JPH01147365A - Leukocyte classifying device - Google Patents
Leukocyte classifying deviceInfo
- Publication number
- JPH01147365A JPH01147365A JP30561087A JP30561087A JPH01147365A JP H01147365 A JPH01147365 A JP H01147365A JP 30561087 A JP30561087 A JP 30561087A JP 30561087 A JP30561087 A JP 30561087A JP H01147365 A JPH01147365 A JP H01147365A
- Authority
- JP
- Japan
- Prior art keywords
- red
- density
- green
- image
- density width
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 210000000265 leukocyte Anatomy 0.000 title claims description 28
- 210000003979 eosinophil Anatomy 0.000 claims abstract description 19
- 210000004369 blood Anatomy 0.000 claims abstract description 18
- 239000008280 blood Substances 0.000 claims abstract description 18
- 238000012758 nuclear staining Methods 0.000 claims description 21
- 210000003743 erythrocyte Anatomy 0.000 claims description 11
- 230000015654 memory Effects 0.000 abstract description 10
- 210000004027 cell Anatomy 0.000 abstract description 2
- 230000003287 optical effect Effects 0.000 abstract description 2
- 230000002950 deficient Effects 0.000 abstract 1
- 210000004940 nucleus Anatomy 0.000 description 18
- 238000010586 diagram Methods 0.000 description 7
- 238000000034 method Methods 0.000 description 6
- 239000008187 granular material Substances 0.000 description 5
- 230000003595 spectral effect Effects 0.000 description 5
- 238000010186 staining Methods 0.000 description 5
- 238000006243 chemical reaction Methods 0.000 description 3
- 238000000605 extraction Methods 0.000 description 3
- 238000010521 absorption reaction Methods 0.000 description 2
- 230000005779 cell damage Effects 0.000 description 2
- 208000037887 cell injury Diseases 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 210000000601 blood cell Anatomy 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
- 238000011179 visual inspection Methods 0.000 description 1
Landscapes
- Investigating Or Analysing Biological Materials (AREA)
Abstract
Description
【発明の詳細な説明】
〔産業上の利用分野〕
本発明は、白血球の分類に有効な特徴量に関するもので
、特に核染色性不良の好酸球と核染色性良好の白血球、
白血球こわれの分類識別に好適な白血球分類装置に関す
る。[Detailed Description of the Invention] [Industrial Application Field] The present invention relates to feature quantities effective for classifying leukocytes, and in particular eosinophils with poor nuclear staining, leukocytes with good nuclear staining,
The present invention relates to a white blood cell classification device suitable for classification and identification of white blood cell damage.
特開昭５９−１１４６６７、特開昭６０−２５０２５５
に記載のように、従来血液の塗抹・染色標本を顕微鏡で
拡大した血液像の画像処理を行う場合、青を除く可視域
（約４５０ｎｗｎ〜７００ｎｙｎ）では、白血球の核が
最も濃度の高い領域であるという仮定をもとに、血液像
の領域分割（画像処理により血液像を白血球の核、白血
球の細胞質、赤血球、血球の存在しない部分の４領域に
分けること）を行い、特徴の定量化を行っていた。JP-A-59-114667, JP-A-60-250255
As described in , when image processing is performed on a blood image obtained by magnifying a blood smear/stained specimen with a microscope, in the visible range excluding blue (approximately 450 nwn to 700 nyn), the nuclei of white blood cells are in the area with the highest concentration. Based on the assumption that there are I was going.
上記従来技術は、核染色性不良の好酸球の場合白血球の
核の濃度より、細胞買上に充満している顆粒の濃度の方
が高くなる点について配慮がされておらず、核染色性不
良の好酸球を分類できないという問題があった。The above conventional technology does not take into consideration the fact that in the case of eosinophils with poor nuclear staining, the concentration of granules filling the cell surface is higher than the concentration of white blood cell nuclei. There was a problem that eosinophils could not be classified.
本発明の目的は、核染色性不良の好酸球を核染色性不良
の好酸球として正しく検出分類し、人間の目視と相関の
良い分類結果を与える白血球分類装置を提供することに
ある。An object of the present invention is to provide a leukocyte classification device that correctly detects and classifies eosinophils with poor nuclear staining as eosinophils with poor nuclear staining and provides classification results that correlate well with human visual observation.
