KR20020013970A - System for microvolume laser scanning cytometry - Google Patents

Abstract

The present invention is to provide an improved integrated system for biological marker identification. The system uses microvolume laser 11 scanning microscopy (MLSC) 16 to measure the expression pattern of biological markers in biological fluid 10. The system includes an improved tool for performing MLSC and also includes an improved method for particle detection and analysis. The system also includes an informatics structure for analyzing data obtained from the MLSC in cooperation with other medical information.

Description

SYSTEM FOR MICROVOLUME LASER SCANNING CYTOMETRY

As a result of recent revolutions in drug development, including genomics, combinatorial chemistry and high throughput screening, many drug candidates useful for clinical trials exceed the development and economic capacity of the pharmaceutical industry. In 1998, the world's largest pharmaceutical and biotechnology companies spent more than $ 50 billion on research and development, with more than one third spent directly on clinical development. As a result of a number of factors, including increased competition and pressure from protected protection agencies and other responsible people, the pharmaceutical industry improves the quality and efficacy of clinical development, including the safety and efficacy of new drugs on the market. To improve.

Thus, the recent drug development revolution is contributing to clinical trial bottlenecks. The number of treatments identified and the number of lead compounds generated far exceeds the pharmaceutical company's capacity to conduct clinical trials they are currently performing. In addition, the average cost of new drug development in the industry is currently estimated at approximately $ 500 million, which is a tremendous price to develop all potential drug candidates.

The pharmaceutical industry seeks comparable technological developments in drug development. Clinical trials are very expensive, very dangerous, and decision making is largely based on subject analysis. As a result, it is difficult to determine the patient population for which a drug is most effective, the appropriate dosage for a given drug, and the potential for side effects associated with its use. This not only leads to failures in clinical development, but can also result in the approval of products that are inappropriately administered, prescribed, or cause dangerous side effects. With the increased number of these developing drugs, pharmaceutical companies require technology to confirm objective measures of the safety and efficacy profile of candidate drugs prior to the drug development process.

Biological markers are properties that, when measured or evaluated, have discrete relationships or correlations as indicators of pharmacological responses to normal biological processes, pathogenesis or therapeutic interventions. Pharmacological responses to treatment interventions include, but are not limited to, responses to general interventions (eg efficacy), dosage response to interventions, adverse reaction profiles of interventions, and drug metabolic rates and drug metabolism. Pharmacokinetic properties such as product identification. The response may be associated with an effective or adverse (eg toxic) change. Biological markers vary in relation to the pathogenesis and include patterns of cells or molecules that have diagnostic and / or prognostic values. Biological markers may include levels of cell populations and their related molecules, levels of soluble factors, levels of other molecules, gene expression levels, genetic mutations, and clinical variables that may be associated with the presence and / or progression of disease. . In contrast to such clinical endpoints as assessment of disease progression or recurrence or quality of life (which typically takes a long time to assess), biological markers provide a faster and more quantitative measure of the drug's clinical profile. Can provide. Single biological markers currently used in both clinical practice and drug development include cholesterol, prostate specific antigen (“PSA”), CD4 T cells, and viral RNA. In contrast to the known correlation between high cholesterol and reduced CD4 positive T cells and viral RNA in heart disease, PSA and prostate cancer and AIDS, biological markers associated with most other diseases have not yet been identified. As a result, both government agencies and pharmaceutical companies intend to develop biological markers for use in clinical trials, but the use of biological markers in drug development has been limited so far.

There is a need for a biological marker identification system that can classify the vast amount of information needed to correlate biological markers with disease, disease progression, and response to treatment. Such biological identification systems are filed under US Provisional Patent Application No. 60 / 131,105, filed April 26, 1999 under the name of "Biologic Marker Identification System," and at the same time as the "phenotype and biological marker identification system." Co-owned US patent applications, all of which are specifically incorporated herein by reference in their entirety. This technique is an instrumentation and assay required to measure hundreds to thousands of biological markers, an informatics system with easy access to this data, software to correlate marker patterns and clinical data, and obtained Include the ability to use information in the drug development process. The system extensively uses microvolume laser scanning cytometry (MLSC).

In a preferred embodiment of the marker identification system, the biological fluid is contacted with one or more fluorescently labeled detection molecules capable of binding specific molecules in the fluid. Typically, the biological fluid is a blood sample and the detection molecule is a specific antibody labeled with a fluorescent dye for cell related molecules present in or in one or more subtypes of blood cells. The labeled sample is then placed into a capillary tube, which is then attached to the MLSC tool. The tool scans the laser beam through a microscope objective onto a blood sample. Fluorescent light emitted from the sample is collected by a microscope objective lens and passed through a series of photomultiplier tubes in which an image of the sample is formed in each fluorescent channel. The system then processes the raw image to identify cells from each channel and then determines the absolute cell number and relative antigen density level for each type of cell labeled with fluorescent antibody.

The marker MLSC can also be used to quantify soluble factors in biological fluids using a microsphere bound primary antibody for the factor and an antibody labeled with a secondary fluorescent color for the factor. This connects the factor to the microspheres and the binding of the secondary antibody labels the bound factor. In this embodiment the system measures the fluorescence signal associated with each bead in the blood sample to determine the concentration of each solubility factor. Using multiple bead types (each conjugated to different primary antibodies) makes it possible to perform multiple assays with the same sample volume. To identify each bead type, different beads may have distinct sizes or may have different internal colors or each secondary antibody may be labeled with a different fluorophore.

Although preferred embodiments of the invention use antibodies to detect biological markers, any other detection molecule capable of specifically binding to a particular biological marker is envisaged. For example, various types of receptor molecules can be detected through their interaction with fluorescently labeled homologous ligands.

Raw data obtained from the MLSC tool is processed by image analysis software to generate data for cell populations and soluble factors that are the subject of the assay. This data is then passed to the database. Other data that may be stored with this cell population and soluble factor data for the purpose of establishing a correlation between the biological marker and the disease or medical condition include: Drug administration and pharmacokinetics Measurement of the concentration of metabolites); Clinical variables, including the age, sex, weight, height, body type, medical history (including co-morbidities, medications, etc.), and the manifestation and categorization of the disease or medical condition, if any And other standard clinical observations made by a physician, and the like. Also included among clinical variables may be environmental and family history factors, as well as other techniques for measuring the concentration of specific molecules present in the body fluids of an individual, including standard ELISA tests, calorie functional words for enzyme activity. Assays, mass spectrometry, and the like. The data also includes information from images such as x-ray photographs, brain scans or MRI, biopsies, EKGs, stress tests or any other measurement of the subject's symptoms.

