Method of counting particles using degree of membership in clustering data points into subgroups

• US 5,555,196 A
• Filed: 06/18/1992
• Issued: 09/10/1996
• Est. Priority Date: 11/27/1991
• Status: Expired due to Term

• Current Assignee: Toa Medical Electronics Co., Ltd.
• Original Assignee: Toa Medical Electronics Co., Ltd.
• Inventors: Asano, Kaoru
• Primary Examiner(s): Trammell, James P.
• Assistant Examiner(s): Wachsman, Hal D.
• Application Number:
  US07/900,449
• Time in Patent Office: 1,545 Days
• Field of Search: 364/555, 364/413.1, 364/413.07-.09, 382/6, 382/128, 382/133, 324/71.4, 436/10, 436/17, 377/10, 377/11, 377/12, 356/39, 395/900, 395/924
• US Class Current: 382/128
• CPC Class Codes: G01N 2015/011 with lysing, e.g. of erythrocytes
• CPC Class Codes: G01N 2015/016 White blood cells
• CPC Class Codes: G01N 2015/1024 Counting particles by non-optical means
• CPC Class Codes: G01N 2015/139 Measuring the ratio of AC/DC impedances
• CPC Class Codes: G06M 11/00 Counting of objects distributed at random, e.g. on a surface
• CPC Class Codes: G06T 2207/30024 Cell structures in vitro; Tissue sections in vitro
• CPC Class Codes: G06T 7/66 of image moments or centre of gravity
• CPC Class Codes: G06V 20/695 Preprocessing, e.g. image segmentation

Abstract

A method of seeking the number of particles of each kind in a group of particles in which plural kinds of particles are intermixed, by measuring, for example, two kinds of properties of each particle by a suitable method or device, plotting each particle on a rectangular coordinate system using the measured properties as its X and Y co-ordinates to obtain a distribution diagram, drafting boundaries to partition the distribution into clusters of the respective kinds and counting the number of particles within the boundary of each cluster, which, especially, even when the clusters of the respective particles are mutually overlapping, can reduce any deviation of the count of particles from its true value as small as possible by defining an optimum boundary. In this method, a center of gravity of each cluster is determined, a degree of attribution of each particle to each cluster is calculated based upon a distance of each particle from the center of gravity and the number of particles belonging to each cluster is counted in consideration of the degree of attribution.

Claims

1. A particle counting method comprising:

passing plural kinds of particles substantially one at a time through a narrow path;

measuring values of at least two kinds of characteristics of each particle in said narrow path;

preparing a scattergram of at least two dimensions using said values as parameters thereof;

partitioning said particles in said scattergram into clusters of said plural kinds;

defining a fixed domain for each cluster of said scattergram so that all of the particles within said domain belong to said cluster without exception and assigning a membership value of one to all particles within said domain;

calculating an initial center of gravity of each cluster based upon the spatial distribution of the particles throughout said cluster;

calculating a distance from said particle distributed outside each said fixed domain in said scattergram to the initial center of gravity of said cluster and calculating the membership value of said particle to each cluster based upon said distance;

correcting said initial center of gravity of each cluster in consideration of the value of membership of each particle distributed outside said fixed domains to each cluster;

calculating an estimated number of particles belonging to each cluster based upon the value of membership of each particle in said scattergram to each cluster.

2. A method as set forth in claim 1, characterized in that said distance is calculated as a half length of the minor axis of an ellipse having its center at said initial center of gravity of each cluster and a predetermined ratio of the major and minor axes and tilt angle of the major axis which are predetermined for each cluster passing said particle.

3. A method as set forth in claim 1, characterized in that said number of particles is calculated as assuming that each particle belongs only to the cluster which is greatest in the calculated value of membership thereof.

4. A method as set forth in claim 1, characterized in that, said number of particles is calculated as assuming that each particle is allotted to each cluster in proportion to the calculated value of membership thereof to said cluster.

5. A method as set forth in claim 1, further comprising the step of repeating said membership value calculating step and center of gravity calculating step until a value of correction of said center of gravity becomes sufficiently small.

6. A method as set forth in claim 1, wherein said sample is blood and said particles are different types of blood cells.

7. A method as set forth in claim 1, further comprising the step of applying direct current and high frequency current to said narrow path.