上記目的は、核染色性不良の良否にかかわらず良く染ま
る好酸球の顆粒、核染性の良好な白血球の核、染色性不
良の白血球の核それぞれの分光特性の相対関係を利用す
ることにより達成される。The above purpose was achieved by utilizing the relative relationship of the spectral characteristics of eosinophil granules that stain well regardless of whether nuclear staining is good or bad, leukocyte nuclei that have good nuclear staining, and leukocyte nuclei that have poor nuclear staining. achieved.
すなわち、第１図の分光特性図に示す如く、好酸球の顆
粒の分光特性１３は、緑波長域（５００ｍ〜５５０ｎｎ
）では、染色性良好の核１２と同程度の濃度を示すが、
赤波長域（５６０ｍｎ〜７００ｒｒｒｎ）では、染色性
不良の核１１と同程度の濃度（染色性良好の核の赤波長
域での濃度の５０％程度）まで大きく変化する。一方、
白血球の核そのものの分光特性１１．１２は、染色性の
良否にかかわらず緑波長域と赤波長域での変化は顆粒に
較べて小さい。That is, as shown in the spectral characteristic diagram of FIG.
) shows the same concentration as nucleus 12 with good staining, but
In the red wavelength range (560 mn to 700 rrrn), the concentration changes greatly to the same level as the nucleus 11 with poor stainability (approximately 50% of the density in the red wavelength range of the nucleus with good stainability). on the other hand,
The spectral characteristics 11.12 of the leukocyte nucleus itself change less in the green and red wavelength regions than in granules, regardless of the quality of staining.
それゆえ、緑波長域での画像の濃度最大値に相当する値
と、赤波長域での画像の濃度最大値に相当する値の比を
求めれば、核染色性不良の好酸球（顆粒と染色性不良の
核から成る）と白血球（染色性良好な核を有し、青波長
域を除けば核が最も高い濃度を示す）・白血球こわれ（
染色性やや不良の核のみから成る）を区別することがで
きる。Therefore, if we calculate the ratio of the value corresponding to the maximum density value of the image in the green wavelength range to the value corresponding to the maximum density value of the image in the red wavelength range, we can calculate the ratio of eosinophils with poor nuclear staining (granules and Leukocytes (consisting of nuclei with poor staining), white blood cells (having nuclei with good staining, and the nucleus shows the highest concentration except in the blue wavelength region), and white blood cells (consisting of nuclei with poor staining)
(consisting only of slightly poorly stained nuclei) can be distinguished.
ただし、白血球のこわれの場合、緑波長域では白血球の
こわれの周辺に存在する赤血球の濃度（分光特性は第１
図１４）の方が濃くなるため、特開昭５９−１１４６６
７等の方法により、血液像中から赤血球の領域を除いた
画像に対して処理を行う必要がある。However, in the case of broken white blood cells, in the green wavelength range, the concentration of red blood cells around the broken white blood cells (the spectral characteristics are
Figure 14) is darker, so JP-A-59-11466
It is necessary to process an image obtained by removing red blood cell areas from a blood image using a method such as No. 7.
血液像中から特開昭５９−１１４６６７の方法により赤
血球の領域を求め、赤血球以外の領域の緑波長域・赤波
長域での濃度ヒストグラムを示したものが第２図である
。ここで第２図Ａａ、Ａｂは核染色性不良の好酸球、Ｂ
ａ、Ｂｂは染色性良好の白血球、Ｃａ、Ｃｂは白血球こ
われの例を示している。FIG. 2 shows a density histogram in the green wavelength region and red wavelength region of the region other than red blood cells obtained by determining the red blood cell region from the blood image by the method of Japanese Patent Application Laid-open No. 59-114667. Here, Aa and Ab in Figure 2 are eosinophils with poor nuclear staining, and B
a and Bb show leukocytes with good stainability, and Ca and Cb show examples of damaged leukocytes.