The informatics system then a) compares the stored profiles obtained from the same individual (for disease progression or treatment evaluation purposes) and / or from another individual (for disease diagnosis); And b) “mines” the data to derive a new profile. In this way, diagnostic and prognostic information can be obtained from a database and derived by the database. US patent application Ser. No. 60 / 131,105, filed April 26, 1999, entitled "Biological Marker Identification System," and co-owned US patent application filed under the name "Phenotype and Biological Marker Identification System" at the same time as this application. Is described in detail for the use of MLSCs in many other applications, all of which are specifically incorporated herein by reference in their entirety. This system can provide robust and consistent assay data even in assays that are hampered by variables between donor samples with conventional systems. Applications include the use of MLSCs to measure cell-type population changes and soluble factor changes during disease progression and treatment. For example, MLSCs can be used to identify novel biological markers for multiple infarction and rheumatoid arthritis.

The present invention relates to the analysis of biological markers using microvolume laser scanning cytometry (MLSC). The present invention includes a tool for performing MLSC, a system for analyzing image data obtained from the tool, and an informatics system for coordinated analysis of biological marker data and medical information.

1 shows the optical structure of an MLSC tool in one preferred embodiment of the present invention.

2A is a partial circuit diagram of a switchable filter scheme.

2B is a partial circuit diagram of a switchable filter scheme.

3 is a flowchart of a surroImage process.

4 schematically illustrates one file storage implementation envisioned by the present invention. The N channels of data are stored in an interleaved format as a binary file designated with an extension, * .sm1. Headers are selected to allow for various data formats.

5 is a flow chart of a baseline analysis process.

6 is a flow chart of a cell detection process.

7 shows a noise analysis process.

8 is a flowchart of a mask generation process.

9 is a flow diagram illustrating an 8-point Connectivity Rule for discovering cells.

10 shows possible types of cell assays predicted by the present invention.

FIG. 11 is a plot comparing the Gaussian fit algorithm for diameter-moment calculations. Image: Each point is the mean diameter value of these particles detected from 1000 particle (cell) artificial images with RMS noise equal to 250 counts.

12 is a flowchart of an informatics structure of a mutual scan system.

The present invention provides an improved system for performing microvolume laser scanning cytometry (MLSC). This system is called the SurroScan system. This includes an improved MLSC tool that can work at variable scan rates and can simultaneously collect data from four different fluorescent channels. The present invention includes improved methods for performing image processing on raw data obtained from MLSC tools, and improved methods for working this data in a linked database. The improvements described herein make it very easy to build and use a fast multi-factorial disease database. This database allows the user to a) compare the blood profile obtained from the laser scanning cytometer to a stored profile of an individual suffering from a known disease to obtain a prognostic or diagnostic result; And b) allow a user to quickly build a new prognosis and diagnostic profile for a particular disease; c) reveal new links between biological marker patterns and diseases in any organism.

Justice

As used herein, the term “biological marker” or “marker” or “biomarker” means a property that is measured and evaluated as an indicator of pharmacological response to normal biological processes, pathogenesis or therapeutic intervention. Pharmacological responses to therapeutic interventions include, but are not limited to, responses to common interventions (eg, efficacy), drug response to interventions, adverse event profiles of interventions, and pharmacokinetic characteristics. The response can be associated with an effective or bad (eg toxic) change. Biological markers change in connection with the pathogenesis and include the pattern or the global effect of a cell or molecule having diagnostic and / or prognostic utility.

Biological markers include, but are not limited to, the number of cell populations and levels of molecules associated with them, levels of soluble factors, levels of other molecules, levels of gene expression, presence and progression of genetic mutations and diseases, normal biological processes, And clinical variables that may be associated with response to treatment. Single biological markers currently used in both clinical practice and drug development include cholesterol, PSA, CD4 T cells, and viral RNA. Unlike the known association between high cholesterol and heart disease, PSA and prostate cancer, and CD4 positive T cells and viral RNA and AIDS, no biological markers associated with most other diseases have yet been identified. As a result, although government agencies and pharmaceutical companies focus their efforts on the development of biological markers for use in clinical trials, the use of biological markers in drug development is still limited.

As a non-limiting example, biological markers are primarily thought to have individual relationships with normal biological conditions, diseases or medical conditions. For example, increased risk of heart disease is associated with high cholesterol, increased risk of prostate cancer is associated with increased PSA levels, and the presence / progression of AIDS is associated with decreased CD4 T cells and increased viral RNA. have. However, useful markers for various diseases or medical conditions may consist of significantly complex patterns. For example, reduced levels of one or more specific cell surface antigens on specific cell types are particularly prognostic of autoimmune diseases if they are found to be associated with increased levels of one or more soluble proteins, such as cytokines. Could turn out. Thus, for the purposes of the present invention, biological markers may be referred to as patterns of multiple indicators.

As used herein, the term “biological marker identification system” refers to a system in which information is absorbed in such a way that it can obtain information from a patient population, correlate data, and identify biological markers. The biological marker identification system includes an integrated database comprising a plurality of data categories, data from a plurality of individuals corresponding to each of the data categories, and processing means for associating data within the data categories, wherein the associative analysis of data categories Can be made to identify data categories or categories in which an individual with the disease or medical condition can be distinguished from an individual without the disease or medical condition, wherein the identified category or categories are the disease or medical condition. Is a marker for. In addition, markers can be identified at different time points, for example, comparing data in various data categories for a single individual before and after administration of the drug. The MLSC system of the present application, named each other scan, is an example of a biological identification system.

As used herein, the term “data category” means any type of measurement that can be recognized for an individual. Examples of data categories useful in the present invention include, but are not limited to, the number and type of cell populations or molecules associated therewith in the individual's biological fluid, the number and type of solubility factors in the biological fluid of the individual, Associated information, cell volumetric counts per milliliter of the subject's biological fluid, the number and type of small molecules in the subject's biological fluid, and genomic information related to the subject's DNA. For example, a single data category represents the concentration of IL-1 in the blood of an individual. In addition, the data category can be the level of the drug or its metabolite in blood and urine. In addition, an example of a data category may be an absolute CD4 T cell number.

