JPS5887649A - Artifact detector - Google Patents

Abstract

PURPOSE:To improve the processing speed, by detecting the artifact at the stage of input of a picture, and transferring to the next objective processing without the characteristic pickup and discriminating processing at the input and succeeding steps. CONSTITUTION:A sample set on a microscope 31 is irradiated with a light source 32, the transmitted light is received at a photoelectric converter 33, a target like a white blood corpuscle is detected and focused with a microscope stage drive controller 34. Further, a picture to detect a white blood corpuscle is inputted from a photoelectric converter 35, the picture is converted into a digital picture at an A/D converter 36 and inputted to a one-frame memory 37. Simultaneously, the picture is applied to a maximum/minimum detection circuit 39 of a picture processor 38. A threshold value suitable for the detection of the nucleus of the white blood corpuscle is applied to a discrimination controller 42 at the device 38, the threshold value calculated at a controller 42 is applied to a binary circuit 40 and a projection histogram producing circuit 41 of the processor 38 to calculate a parameter defined with the controller 42, the parameter is applied to the controller 34 to detect the articraft at the stage of input.

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention a relates to automatic blood image analysis and turtle artifact detection, and is particularly suitable for detecting as an artifact a plurality of white blood cells (hereinafter referred to as plural) in close fist contact. To make an imitation of a movie.

An object of the present invention is to use an automatic white fluid image analyzer to detect artifacts other than white blood cells, which are not the target of analysis, especially plur.
alt-detecting robe 11t-to provide.

As a method to distinguish white blood cells from artifacts (plural, broken blood cells, staining scum, etc.).

A possible method is to distinguish between the two in the characteristic space of white blood cells using parameters that characterize white blood cells, such as nuclear area, average nuclear concentration, and nuclear cell ratio. but.

Plural has multiple 11 white blood cells with perfect morphology!
If the number of white blood cells exists in close proximity to each other, it can be detected like a body without waiting for the extraction of the above feature parameters, from the information to find the number of white blood cells in a rounded number of -1 elephant human power. be. That is, considering the ratio of the nucleus of a white blood cell to the input extraction region, when a plurality of white blood cells are present, the above ratio is smaller than when a white blood cell with a large plate area or a fill is present.

Information for determining the position is as shown in Figure 1 for each prefecture.In order to determine the position of the white blood cells present on the input screen 1, set an appropriate concentration level and apply the information to the binarized nuclear button 21 horizontally and vertically. It is obtained from the projection histogram 3.4 obtained by counting the projections. In the above projection histogram, a counting threshold value 5 is set in order to remove the platelet equal ratio particles, and the minimum straight line 61 maximum ti? of the horizontal projection histogram is set. , the minimum of the direct projection histogram is 1118. 1ik 9 t'' is calculated, and from this the cut out area 10 in which the nucleus of the white blood cell is contained is calculated.
．． Then, an input region 11t in which all the white blood cells are included is determined by enlarging the cutting region 10t by a certain value.

From the above quantities, 5plural t-detection is performed as follows... (1) However, vpRJ(suI vertical projection histogram 4HPR,J(j
): Horizontal projection histogram 3VMIN! Cutting area horizontal maximum value 8VMAX: Cutting area horizontal direction maximum value 1 to 9HMINS Cutting area horizontal direction minimum value 6HMAX: Cutting area horizontal maximum value 7. However, this filling rate varies greatly depending on the relative position of the number of white blood cells, so this relative position is calculated as the projection area aspect ratio:
When taken on the horizontal axis, the plural
The detection area 21 is determined. This is when the template pitch on the specimen is 0.25 μm.

White blood cells with a denominator of 3,600 or more in the photon ratio definition formula.

This is the result of finding one distribution for plural, and anything that enters here is detected as Iplural and # is rejected.

