Identification of particles in fluid
First Claim
1. A method for identifying unknown particles that are present in a fluid, which includes interrogating particles by directing a beam of light along a beam direction through the fluid and detecting light scattered in a plurality of different directions that are angularly spaced from said beam direction by a plurality of detectors as a result of an event, which is when a particle passes through a detection zone that lies along said beam, and recording the outputs of said detectors for an event to produce an eventvector, comprising:
- producing multiple events for particles that are of a known first specie, to produce multiple eventvectors for that known first specie, each eventvector for said known first species including a plurality of numbers that represent the outputs of a plurality of said detectors and at least the angular positions of said detectors;
producing an event for an unknown particle of an unknown specie to produce a eventvector for the unknown particle, said eventvector for said unknown particle including a plurality of numbers that represent the outputs of a plurality of said detectors;
comparing the multiple eventvectors for said known first specie to said eventvector for the unknown specie, to determine whether the unknown particle is of said known first specie.
2 Assignments
0 Petitions
Accused Products
Abstract
A method for the identification of unknown microscopic particles in a fluid. A laser beam (104) is directed through the fluid while particles move through the beam. Light scattered by each particle moving through a short detect zone (114) along the beam (an “event”) is detected by some of sixteen detectors DA-DP to generate data for the event. First, a known specie of particles (e.g. a particular pathogen specie) is placed in pure fluid and a subpattern, or eventvector, of data is recorded for each of multiple events for that specie. Each eventvector represents the outputs of all detectors. The process is repeated for other known species (e.g. algae) likely to be encountered. The group of eventvectors for all selected species is analyzed by an algorithm that determines a projection direction that results in the closest grouping of eventvectors of the same specie and greatest of separation of groups of eventvectors of different species. When an unknown particle is detected and produces an unknown particle eventvector, the program views the eventvector from the previously determined projection direction, to determine whether or not the unknown particle eventvector falls into one of the groups of known eventvectors so the particle can be identified as one of the known species or not.
-
Citations
9 Claims
-
1. A method for identifying unknown particles that are present in a fluid, which includes interrogating particles by directing a beam of light along a beam direction through the fluid and detecting light scattered in a plurality of different directions that are angularly spaced from said beam direction by a plurality of detectors as a result of an event, which is when a particle passes through a detection zone that lies along said beam, and recording the outputs of said detectors for an event to produce an eventvector, comprising:
-
producing multiple events for particles that are of a known first specie, to produce multiple eventvectors for that known first specie, each eventvector for said known first species including a plurality of numbers that represent the outputs of a plurality of said detectors and at least the angular positions of said detectors;
producing an event for an unknown particle of an unknown specie to produce a eventvector for the unknown particle, said eventvector for said unknown particle including a plurality of numbers that represent the outputs of a plurality of said detectors;
comparing the multiple eventvectors for said known first specie to said eventvector for the unknown specie, to determine whether the unknown particle is of said known first specie. - View Dependent Claims (2, 3)
said step of producing multiple events for particles of said known first specie includes placing multiple particles of said first specie in a first quantity of fluid that is originally substantially devoid of particles that would produce an eventvector, and conducting said step of interrogating for particles of said known first specie; and
includingproducing multiple events for particles that are of a known second specie, to produce multiple eventvectors for that known second specie;
performing a comparison of the multiple eventvectors for said known first specie and the multiple eventvectors for said known second specie by an algorithm that groups the eventvectors of said first and second known species to produce first and second groups with largely maximum separation of said groups while producing largely minimum separation of eventvectors of the same group;
said step of comparing includes comparing the eventvector of said unknown particle to the eventvectors of said first and second groups of eventvectors to determine whether said eventvector of said unknown particle lies in one of said groups of eventvectors.
-
-
3. The method described in claim 2 wherein:
said step of performing a comparison includes comparing by the technique of Multiple Analysis Of Variances.
-
4. A method for identifying unknown particles that are present in a fluid, which includes interrogating particles by directing a beam of light through the fluid and detecting an event which occurs when a single particle transits a detection zone along said beam of light, said detecting of an event including detecting an eventvector representing at least the direction of light scattered by a particle during an event, as detected by each of a plurality of angularly spaced detectors, and recording said eventvectors that each represents the outputs of said detectors as a recording of a particle interrogation, including:
-
for each of a plurality of known species of particles, placing multiple particles of a selected one of said known specie of particle in a quantity of fluid to create a selected known quantity of fluid, and conducting said step of interrogating particles in said selected known quantity of fluid including recording the eventvector detected by said detectors, for each of a multiplicity of known-particle interrogations of said selected known particles of said selected known specie, where said multiplicity is more than 10, to create a selected known-particle pattern that includes a group of known eventvectors for each of said plurality of known species of particles;
conducting said step of interrogating, for an unknown quantity of fluid that contains at least one unknown specie of particle, including generating an eventvector of light detected by said detectors when an unknown particles is detected, to generate an unknown particle eventvector for the unknown particle where the unknown particle eventvector includes a plurality of numbers that each represents the output of a different one of said detectors that were positioned to detect light scattered in different directions, determining correlations of the plurality of numbers representing the unknown particle eventvector with each of said groups of known-particle eventvectors, and generating an output that indicates that the unknown particle is a first of said known species when the unknown particle eventvector of the unknown particle closely matches a first group of known eventvectors of a first of said known species of particles. - View Dependent Claims (5, 6, 7)
said step of determining correlations includes determining a largely optimum projection of said groups of known particle eventvectors in a multi-dimensional space to minimize the distance between eventvectors of the same group within the space and to maximize the distance between said groups, and said step of determining correlations includes determining whether or not the unknown particle eventvector lies in the space of one of said known groups.
-
-
6. The method described in claim 4 wherein:
said step of determining correlations includes analyzing said eventvectors by the technique of Multiple Analysis Of Variances.
-
7. The method described in claim 4 wherein:
-
said step of determining correlations includes grouping eventvectors for known particles of each of said plurality of known species of particles, into each of a plurality of multi-dimensional volumes;
in the event that two of said volumes would overlap if all eventvectors that fit into a group were included in that group, performing the step of eliminating from each volume that portion that would overlap the other volume, so no pair of volumes overlap.
-
-
8. Apparatus for identifying a specie of particle that is present in fluid that is to be analyzed, which includes means for generating a light beam and a plurality of detectors that each detects light scattered in a different direction from a detection zone lying along the beam when a particle enters the detection zone and thereby produces an event, the outputs of the detectors for an event constituting an eventvector for the event, comprising:
-
a memory that stores a plurality of known-particle patterns for particles of each of a plurality of known species, where each known-particle pattern comprises a group having multiple eventvectors that each includes a plurality of numbers representing the outputs of a plurality of detectors and the identity of the detectors, that were each recorded when a different particle of the same known specie of particle entered the detection zone to create an event;
pattern recognition means for indicating whether or not an unknown eventvector produced by an event of an unknown particle, wherein each eventvector includes a plurality of numbers that each represents the output of one of said detectors and that indicates the detector, is sufficiently similar to the eventvectors of one of said known-particle patterns to indicate that the unknown particle that produced the unknown subpattern is of one of said known species. - View Dependent Claims (9)
said pattern recognition means is constructed to arrange each group of eventvectors in multiple dimensions and calculate an approximately optimum projection of the groups of known eventvectors where, along said optimum projection, the eventvectors of each group lies in a smallest space and the spaces of said groups are furthest apart, with said means being constructed to view said unknown eventvectors along said optimum projection to determine whether or not said unknown eventvector fits within a space of one of said known groups of eventvectors.
-
Specification