As used herein, the term “biological fluid” is not limited to, but is not limited to, blood (whole blood, leukocytes produced by red blood cell lysis, peripheral blood mononuclear cells, plasma and serum), sputum, saliva, urine, semen , Any biological material, including cerebrospinal fluid, bronchial air, sweat, feces, synovial fluid, lymphatics, tears, and soft tissue from any organism. Biological fluids typically include cells and their related molecules, solubility factors, small molecules, and other materials. Blood is a preferred biological fluid in the present invention for many reasons. First, it is readily available and can be extracted many times. Blood replenishes to some extent from the precursors of the bone marrow over time. Blood is reactive to challenge and has a memory of challenge. Blood is centrally located and circulated and potentially records changes throughout the body. Blood includes a large number of cell populations, including surface molecules, internal molecules, and secreted molecules associated with individual cells. The blood also contains both foreign-derived soluble factors such as self-derived and chemicals such as cytokines, antibodies, acute phase proteins and the like and products of infectious diseases.

As used herein, the term "cell population" refers to a group of cells having common characteristics. Such properties may include the presence, level, size, etc. of one, two, three or more cell-related molecules. One, two or more cell-related molecules can define a cell population. In general, some additional cell related molecules can be used to further subset the cell population. Cell populations are identified at the cellular level and not at the protein level. Cell populations may be defined by one, two or more molecules. Any cell population is a potential marker.

As used herein, the term "cell related molecule" means any molecule associated with a cell. This is not limited to: 1) endogenous cell surface molecules such as proteins, glycoproteins, lipids and glycolipids; 2) foreign cell surface molecules such as cytokines bound to their receptors, immunoglobulins bound to Fc receptors, foreign antigens bound to B cell or T cell receptors, and self-antibodies bound to self-derived antigens; 3) endogenous internal molecules such as cytoplasmic proteins, carbohydrates, lipids and mRNA, and nuclear proteins and DNA (including genomic and somatic nucleic acids); And 4) exogenous internal molecules such as viral proteins and nucleic acids. Preferred cell related molecules are typically cell surface proteins. For example, there are hundreds of leukocyte cell surface proteins or antigens, which are leukocyte differentiation antigens (currently including CD antigens via CD 166), antigen receptors (eg, B cell receptors, and T cell receptors), and major histocompatibility. Complexes, each of which contains a vast number of proteins.

As used herein, "soluble factor" refers to any soluble molecule found in biological fluids, typically blood. Solubility factors include, but are not limited to, soluble proteins, carbohydrates, lipids, glycoproteins, steroids, other small molecules, and any complexes of such components, such as cytokines and soluble receptors; Antibodies and antigens; And drugs complexed with any. Soluble factors can be self-derived such as cytokines, antibodies, acute phase proteins and the like and foreign ones such as products of chemicals and infectious diseases. Soluble factors may be endogenous produced by an individual or may be exogenous such as viruses, drugs or environmental toxins. Solubility factors can be small molecule compounds such as prostagladins, vitamins, metabolites (eg, iron, sugar, amino acids, etc.), drugs, and metabolites of drugs.

As used herein, the term “small molecule” or “organic molecule” or “small organic molecule” means a soluble factor or cell related factor having a molecular weight of 2 to 2,000. Small molecules may include, but are not limited to, prostagladins, vitamins, metabolites (eg, iron, sugars, amino acids, etc.), drugs, and drug metabolites.

As used herein, the term "disease or medical condition" refers to the disruption, interruption, impairment, or change in body function, system or organ in any organism. Examples of diseases or medical conditions include, but are not limited to, immune and inflammatory symptoms, cancer, cardiovascular diseases, infectious diseases, psychiatric symptoms, obesity and other diseases. By way of explanation, immune and inflammatory symptoms include autoimmune diseases, which also further include rheumatoid arthritis (RA), multiple sclerosis (MS), diabetes and the like.

As used herein, the term “clinical variable” means information that can be obtained in clinical settings in connection with a disease or medical condition. Examples of clinical variables include, but are not limited to, age, sex, weight, height, body type, medical history, ethnicity, family history, genetic factors, environmental factors, symptom manifestation or categorization of disease or medical condition, and blood pressure And any results of clinical trials such as MRI, x-rays, and the like.

As used herein, the term “clinical endpoint” means a characteristic or variable that measures how a patient feels, functions or survives.

As used herein, the term "microvolume laser scanning cytometry" or "MLSC" or "MLSC system" employs fluorescently labeled detection molecules and the sample is treated with optical scanning in which fluorescence emission is recorded to produce a small amount of A method of detecting the presence of a component in a sample. The MLSC system has several important features that distinguish it from other technologies: 1) only a small amount of blood (5-50 μl) is required for many assays; 2) obtaining absolute cell count (cells / μl); And 3) the assay can be performed directly in whole blood or in purified white blood cells. Implementation of this technique will facilitate the measurement of hundreds of different cell populations from a single recovery of blood. MLSC technology is described in US Pat. Nos. 5,547,849 and 5,556,764 and by Daitsu et al., Cytometry 23: 177-189 (1996), Aug. 21, 1998, "Laser-scanner confocal time-resolved fluorescence spectroscopy system." US Provisional Application No. 09 / 378,259, filed under the application of US Provisional Application No. 60 / 097,506 and "Novel Optical Structures for Microvolume Laser-Scanning Cytometry", issued August 20, 1999 Each of which is incorporated herein in its entirety. Laser scanning cytometry using microvolume capillaries provides a powerful way to monitor fluorescently labeled cells in whole blood, processed blood and other fluids, including biological fluids. The present invention also improves MLSC technology by improving the dose of MLSC tools that simultaneously measure multiple biological markers from small amounts of blood. The improved MLSC system of the present invention is named "Surroscan System".

As used herein, the term “detecting molecule” refers to a molecule of interest, particularly any molecule capable of binding to a protein. Preferred detection molecules are antibodies. The antibody may be monoclonal or polyclonal.

As used herein, the terms "dye", "fluorophore", "fluorescent dye" are used interchangeably to mean molecules that are excited by a laser and can fluoresce. Such dyes are typically directly linked to the detection molecules of the present invention, but indirect bonds are also encompassed herein. Many dyes are known in the art. In a particularly preferred embodiment, fluorophores are used which can be excited in the red region of the spectrum (> 600 nm). Two red dyes are typically used for Cy5 and Cy5.5. They have emission peaks 665 and 696 nanometers, respectively, and can be easily bound to the antibody. All of these can be excited at 633 nm with a helium-neon laser. Three sets of red dyes that may be used include Cy5, Cy5.5 and Cy7 or Cy5, Cy5.5 and Cy7-APC. See U.S. Provisional Application No. 60 / 142,477, filed July 6, 1999 entitled "Combined Fluorescent Dyes, Methods for Making and Uses in Assays".

As used herein, the term "particle" refers to any macromolecular structure that is detected by MLSC to obtain information about biological markers. In some embodiments, the particle to be detected is a cell, and in other embodiments, the particle to be detected is antibody-labeled beads.