An embodiment f of the present invention will be described below with reference to FIG. A specimen (not shown) set in a mirror microscope 31 is irradiated with a light source 32, and the transmitted light is received by a light cage lens 33 which has an optical filter on the front.
#! Detect what appears to be white blood cells with a direct 34 and complete focusing. Then, a photoelectric conversion device 35 having an optical filter on the front side inputs an image for determining the white blood cell level III, and an A/D converter 36 converts the image into a digital image. The signal converted into an image is input to the l-frame memory 37. At the same time as the input to the memory 37, the above signal is input to the maximum/minimum concentration detection circuit 39 included in the image processing 113B to obtain the minimum and maximum concentration iiL within one image, and from this, the nuclei of white blood cells are extracted. fI-Identify the cabinet value? It is calculated by the fllj control device 42. Next, the threshold value for extracting the nucleus is set as the reference value of the binarization circuit 400, and the image in the memory 37 is input to the binarization circuit 40 to obtain the binarization pattern of the nucleus. It is input to the output tube projection histogram generation circuit 441 and outputs a projection histogram in the horizontal and vertical directions. The above-mentioned projection sest game is manually inputted into the identification ItlJ# device, the parameters defined by equations (1) and (2) are calculated, and it is determined whether it is a white blood cell or a plural cell. plu
If ral, tP is essential for classifying leukocytes! t41.
In order to skip the harameter extraction process and search for the primary object, the stage is moved via the copying mirror stage drive control device 1k34. This operation is repeated until the number of white blood cells is input as 9 and reaches a constant value of 1.

Here, a small computer can be considered as the identification control device 42. In addition, the photoelectric conversion device [33 can be used in combination with the photoelectric conversion device [35], or the memory 37 can be deleted, the stage can be stopped, and the quality of the specimen itself can be entered as many times as necessary. Although it is possible to consider a configuration in which the identification control device 42 takes over the wear and tear of the &

According to this example, artifacts, especially plurals, can be detected at the input stage of one image, and % image extraction and
Since it is possible to move on to the processing of the primary object without performing identification processing, it is effective in improving the speed and accuracy of escape. In addition, in one example, about 90% of plural was able to be detected.

As described above, according to the present invention, the processing speed can be significantly improved by detecting pi ural at the image input stage without performing 1% feature extraction and classification processing.

[Brief explanation of drawings]

Figure 1 shows the various quantities for determining the internal 1ilLf of leukocytes, and Figure 2 shows the fcIIk formula.
FIG. 3, which shows the ural detection area, is a block diagram showing an embodiment of the present invention. l...Entire screen, 2...Cut to nuclear 211, 3.
. . . Horizontal projection histogram, 4. Vertical projection histogram, 5. Low frequency exclusion aperture, 6. Minimum value in the vertical direction of the cutting area, 7. Maximum value in the horizontal direction of the cutting area. 8... Minimum value of cutting area in vertical direction, 9... Maximum value of cutting area in vertical direction, 10... Cutting area, 11
...input area, 21...plural detection area, 31
...Microscope, 32...Light source, 33...Photoelectric transformation AI
Device, 34...Yu & theory stage drive system ##C11,3
5...Photoelectric conversion device. 36... A/Di Kasegake, 37... l frame and memory. 38...Image geographic table location, 39...Maximum/minimum lateral rotation k. 40... 21 segmentation times &, 41... Projection histogram generation circuit, 42... Identification twlJ# device. VJI figure 71F 'vl 1 to image type ratio figure 3

Claims (1)

[Claims] means for creating a binarized image of the image inputted by the converting means into light, means for obtaining a projection histogram of the binarized image, and a means for determining the projection histogram and a predetermined threshold
k, and the above projection histogram is
[Using the t-calculating means, the projection histogram, the minimum value, and the maximum value, calculate the proportion occupied by the object existing within the area limited by the minimum 111-maximum value and the aspect ratio of the area. 1. An artifact detection device, comprising: means for detecting a plurality of objects that are present in close proximity or contact in the input image using the output of the calculation means.