example, 17 inches) of the minimum value min(g) and the preset upper limit UL of the image. . (FIG. 3(a)) In the case of white blood cells with a low cytoplasm concentration in the dashi 2 image, a portion of the cytoplasm C may be mixed into the area below the threshold value T. (1st
Figure (b)) 2. Pixels existing in the area below the threshold value T in 2 images,) b image f (7) Obtain the density, ゜g, and silica, and calculate the center of the histogram by the half width of this histogram H. A value higher in concentration than the above value is set as the threshold +mT for dividing background BG and red blood cell RBC. (Figure 3(b)) In this case, even if the low-concentration part of cytoplasm C is mixed into the background BG, the concentration of both is almost the same in the 5 images, so T
It is not a factor that changes the

3. Obtain the density histogram H in the r-5 image (the difference image between the 1st image and the 5th image) of the pixels existing in the area above the threshold value T in the 2nd image, and calculate the white blood cell nucleus N. A threshold value T for dividing cytoplasm C and red blood cell RBC and a threshold value 7 for dividing white blood cell nucleus Nb and red blood cell RBC are set. (
Figure 3 (C)) Remove pixels existing in area b, which is below the threshold T in the 4 and 5 images, and in the r-b image, and above the threshold T in the image, that is, red blood cells R and BC. After determining whether the region segmentation is more stable if the threshold value is determined from the density histograms of the first and second images, the white blood cell nucleus N
The threshold values T and T that divide the cytoplasm C and the background BG are determined. (Figure 3 ((phantom, (d') Here, as a method of selecting r, 2 images, the density histogram H (1 = r, g) is divided into six types as shown in Figure 4 ta) to ge). between these six types, from ta) to (f
). That is, if the density histogram of one image is 30-inch and the density histogram of two images is 20-inch, the r image is selected. The eosinophil whose cytoplasm is dark orange in FIG. 2 corresponds to this example. Further, if the density histogram type of one image and the density histogram type of two images are the same, selection conditions are provided for each type.

For example, in the case of type 30, the peak position of the concentration distribution in the center of the histogelium (cytoplasm C) and the left and right distribution (background BG,
The positional relationship of the peak positions of nucleus N), and the peak value of the central distribution and each LV at both ends thereof.

Sometimes two evaluation variables are selected: the ratio of the sum of the UV values. For example, in the case of neutrophils whose cytoplasmic dot tone is light orange in FIG. 2, two images are selected based on the above two evaluation variables.

Here, type classification of density histogram and TN, T
', TN, Tc, TN, Tc (71) For details on how to obtain them, please refer to the specification of Japanese Patent Application No. 57-99431.

According to this embodiment, division of certain blood cells into four regions on the r, b two-dimensional plane is achieved according to the color tone and density of the cytoplasm, and for other blood cells, the threshold value of the r-5 image is The division into four regions is achieved by mapping T , T to the g,b2rb rb dimensional plane. (Fig. 3te), (e')) The present invention is implemented using an electronic computer program. FIG. 5 is a diagram showing an example of the configuration of an apparatus for implementing the method of the present invention.

In the figure, 1 shows a photoelectric conversion device (
2 is a two-image image memory that stores a concentration signal converted by an image pickup tube, etc.), and 5 stores a concentration signal converted from the blood sample through a blue filter (main wavelength around 420 nm). The image memory 3 is a one-image image memory that stores a concentration signal obtained by converting the blood sample through a red filter (main wavelengths around 590 nm and 590 m).

Reads two images from image memory 1, and minimum value detection circuit 4
Find the minimum density min (g) of the two images, input this into the computer (for example, μ computer) 20, and use the above minimum density min (g) and the predetermined maximum density UL of the two images. , T = (UL-min(g))
Calculate xω++n i n (g) and find the threshold value T:
”. However, ω must be a constant, for example 0.17
It is. Next, calculate the threshold T of the threshold circuit 9 using the calculator 2.
Threshold value rr obtained by the above processing according to the signal from 0 B
G, and the logic of the logic circuit 11 is set to be less than or equal to the control signal from the computer 20 T1. Switching circuit 12
is set to use the output of the logic circuit 11 as a control signal for the histogram creation circuit 13 in accordance with the control signal from the computer 20, and the selection circuit 6 receives the control signal from the computer 20 in accordance with the image memory 2. The selection circuit 7 is set to select two images in the image memory 1 in response to a control signal from the computer 20 so that an image can be selected. In the following processing as well, the settings of each circuit are performed using control signals from the computer 20. Then, when images 2 and 5 are simultaneously read out from the image memories 1 and 2, a density histogram H'' of the 5 images in a region including the background B() and part of the cytoplasm C is obtained as the output H of the histogram creation circuit 13. This density histogram H" is input into the computer 20, and the half-value width is calculated to obtain the threshold value T.