The present invention provides an improved microvolume laser scanning cytometry (“MLSC”), referred to as a mutual scan system, or simply referred to as a mutual scan. Conventional systems are described in US Pat. Nos. 5,547,849 and 5,556,764, filed April 26, 1999, entitled "Biological Marker Identification System," US Provisional Patent Application 60 / 131,105, filed August 21, 1998. US Provisional Patent Application 60 / 097,506, Dieto et al. (Cytometry 23: 177-186 (1996), filed under the name of a laser-scanner confocal time-resolved fluorescence spectroscopy system) and August 20, 1999. It is described in US patent application Ser. No. 09 / 378,259, filed on the date of "Novel Optical Structures for Microvolume Laser-Scanning Cytometers," each of which is incorporated herein by reference in its entirety. The Imagn 2000 system commercially available from Biometric Imaging Inc. is an example of a conventional MLSC system.

The improved MLSC system of the present invention includes the following components.

(a) an MLSC tool comprising an electronic control system for obtaining raw data from an analytical sample;

(b) an image analysis system for collecting and enriching raw data from MLSC tools; And

(c) Integrated informatics structure for multi-parameter assay design, tool control, final data analysis, and data archiving.

The present invention provides significant improvements in several important aspects of the operation of MLSC: a) MLSC optics; b) MLSC control electronics; c) image display and analysis algorithms; And d) informatics structure. The present invention also provides an improved method for image display and for converting data into an industry standard Flow Cytometry Standard (.FCS file format).

MLSC tools

The interscan system provides a significant improvement in the optical structure of the MLSC means. Previous MLSC tools typically can detect fluorescence signals in two channels, which limits the number of analytes that can be detected simultaneously in a single experiment. In some applications, it is necessary to detect two or more different fluorescent signals to identify specific cells. For example, simultaneous determination of three or more antigens is required to identify primitive T cells expressing some cell populations, such as CD4, CD45RA and CD62L. The improved interscan tool of the present invention can detect at least four separate fluorescence signals, thereby allowing the use of at least four separate fluorescence reagents in a single experiment. One embodiment of an improved optical arrangement is shown in FIG. 1. Capillary array 10 includes a sample for analysis. In a preferred embodiment, collimated excitation light is provided by one or more lasers. In a particularly preferred embodiment, 633 nm excitation light is provided by the He-Ne laser 11. This wavelength avoids the problems associated with autofluorescence of biological materials. The power of the laser is increased to 3 to 17 mW. High laser power has two potential advantages, increased sensitivity and increased scanning speed. The collimated laser light is deflected by an excitation dichroic filter 12. By reflection, the light beam is projected onto the galvanometer-driven scan mirror 13. The scan mirror can be vibrated rapidly over a fixed range of angles, for example +/- 2.5 degrees, by the galvanometer. The scanning mirror reflects the projection light into two relay lenses 14 and 15 in which the scan mirror forms an image on the entrance pupil of the microscope objective lens 16. This optical placement converts a particular scanned angle in the mirror to a specific field position at the focal point of the microscope objective. The +/- angle sweep results in a 1 mm scan width at the focus of the objective lens. The relationship between the scan angle and the field position is essentially linear in this angle range of this arrangement. In addition, the microscope objective aims the incoming collimated beam at a point in the focal plane of the objective. The spot diameter that defines the optical resolution is determined by the diameter of the collimated beam and the focal length of the objective lens.

Fluorescent samples located in the passage of the cucked excitation beam emit Stokes-shifted light rays. This ray is collected and collimated by the objective lens. This collimation light still appears from the two relay lenses 14 and 15 collimated and impinges on the scan mirror which reflects and descans it. The Stokes-shifted light is then passed through a uniquely colored excitation filter (which reflects short wavelength light and allows long wavelength light to pass), followed by a first long pass which further provides to filter out any reflected excitation light. Pass through the filter (17).

An improved means of the present invention uses another series of dissimilar filters to separate Stokes-shifted light rays into four different emission bands. The first fluorescent dichroic 18 divides the two bluest fluorescent colors from the two reddest colors. The two bluest colors are then focused on the first aperture 19 through the first focus lens 20 to significantly reduce any fluorescence signal out of focus. After passing through the hole, the second fluorescent unique color 21 further separates the individual blue colors from each other. The individual blue color is then parsed into two separate photomultipliers 22 and 23. The two most reddish colors are divided from the two most blue colors by the first fluorescence unique color 18 and then the second through the second long pass filter 25, the mirror 26 and the second focus lens 27. The hole 24 is in focus. After passing through the hole 24, the reddest colors are separated from each other by the third fluorescent unique color 28. The individual red color is then parsed into the photomultiplier tubes 29 and 30. In this way, four separate fluorescence signals can be sent from the sample fixed to the capillary to the individual photomultiplier tubes at the same time. First of all, this improvement allows the monitoring of four separate analytes simultaneously. Each photomultiplier generates an electron current in response to the incoming fluorescent photon flux. These individual currents are switched in order to isolate the voltage by one or more preamplifiers in the detection electronics. The voltage is sampled at regular intervals by an analog to digital converter to determine the pixel intensity value for the scanned image. The four channels of the present invention are named channels 0, 1, 2 and 3.

In order to make the data obtained using a single excitation wavelength, for example, 633 nm from a He-Ne laser, a dye that can be excited from a single excitation wavelength and emits a distinctly minimally overlapping wavelength is desired. For three channel detection systems using He-Ne lasers, one suitable triple combination dye is Cy5 (emission peak at 670 nm), Cy5.5 (emission peak at 694 nm) and Cy7 (emission peak at 767 nm). There is). In an alternate embodiment, allophycocyanin (APC) is substituted with Cy5. Since the absorption peak for Cy7 (743 nm) is far from the wavelength of the He-Ne excitation laser (633 nm), those skilled in the art do not generally regard Cy7 as useful in He-Ne excitation systems. However, we found that Cy7 can be properly excited at 633 nm to enumerate specific cells in whole blood. This excitation long excitation tail as described by Mujumdar, RB, LA Ernst, SR Mujumdar, CJLewis, and ASWaggoner, 1993, Cyanine dye labeling reagents: sulfoindocyanine succinimidyl esters, Bioconjug Chem. 4: 105-11. ), Which is incorporated herein by reference in its entirety. Excitation and detection of Cy7 can be improved by increasing the laser power and using more sensitive detectors in the red region of the spectrum.