分類に必要な特徴量である上記濃度最大値に相当する値
をノイズに影響されずに安定かつ容易に求める方法とし
て、これらの濃度ヒストグラムから、所定の頻度値（オ
フセット値２１）以上の最大濃度値と最小濃度値の差２
２．２３ （濃度中）を求める方法が考えられる。上記
、赤波長域での濃度中２２と、赤波長域での濃度中２２
・緑波長域での濃度中２３の比の２つの特徴量を用いた
空間での、核染色性不良の好酸球と白血球、白血球こわ
れの分布３１〜３３を示したものが第３図である。As a method to stably and easily obtain the value corresponding to the above-mentioned maximum density value, which is a feature quantity necessary for classification, without being affected by noise, we can calculate the maximum density value above a predetermined frequency value (offset value 21) from these density histograms. Difference between value and minimum concentration value 2
2.23 (in concentration) can be considered. Above, 22 in the density in the red wavelength range and 22 in the density in the red wavelength range
・Figure 3 shows the distribution of eosinophils and white blood cells with poor nuclear staining, and the distribution of white blood cell damage 31 to 33, in a space using two feature quantities with a ratio of 23 in the concentration in the green wavelength range. be.
この空間上で核染色性不良の好酸球を核染色性不良の好
酸球として検出分類できる。In this space, eosinophils with poor nuclear staining can be detected and classified as eosinophils with poor nuclear staining.
以下、本発明の一実施例を第４図により説明する。第４
図は、白血球分類装置の全体構成を示した図である。光
ｇ４１からの白色光を顕微鏡４２に導き、顕微鏡上にセ
ットされた血液塗抹標本４３の透過光を光分光器４４で
赤・緑・青の３波長域に分光する。上記分光された光を
３つの光電変換素子４５１〜４５３にて電気信号に変換
し、Ａ／Ｄコンバーター４６１〜４６３でディジタル信
号に変換後、画像メモリ４７１〜４７３に記憶する。こ
の結果、画像メモリ４７１には赤波長域での血液像、画
像メモリ４７２には緑波長域での血液像、画像メモリ４
７３には青波長域での血液像が記憶される。An embodiment of the present invention will be described below with reference to FIG. Fourth
The figure is a diagram showing the overall configuration of a leukocyte classification device. The white light from the light g41 is guided to a microscope 42, and the transmitted light of a blood smear 43 set on the microscope is separated into three wavelength regions of red, green, and blue by a light spectrometer 44. The separated light is converted into electrical signals by three photoelectric conversion elements 451-453, converted into digital signals by A/D converters 461-463, and then stored in image memories 471-473. As a result, the image memory 471 has a blood image in the red wavelength range, the image memory 472 has a blood image in the green wavelength range, and the image memory 472 has a blood image in the green wavelength range.
73 stores a blood image in the blue wavelength range.
特徴抽出装置４８において、例えば特開昭５９−１１４
６６７の方法により、上記３つの画像メモリの内容をも
とに血液像中の赤血球領域を抽出する閾値を求め、赤血
球領域が＃Ｑ＃、赤血球以外の領域が１１“となる二値
マスクを生成する。ついで、特徴抽出装置４８において
、上記二値マスクと画像メモリ４７１に記憶された赤波
長域での血液像および画像メモリ４７２に記憶された緑
波長域での血液像から、第２図に示した濃度ヒストグラ
ムを生成する。さらに特徴抽出装置４８にて濃度ヒスト
グラムの解析を行い、濃度中２２，２３を求め、その結
果を分類装置４９に出力する。分類装置４９では、ａ度
山２２および濃度中２２と濃度中２３の比を求め、第３
図の空間のどこに位置するかで、核染色性不良の好酸球
か否かを分類する。In the feature extraction device 48, for example,
Using the method of 667, a threshold value for extracting the red blood cell area in the blood image is determined based on the contents of the three image memories mentioned above, and a binary mask is created in which the red blood cell area is #Q# and the area other than red blood cells is 11". Then, in the feature extracting device 48, from the binary mask, the blood image in the red wavelength range stored in the image memory 471, and the blood image in the green wavelength range stored in the image memory 472, The density histogram shown in FIG. Find the ratio of 22 in concentration and 23 in concentration, and calculate the third
Depending on where it is located in the space of the diagram, it is classified as whether it is an eosinophil with poor nuclear staining or not.
ここで、濃度ｉ］２２に関する識別閾値は、例えば吸収
率０％から１００％を２００レベルに量子化した場合で
７０程度、濃度中２２と濃度中２３の比は、０．７程度
である。Here, the discrimination threshold for density i]22 is about 70 when the absorption rate from 0% to 100% is quantized to 200 levels, and the ratio between medium density 22 and medium density 23 is about 0.7.