Next, the logic of the logic circuit 11 is set to T□ or more, and the selection circuit 6 is set to select the output of the arithmetic circuit 5. Do not change the settings of other circuits. Then, when 2 images, 5 images, and 1 image are simultaneously read out from the image memory 2.3, the density histogram H of the r-5 image in the area excluding the background BG is obtained as the output H of the histogram creation circuit 13. This density histogram H is input into the calculator 20, and the density value of the rb valley with low density is set as a threshold value T, and the value obtained by subtracting a fixed value (for example, 10) from the density value of the valley with high density is set as TN.

Next, the threshold value T of the threshold circuit 9 is set to the threshold value T, the threshold value T2 of the threshold circuit 10 is set to the threshold rb value T, and the logic of the logic circuit 11 is set to T□ or more, 12
Set as the logical sum of the above. The selection circuit 6 selects one image of the image memory 3, the selection circuit 7 selects the output of the subtraction circuit 5, and the selection circuit 8 selects one image of the image memory 2 according to a control signal C from the computer 20. Set to select images. Settings of other circuits cannot be changed. Then, when 5 images and 1 image are simultaneously read from image memories 2 and 3, the area i excluding red blood cells RBC is output H from the histogram creation circuit.
The density histogram H in the region c is determined. This density histogram H"1 is input into the computer 20 and stored.

Next, the selection circuit 6 is set to select two images in the image memory 1, and the settings of other circuits are not changed. And image memo '7
When 2 images, 5 images, and 1 image are simultaneously read out from 1, 213, a density histogram H of i is obtained as the output H of the histogram creation circuit 13 in a region excluding red blood cells RBC. It is determined which of these concentrations H and H should be used to determine the threshold value for dividing the nucleus N and the cytoplasm C1 background BG, and the valley concentration values T 1 , T or T 2 , T 2 are determined.

To divide the region of the nucleus N, for example, based on the above thresholds, if H is selected as the concentration histogram i for threshold calculation, the threshold value of the threshold circuit 9 must be Set the threshold T as T, set the threshold T as the threshold T2 of the threshold circuit 10,
The logic of the logic circuit 11 is T or more.

It is set as a logical product of 1 above and 12 or above. Selection circuit 7
selects one image in the image memory 3, and the selection circuit 8 is set to select the output of the subtraction circuit 5. Then, by simultaneously reading five images and one image from the image memories 2 and 3 and counting the output of the logic circuit 11 through the switching circuit 12, the area of the nucleus N can be determined.

The same applies to the region including the nucleus N and cytoplasm C, and the background BG region.

In the above embodiment, the area division method is implemented using two systems of threshold circuits, a histogram creation circuit, and a computer, but partially parallel processing is possible by providing two systems of histogram creation circuits, for example.

"Effects of the Invention" According to the present invention, the threshold value that separates the components of a blood image, the nucleus N of white blood cells, the background BG of cytoplasm C1, and the red blood cells RBC, is determined by rl.
Since the settings can be made according to the distribution in the g and b three-dimensional color space, the area division of each component can be performed accurately.

FIG. 2 is a diagram for explaining the type of thin duram, and FIG. 5 is a block diagram of an apparatus for realizing an embodiment of the present invention.

1.2.3... Image memory, 4... Minimum detection circuit,
5... Subtraction circuit, 6.7.8... Selection circuit, 9,1
0... Threshold circuit, 11... Logic circuit, 12...
・Switching circuit, 13... Histogram creation circuit, 20・
... Computer lol 12 ((1) Usu Figure 2 Figure 3 Figure 4 Figure S Figure 1 pirg) H Procedural amendment (method) Indication of the case Patent Application No. 223820 of 1988 Invention Relationship with the person who amends the name area division method Patent application Person name
15101 Hitachi Co., Ltd. Written by 'A' Agent Column for detailed explanation of the invention in the specification subject to personal amendment 2 Column for brief explanation of drawings and 7□-2 Drawings.

Contents of correction 1: "d'" on line 14 of page 6 of the specification is corrected to rfJ.

2. Corrected "e'" in the second line of page 8 of the same book by rgJ table correction.

3. In the same book, page 13, line 8, 1. Figure 3 (al~(el, (d'), (e') are'') is corrected to ``Figure 3 (a)~(gl is'').

4. Amend Figure 3 of the drawings as shown in the attached sheet.

83〕A d; 83 old

Claims (1)

[Claims] Using red component, green component, and blue component blood cell images (hereinafter abbreviated as 1 image, 2 image, and 5 images, respectively) inputted by photoelectric conversion means for each wavelength, blood cells are separated from the background and red blood cells from a white blood cell stogram. A difference image between the first image and the fifth image (hereinafter abbreviated as r-b image) excluding the background is obtained.
From the density histogram, find the second threshold that separates the cytoplasm of white blood cells and red blood cells, and the third threshold that separates nucleus and red blood cells, and calculate either the density histogram of two images excluding red blood cells or the density histogram of one image. Select and
A fourth threshold value that separates the background from the cytoplasm and a fifth threshold value that separates the cytoplasm from the nucleus are determined, and a fourth threshold value that separates the cytoplasm from the nucleus is determined. Second. Third
．． A region dividing method characterized in that the region is divided into the four regions using fourth and fifth thresholds.