In another embodiment, Cy7 can be excited at APC excitation wavelength but is coupled to APC to make a series of dyes emitting at Cy7 emission wavelength. This series of dyes is described in Beavis, AJ, and KJPennline, 1996, Allo-7: a new fluorescent tandem dye for use in flow cytometry, Cytometry . 24: 390-5; and Roederer, M., ABKantor, DR Parks, and LAHerzenberg, 1996, Cy7PE and Cy7APC: bright new probes for immunofluroescence, Cytometry , 24: 191-7, as described in Donor (APC). Energy transfer), all of which are incorporated herein by reference in their entirety.

In some embodiments of the invention, one or more excitation wavelengths are used. By using more than one excitation wavelength it is possible to use a wider variety of fluorescent dyes, and each dye need not have the same excitation requirements. Multiple excitation wavelengths can be obtained in at least three ways: (1) using an Ar-Kr laser as the excitation source with excitation wavelengths 488 nm, 568 nm and 647 nm, thus being used for triple excitation of three different fluorescent groups (Eg, fluorescein, rhodamine, and Texas Red?); (2) using one or more laser sources, each supplying a different wavelength of collimated excitation light; (3) Use a laser that generates a femto-second pulse for multiphoton fluorescence excitation, for example, a Ti-S laser (˜700 nm excitation light) or an Nd: YLF laser (1047 nm excitation light).

Although the embodiments of the present invention described above use four separate channels, the optical structure described herein allows for the design of a tool having a larger number of channels.

In a preferred embodiment, the sample to be scanned is installed on a stage that can be automatically recognized in the X, Y and Z planes. Galvanometer driven mirror scans the excitation beam in the Y axis; The stage moves the sample on the X axis at a constant speed. The sample interval of the analog-to-digital converter, increased by the beam ratio, determines the pixels that are spaced apart on the Y axis of the image. The X stage scan rate divided by the line ratio determines the pixels spaced apart in the X axis of the image.

The stage not only scans individual samples on the X axis but also carries many samples on the microscope objective. In this way, many individual samples can be scanned continuously by the computer controller without any operator intervention. This will greatly increase the throughput of the tool and allow the tool to better follow high-speed automated analysis of blood samples in clinical settings.

In a preferred embodiment of the invention, the mutually scanned MLSC stages hold one or more capillary arrays, each of which has a footprint of 96-well plates. Each capillary holds a sample to be analyzed. Disposable capillary arrays, each with 32 fixed capillaries, are described in "Disposable Optical Cuvette Cartridges", "Spectrophotometric Analysis Systems Using Disposable Optical Cuvette Cartridges", dated April 23, 1999, Patent Attorneys 032517-004, 005 and 006; US Provisional Application, filed under the name "Vacuum Chucks for Thin Film Optical Cuvette Cartridges" and co-owned US Application, filed under the name "Disposable Optical Cuvette Cartridges" on April 21, 2000, all of which are incorporated by reference in their entirety. It is included herein for the purpose. Each array consists of two layers of Mylar sandwiched with a double adhesive adhesive layer, which is die-cut to clarify the capillary internal dimensions. The resulting cartridge, named Flex-32, can be manufactured at high cost with low cost. The cartridge is flexible, which allows it to be fixed on an optically flat baseplate by vacuum pressure, eliminating the requirement for flatness in the manufacturing process. The spaced capillaries are designed to maintain compatibility with multi-channel pipettes and robots.

In a preferred embodiment, the operator can load two plates of 32 capillaries at a time. No operator intervention is required while the plate is scanned and the image is processed. Optionally, 16 capillaries designed for the Imagn 2000 (VC120) are loaded into another holder.

The Z motion of the stage provides a means for placing each sample in the focal plane of the objective lens. Z motion can also be scanned to obtain a stack of focal plane images for each individual sample. The optimal focal position for each sample can be determined from this scanned Z image, and preferably by a computer control system to eliminate the need for operator intervention. In addition, the optimal focus can be determined for the two ends of the sample. As the sample is scanned in the X-axis, the stage is moved through the focus difference between the two ends at a constant speed, thus correcting for any tilt that may be present in the sample or fixture.

The scan rate of the laser beam determines the amount of time spent integrating the optical signal at each pixel; The longer the integration time, the better the signal-to-noise ratio. The scan rate is also proportional to the processing rate of the system. Previous MLSC means scan samples at a single rate. Although this is suitable in many applications, the present invention contemplates the use of variable scan rate systems. This variable scan rate system optimizes system sensitivity for each individual sample. For example, some assays may include detection of analytes present at very low concentrations in the sample. The fluorescence signal will be correspondingly lowered in proportion to the background noise from the low concentration of analyte. In this case, the system sensitivity can be increased by scanning slowly, which takes more time to integrate the optical signal at each pixel. This results in a much improved signal to noise ratio. On the other hand, some assays may also include detection of brighter fluorescence signals, which is possible due to the relatively high concentration of the particular analyte to be detected in the sample. In this case, a higher scan rate would be desirable: the time required to integrate the signal at each pixel is shortened to obtain a satisfactory signal-to-noise ratio. Higher scan rates also allow more samples to be processed. Thus, the variable scan rate system contemplated herein is a significant improvement over conventional fixed scan rate systems, which a) optimizes the signal-to-noise ratio for each analyte, thereby for each analyte Collect the highest possible query data; b) allows the system to function at the most effective throughput possible. In all these cases, the scan rate can be varied by adjusting the scan rate of the galvanometer-installed mirror and by adjusting the rate at which the stage moves in the X axis during sample imaging.

In order to optimize the system sensitivity at each scan rate, the mutual scan system also provides a novel switchable filter scheme that is incorporated into the analog processing circuitry. Low-pass filters are commonly used to reject unwanted high frequency noise that passes through the target signal and is created during the measurement process. In each other scan system, the optimal filter bandwidth for each scan rate is different and is generally proportional to the scan rate. In a preferred embodiment, at least two bandwidths are provided for each channel by a switchable filter. In a particularly preferred embodiment, four bandwidths are provided. 2A and 2B show circuit diagrams for a switchable filter scheme that provides a bandwidth of 4, 8, 12, and 16 kHz (they correspond to optimal bandwidths for scan rates of 64, 128, 192, and 256 Hz, respectively). Gram. In a preferred embodiment, this filter bandwidth conversion scheme is associated with each photomultiplier channel.

Thus, the present invention is significantly improved over conventional MLSC systems because the system is optimized in two separate ways: 1) the scan rate of the system is varied to optimize the signal to noise ratio; 2) The bandwidth of each analog filter in each signal channel is also varied to further optimize the signal-to-noise ratio. This novel combination synergistically enhances the sensitivity and efficacy of MLSC tools and systems.