以上が１実施例であるが、ここで特徴抽出装置４８は計
算機あるいは画像処理専用装置が考えられ、分類装置４
９としては計算機が考えられる。Although the above is one embodiment, the feature extraction device 48 may be a computer or a device dedicated to image processing, and the classification device 48 may be a computer or a device dedicated to image processing.
A computer can be considered as 9.
また、他の実施例としては、
１．３個の光電変換素子４５１〜４５３をＲＧＢ出力端
子を有した単管カラー・カメラに置換えたもの。In another embodiment, 1.3 photoelectric conversion elements 451 to 453 are replaced with a single-tube color camera having RGB output terminals.
２．３個のＡ／Ｄコンバータを１個とし１時系列的に赤
・緑・青の信号を変換するもの。2. A device that converts red, green, and blue signals in one time series using three A/D converters.
３、′ａ度ヒストグラムの濃度中２２．２３の替わりに
、濃度ヒストグラムの高濃度部、低濃度部のピークを検
出し、各ピークに対応する濃度値の差を用いるもの。3. In place of 22.23 in the density of the 'a degree histogram, the peaks of the high density part and the low density part of the density histogram are detected, and the difference between the density values corresponding to each peak is used.
４、赤血球領域を除いた感度ヒストグラム（第２図）か
ら得られる特徴量の替わりに、核相当領域での赤波長域
・緑波長域での吸収率平均値を用いるもの。ただし、核
相当領域（核染色性不良の好酸球の場合は顆粒が核とし
て抽出され、他の場合は核が正しく核として抽出された
もの）は、例えば特開昭５９−１１４６６７の方法にお
いて、ＴｒｂＮをＴｒｂ’まで下げることにより抽出で
きる。4. Instead of the feature values obtained from the sensitivity histogram (Figure 2) excluding the red blood cell region, the average value of absorption in the red and green wavelength regions in the region corresponding to the nucleus is used. However, the region corresponding to the nucleus (in the case of eosinophils with poor nuclear staining, the granules are extracted as the nucleus, and in other cases, the nucleus is correctly extracted as the nucleus) can be obtained by, for example, the method of JP-A-59-114667. , can be extracted by lowering TrbN to Trb'.
等があげられる。etc. can be mentioned.
本発明によれば、従来の方法では検出できなかった核染
色性不良の好酸球を精度良く検出分類でき１人間の目視
結果との相関が良くなるという効果がある。According to the present invention, eosinophils with poor nuclear staining, which could not be detected by conventional methods, can be detected and classified with high accuracy, and the correlation with visual inspection results obtained by one person can be improved.
第１図は、血球各部の可視域での分光特性図。
第２回は赤血球領域を除いた場合の血液像の濃度ヒスト
グラム、第３図は核染色性不良の好酸球を検出分類する
のに好適な濃度中空間の図、第４図は本発明の一実施例
の製造の全体構成を示すブロック図である。
４１・・・光源、４２・・・顕微鏡、４３・・・血液塗
抹標本、４４・・・光分光器、４５１〜４５３・・・光
電変換素子、４６１〜４６３・・・ＡＤコンバータ、４
７１〜４７３・・・画像メモリ、４８・・・特徴抽出装
置、４９・・・分類ン
第 １ 図
４θρ ９ρρ Δρρ
Ｚρθ坂五飄処）
ガＶ肴に買欝−１
第２図
２１ オフ＋！、・！／） ４Ｊ
弄５図FIG. 1 is a diagram showing the spectral characteristics of each part of blood cells in the visible range. The second part is a density histogram of a blood image when the red blood cell area is excluded, Figure 3 is a diagram of the density space suitable for detecting and classifying eosinophils with poor nuclear staining, and Figure 4 is a diagram of the density histogram of the present invention. FIG. 1 is a block diagram showing the overall configuration of manufacturing in one embodiment. 41... Light source, 42... Microscope, 43... Blood smear, 44... Optical spectrometer, 451-453... Photoelectric conversion element, 461-463... AD converter, 4
71-473...Image memory, 48...Feature extraction device, 49...Classification 1st Figure 4θρ 9ρρ Δρρ
Zρθ Saka Gokudokoro) Ga V appetizers - 1 Figure 2 21 Off +! ,・! /) 4J play 5 figure
Claims (1)
、上記入力手段による赤画像・緑画像・青画像から赤血
球領域を抽出する手段を有し、赤血球領域を除いた赤画
像・緑画像の濃度情報をもとに、核染色性不良の好酸球
と核染色性良好の白血球、白血球のこわれを分類する手
段を有することを特徴とする白血球分類装置。 ２、血液像を赤・緑・青に分光して入力する手段を有し
、上記入力手段による赤画像・緑画像・青画像から白血
球の核相当域を抽出する手段を有し、白血球の核相当域
の赤画像・緑画像の濃度情報をもとに、核染色性不良の
好酸球と核染色性良好の白血球、白血球のこわれを分類