In a preferred embodiment of the present invention, the optimum scan rate and filter bandwidth of each other's scan system are determined for each particular assay to be performed. These variables are stored in a clinical protocol database (see below), which can automatically select these settings if the operator later chooses to perform the same assay again. In this way it is possible to have many different assays present on the same stage; The computer can automatically select the optimal scan rate and filter settings that are scheduled for each sample. This progress will depend largely on the flexibility of each other's scan system.

All embodiments described above use laser excitation of fluorophores emitting in the visible or near infrared portions of the electron spectrum to detect particles. However, the present invention also contemplates the use of other types of electron emission and emission probes, such as infrared radiation. In addition, the present invention contemplates the use of an assembly of probes rather than a single probe. In addition, the present invention contemplates the use of light scattering modes other than fluorescence and includes, but is not limited to, Raman scattering, Mie scattering, luminescence, and phosphorescence.

Image to Image Analysis Software

Image processing is an important requirement in laser scanning cytometry. Image processing programs are needed to handle multiple binary images representing different spectral regions of fluorescence (channels) of cells or other particles; This is necessary to determine the background fluorescence level in each channel; Total noise in each channel is necessary for it to be able to enumerate cells or other particles from the noise; This is necessary to ignore external signals such as bubbles, dust particles or other "blob" or "grunge" sources; And it is possible to record each recognized cell or particle to record parameters including, for example, but not limited to, weighted flux, size, ellipticity, ratio and ratio and correlation between signals of different channels at the same location. Need to be specified. The mutual scan system includes an image processing and a particle detection system named mutual image system that fit the reference point and output the results of the analysis in a text list-mode format, and are named mutual image systems.

The following description of the mutual image system appears in a functional format and begins with a binary image input file (.sm1) for a text list-mode output file (.1sm) with a description and discussion of the various algorithms involved. 3 shows a flowchart of the operations performed by each other's image system. Also, in the description that follows, each other imaging system is described in a cell-detecting context. However, as described above, the mutual imaging system can detect any structure with predetermined physical parameters, eg, antibody-labeled beads. The mutual image system is contemplated for use in any embodiment of the MLSC described in the prior art, which is co-owned with US Provisional Patent Application No. 60 / 131,105 ("Biomarker Marker Identification System") and co-owned simultaneously with this application. Includes, but is not limited to, the embodiments described in US patent application "Phenotype and Biological Marker Identification System".

input

In a preferred embodiment of the present invention, a binary interlaced format is used to store image data. Multiple 16-bit data channels (images) may be intertwined in the format shown in FIG. The channel image array is stored alongside each row (Row 0: Col0, Col1, Col2, ..., Col nCol; Row1: ... to RownRow) where nCol is typically 250 pixels and nRow is typical 10000 pixels. The SM1 header shown in FIG. 4 has 28 bytes in the header having four bytes per descriptor. Each file descriptor is arranged in a low-high word format. The "four character descriptor" may be any four characters describing a unique image type, for example "SM01."

In an embodiment of the invention, the system uses 2 bytes or 16 bytes per pixel, so each pixel will have an arbitrary 65536 value. However, field descriptors, "byte (Byte per pixel) per pixel" is the image to the float (float) from the word (WORD) - and allows the flexibility to extend the type or any other data format. In addition, the variable field, "Bytes in Header", may allow for the addition of additional field descriptors. For example, four byte float images using this format specify BytePerPixel = 4, and then perhaps additional descriptor fields will be added to describe the format type as a float. The "interleave" field gives the option of writing channel in sequential mode. For example, in some implementations of the invention, the scanning system stores all data in a sequential manner rather than simultaneously, for example, preferentially in channel 0, then in channel 1, and so forth. Gather information 4 graphically illustrates a preferred file format.

In a preferred embodiment, the * .SM1 files are read as images from each other, and each channel is stored in a memory having a handle descriptor. Information for each data channel are stored in the image property handle part of the structure (image handle property), classification given SmImageInfo having hIm.

Execution: Optional parameters

In a preferred embodiment, the images of each other are executable command lines. The following format can be used to execute the program. If no parameter is given, the current parameter default is shown.

C: ＞ SurroImage { SM1 Input File } { Optional LSM Output File } { Optional Parameter List }

here,

SM1 input file : fully qualified path to * .sm1 file

Optional LSM Output File : Optional fully qualified path specifying the * .1sm output location. If this parameter is omitted, then the same path is used as * sm1 including the base name, * ,.

Optional parameter list : Multiple parameters can be assigned and separated by spaces. Example formats are:

SurroImage C: / SM1_Files \ Image1.sm1 C: \ LSM_Files \ Image1.lsm ThreshRatio = 1.2 Write RAWFiles.

Optional parameters include, but are not limited to:

Neu used to determine ThreshRatio cell detection threshold level

S increase factor

Correlation from iNumCorrelations Number of iNumCorrelations Channel

offer

UseBandPassForBlob 1 = image filtered to detect cells

Use (must be excluded from each other in UsePeaksForBlobs)

UsePeaksForBlobs 1 = center of the 5x5 kernel to detect cells

Use the difference between external pixels

UseFullPerimDetect 1 = all associated with the center to locate the cell

Use external peripheral pixels

Minimum cell diameter to detect Blobarealo

The fixed diameter of the cell for the MaxCellSize is

It is larger than MaxCellSize.

RowsPerNoiseBlock Used per block in peak-peak noise calculation

Number of rows

SampleRowsPerNoiseB1ock For samples in each block for noise calculations

Number of rows

MaxBlobPix Limit Median-Removed source image image

Specific as a "blob" that can be added to the mask

The number of adjacent pixels specifying a segment

MaxBubblePix negatively median limit-removed source image

As a "bubble" that can be added to an unknown mask

The number of adjacent pixels specifying a particular segment

BubbleThreshFactor Applies to median-rejected source images for bubble detection

Acceptable limit * noise factor. Optionally, noi

The argument can be replaced with a baseline value (tex

Reference).

BlobThreshFactor Applies to the median-removed source image for blob detection

Acceptable limit * noise factor. Optionally, noi

The argument can be replaced with a baseline value (tex

Reference).

MaskDilationPix Final mask image is expanded with MaskDilationPix pick

It is a cell.

WriteRAWFiles Diagnostics: All intermediate image files are placed in the C: \ A directory.

Boolean variable indicating if it should be written

In the SameCellRadius replacement channel, cells are killed in these centers (float format).

Its distance is less than or equal to SameCellRadius

In one case it is considered the same cell.

The following three parameters for NomCellMicrons are used for all cell counts.

Determine the kernel size used:

BeamMiocrons NomCellPix = Highports (NomCellMicrons,

Beam Microns) / MicronsPerPix;

MicronsPerPix iNomCellPix = (int) (NomCellPix + 1); iNomCellPix

(Kernel size -1) / 2

Enumerated variables that determine the text output format of a PrintMode LSM file:

0 = Human readable, 1 = Tab delimited, 2 = Comma delimited

Source image processing from each channel:

The central routines in each other's images are specified by SMProcessImages () . In a preferred embodiment, each other image system comprises a number of functions on each source image, e.g., from each channel, including but not limited to filtering, masking, blob and bubble positioning, and initial cell list construction. Do this. The central feature of each other's image system is that each channel is analyzed independently without the addition of individual channels. Briefly, each other image system independently performs multiple manipulations on each source image to remove noise and background features (e.g. bubbles and blobs) and enhance the features with the spatial characteristics of the particles being identified. The system also determines a limit for particle determination in each channel, and independently identifies and analyzes particles in each channel based on this limit and particle parameter. The system then finds the same pixel in the remaining channel (where particles are not detected because this is below the limit for that channel), and also measures the parameters of the particles in these channels. In this way, the image systems of each other collect data for each identified particle even in those channels where the particle is not originally identified.

In a preferred embodiment, the mutual image system is used to store 1) filtered source images (application of the line kernel), 2) intermediate removed source images, and 3) working images, and used for temporary storage. Start by opening the handle to the floating point image. In addition, a number of BYTE images are created to store a limit version of the floating point image (including the MASK image discussed later).

In each channel, the routine preferably begins with performing a baseline analysis. This subroutine call gives a statistic about the overall change of the baseline for y (Note: for future reference, x is in the long capillary direction, typically 40 mm or nRows = 10000 pixels and y is In the galvo-scan direction, typically 1 mm or nCols = 250 pixels). This statistic can be stored globally, including the boolean value, BaselineErrorFlag, indicating that the baseline changes past a predetermined limit (usually a maximum-minimum> 0.3 median). 5 shows this process in a flow chart format.

In a preferred embodiment, a 15 × 15 median kernel is then applied to each source image using a fast median algorithm designated by TurboMedian () . The kernel works by replacing the center pixel in the 15x15 kernel with the median of all the pixels in the kernel. The application of this median kernel to each pixel acts to "smooth" out of the gradual change in pixel intensity that occurs alongside the image in the y-axis. The primary role of the smoothing operation is to remove the intensity contribution attributable to the cell, and effectively obtain a background representation of the image. The median image can then be removed from the source image and stored in the global handle designated hImbgnd . This image can be used later after the cell list has been calculated to determine cell parameters (including, but not limited to, overall flux, ellipticity, and cell diameter (or pit area)).

In a preferred embodiment, the multiple images are then wound into a predetermined kernel and stored in the global handle specified by imBlobSrc . Such circuit kernels are known in the art. The selected kernel structure (size of the kernel and weighting values in the kernel) depends on the particles to be detected. For example, for blood cell determination, a 7x7 kernel is typically used where this kernel is approximately the size of a blood cell. For the purposes of this description, it may be assumed that the circuit kernel is a 7x7 kernel, but it should be considered that other kernels may be useful in other implementations. The result of this circuit is a filtered image that augments these features with predetermined spatial components corresponding to the cell-types detected. The limit version of this image can be used for cell detection and further for weighted flux calculations.

In some embodiments, the “perimeter” method, rather than the convolution method described above, is used for the initial reinforcement of these features with predetermined spatial components corresponding to the cell-type being detected. The perimeter method produces a differential source image (“difference” image) and can be performed by two distinct methods. In some implementations of the perimeter method, every pixel is set to the smallest difference between this and the four pixels of the outer surface of the 7x7 kernel. In another implementation, each pixel is set to the smallest difference between its value and the outer pixel of every 7x7 kernel. The use of these "difference" images rather than the wrapped images is specified through boolean command line arguments specified with UsePeaksForBlobs . Again, the enhanced image is stored in the global handle imBlobSrc . 6 illustrates the use of a perimeter method and a line filter method in a flow chart format.

Whatever method is used for initial reinforcement, the resulting image is limited, and segmentation analysis is done to determine cell location. To establish a threshold for cell detection, noise must be identified in each source image. In a preferred embodiment, an algorithm is used to calculate peak-peak noise over a segment or block of the image. 7 shows this process in flow chart format. Each block is nCols wide (the full width of the image) and RowsPerNoiseBlock (command line argument) long. Each noise value for each block is stored in an array with (int) ( nRows / RowsPerNoiseBlock ) elements. This array is then augmented by a thresholder (command line argument) and inserted into an nRow length array used for thresholding. Limiting subroutines use a wound or "difference" image to create a limiting BYTE image, imBlobSeg .

In a preferred embodiment, a subroutine named MaskGrungeAndBubbles () is requested before performing segmentation or cell-detection in imBlobSeg if the source image is associated with channel 0. 8 shows this subroutine in a flow chart format. Preferably, channel 0 is used to find bubbles and blobs, and these areas are added to the mask image. This is because dust in the sample tends to emit consistently into this channel, which matches the shortest emission wavelength from the sample. However, in other implementations, other channels (one or more) may be used for the mask image.

Mask byte images are added via three different conditions. MaskGrungeAndBubbles () tests these conditions. This uses the image, hImbgnd , the median- rejected source image, to apply the bubble and blob limits, BubbleThreshFactor and BlobThreshFactor (increased by the peak-peak noise value), respectively. For example, with respect to the bubble, * -1 * BubbleThreshFactor p-pNoise for a part of hImbgnd a particular block of the source image (the bubble is represented by the background fluorescent member) and if the total number of neighboring pixels or less than the MaxBubblePix In this case, these corresponding pixels are then fixed in the mask image to a specific value representing "bubble". Similarly, blob detection is done using BlobThreshFactor * p-pNoise and MaxBlobPix . In another preferred embodiment, bubble and blob limiting is based on a percentage of the average baseline value rather than a factor of the peak-peak noise level. Thus, the bubble and blob threshold levels are provided by BubbleThreshFactor * BaseLine (y) and BlobThreshFactor * BaseLine (y) , respectively, where Baseline (y) is a given y value (e.g., the width of the capillary). Is the median of the baseline evaluated for the x range of pixels. The final addition to the mask is made based on the granular filtered imBlobSeg image. It also uses the same tracer as provided in the command line, but only adds to the mask if MaxBubblePix is exceeded. Finally, n = MaskDilationPix pixel expansion (binary expansion fixes an arbitrary background pixel "on" if that pixel already contacts some other pixel in the region). To ensure that they are not identified on the edge. An artifact of the convolution filter tends to wrap around the rim of the bubble and can be misidentified as a cell. The expansion tends to suppress this error.