する手段を有することを特徴とする白血球分類装置。[Scope of Claims] 1. It has means for inputting a blood image by dividing it into red, green, and blue, and it has means for extracting a red blood cell region from the red image, green image, and blue image obtained by the input means, A white blood cell classification system characterized by having a means for classifying eosinophils with poor nuclear staining, white blood cells with good nuclear staining, and damaged white blood cells based on density information of red and green images excluding red blood cell areas. Device. 2. It has a means for inputting a blood image by dividing it into red, green, and blue, and it has a means for extracting a region corresponding to the nucleus of a white blood cell from the red image, green image, and blue image obtained by the input means, A white blood cell classification device characterized by having a means for classifying eosinophils with poor nuclear staining, white blood cells with good nuclear staining, and damaged white blood cells based on density information of red images and green images in corresponding regions.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP30561087A JPH01147365A (en) | 1987-12-04 | 1987-12-04 | Leukocyte classifying device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP30561087A JPH01147365A (en) | 1987-12-04 | 1987-12-04 | Leukocyte classifying device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| JPH01147365A true JPH01147365A (en) | 1989-06-09 |
Family
ID=17947214
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| JP30561087A Pending JPH01147365A (en) | 1987-12-04 | 1987-12-04 | Leukocyte classifying device |
Country Status (1)
| Country | Link |
|---|---|
| JP (1) | JPH01147365A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108693115A (en) * | 2017-12-13 | 2018-10-23 | 青岛汉朗智能医疗科技有限公司 | NK cell detection methods and system |
-
1987
- 1987-12-04 JP JP30561087A patent/JPH01147365A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108693115A (en) * | 2017-12-13 | 2018-10-23 | 青岛汉朗智能医疗科技有限公司 | NK cell detection methods and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US3999047A (en) | Method and apparatus utilizing color algebra for analyzing scene regions | |
| Li et al. | Segmentation of white blood cell from acute lymphoblastic leukemia images using dual‐threshold method | |
| US3851156A (en) | Analysis method and apparatus utilizing color algebra and image processing techniques | |
| US5123055A (en) | Method and an apparatus for differentiating a sample of biological cells | |
| Funck et al. | Image segmentation algorithms applied to wood defect detection | |
| Angulo et al. | Automated detection of working area of peripheral blood smears using mathematical morphology | |
| Ren et al. | Fusion of intensity and inter-component chromatic difference for effective and robust colour edge detection | |
| US9239281B2 (en) | Method and device for dividing area of image of particle in urine | |
| US4045655A (en) | Automatic cyto-screening device | |
| CN104990892B (en) | The spectrum picture Undamaged determination method for establishing model and seeds idenmtification method of seed | |
| CN115619793B (en) | Power adapter appearance quality detection method based on computer vision | |
| Şengür et al. | White blood cell classification based on shape and deep features | |
| CN115082451A (en) | Stainless steel soup ladle defect detection method based on image processing | |
| CN114638832A (en) | DPCR liquid drop fluorescence detection method based on watershed algorithm | |
| JP3653804B2 (en) | Particle image region segmentation method and apparatus | |
| JPH0584544B2 (en) | ||
| JPH01147365A (en) | Leukocyte classifying device | |
| Gavrilovic et al. | Quantification of colocalization and cross‐talk based on spectral angles | |
| JPS5830049B2 (en) | Automatic reticulocyte measurement device | |
| CN117934471B (en) | Image digital analysis method for colloidal gold detection result diagram | |
| JPH0443311B2 (en) | ||
| Clocksin et al. | Automatic Analysis of Fluorescence In-Situ Hybridisation Images. | |
| Gauch | Segmentation and edge detection | |
| Afrakhteh et al. | Pedestrian detection with minimal false positives per color-thermal image | |
| JPH04238207A (en) | Defect inspecting device |