In a preferred embodiment, cells in the imBlobSeg image are then recorded using the 8-point connectivity rule. 9 shows this process in flow chart format. Multiple adjacent pixels are added to the cell list, and basic parameters are determined for each. They use, but are not limited to, the index, the maximum x and y pixel values, the total number of pixels, the xy center value based on the uniform limit cell area, and the same pixels exceeding the limit, and in the source image It includes a weighted center that weights these locations with pixel values. This center value is the floating point value used for all future calculations. The center value is ratio in the mask (remembering that each addition to the mask represents these pixels with a different "idientifier" so that those added due to the bubble can be distinguished from those added due to the blob). If present in the zeroin region, the cells can then be detected from the cell list. The last part of the calculation done in SMProcessImages is a histogram of the mask image that determines the percentage of the image that is hidden due to each of the above factors (blobs, bubbles and filter artifacts). The overall overall image masked parameter is also calculated. This allows the volume of the capillary to be recalculated when the salient part is masked.

As above, the MLSC system also stores parameters such as scan rate, filter bandwidth values in a clinical protocol database for the operation of the MLSC tool. The ability to harmonize the operating parameters of MLSC tools with each other's imaging system allows each assay to be performed in the most efficient and sensitive way possible.

Cell analysis and lsm output

In a preferred embodiment, the majority of the cell analysis and file output in each other image system takes place in the routine, WriteLsmFile () . The purpose of this routine is to output a text-based list of all cellular events to be detected in any channel. In addition, the header portion of the * .LSM file includes image statistics (measured noise level, mean, median, standard deviation statistics at baseline level, image percentage masked due to bubbles and blobs, and image creation date), as well as overall Cell statistics (number of cells detected in each channel, and minimum and maximum size). If only one channel has a "blob" that exceeds the detection limit for a given channel, cell characteristic information is output for all channels. For example, if a "blob" is detected in channel 1 and the blob has a weighted center value of (x = 22.4, y = 2342.3), then the center of the 7x7 kernel is (22,2342) The cell statistics calculated across the 7 × 7 arrays are determined on all channel images, regardless of whether the channel has cells exceeding the limit. This coordinated analysis of each channel improves the accuracy of the MLSC system by collecting and ensuring all fluorescence data for each cell. In this way, very weak fluorescence signals, which may not provide meaningful information, are not ignored, for example, even when the detected molecules are present in very low concentrations. An example of a portion of a * .LSM file for a two channel scan is shown in Table 1. The examples in Table 1 list the cellular data for two independent cellular events. In a particular embodiment, the first cell is detected in both channels as indicated by the parameter Event Source (1 = CH0, 2 = CH1, 4 = CH2, etc., and multichannel detection is indicated by the sum of the values. ). However, the second cell was only detected in channel 0 and also the parameters were still calculated for the same location of channel 1. While this is not apparent in this embodiment, the data output in * .LSM is completely stored by the y-centric value. In a preferred embodiment, a description of how this data arises from an individual channel cell list is given below.

The routine begins by sorting the list of cells in each channel. Since the "FindCell" routine adds to the list of any cell parameters that are first placed by "walking" in the y direction, it is not necessarily sorted by the y center value. Thus, bubble classification is used to generate this list (bubble classification is the best classification algorithm when a small number of rearrangements need to occur).

The next step is to create a general cell list that merges the cells in the channel and is also sorted by the y-center. The details of this routine are as follows: The index for the next useful cell to be processed is created for each channel and named CallFirstAvailIndex [Channel] . The routine cyclizes for the channel that places the cell at the lowest y-center value, which is printed soon. This cell index and their corresponding channel number are then stored in a temporary set of variables. The list, CellPrintListIndex [ChannelsMax], is created containing the indices of the cells in the replacement channel within the SameCellRadius of the cells whose centers were previously located. To populate the nChan element of this list, the routine cyclizes through the cells in all channels. However, if a cell in the replacement channel is already "marked" as it is being analyzed, it is skipped and moved to the remaining cells in that particular channel. As this loop enters, note that the source cell index is first added to the CellPrintListIndex [source_channel] element (eg, marked as being "analyzed"). Cells less than SameCellRadius whose center is the distance from the original cell have their index added to the CellPrintListIndex array.

If a single cell event matches an associated channel, it tries to output it as a text-based .LSM file. The subroutine, PrintCell (), is required from the WriteLsmFile () routine and takes two arguments, a CellPrintListIndex array that contains the indices in the cell channel list, and the current cell result count. The routine loops through all the channels and uses the centers of these cells indexed in the CellPrintListIndex array. This routine then calculates the mean median in x and y between the channels for the particular cell to be evaluated. This result is used to call another routine called AnallyzeCell () that rounds to the entire pixel closest to X and Y, and computes the cell parameters at 7x7 pixels centered at X, Y. This routine is requested on a loop over channel number. C ++ cell structure AnalyzeCell () fill is as follows.

typedef struct

{

double x, y

Area,

TotalFlux,

WeightedFlux,

Diameter,

Ellipticity,

Brightest;

int Printed: / * TRUE if printed already * /

} CELLINFO

AnalyzeCell () takes a pointer to an imBlocSrc image and begins by repositioning the pointer to the cell's X, Y position. One of the parameters passed to AnalyzeCell (), in addition to the X, Y position and the number of request channels, is a boolean flag indicating whether or not this particular channel is a "source" channel (e.g. whether a cell is actually detected in this channel). ). If this is the source channel, then the position of the maximum value found in the 7x7 region-of-interest (ROI) in the imBlobSrc image is returned. If this mismatches the center X, Y positions of the kernel, then the global parameter, nBlobsOffsetFromPeak , for these particular channels is increased. In this way, the method used to determine the center cell location can be evaluated. In addition, this parameter can be added to the cell structure as the mean of the doublets that themselves are revealed.

Regardless of whether a cell is detected or not detected in a channel called AnalyzeCell () , the weighted flux is calculated by simply evaluating pixel values at the X, Y position of the imBlobSrc image. This pixel value represents the weighted sum of all source image pixel values in the 7x7 area weighted by the predetermined 7x7 kernel provided in Table 2 below.

In another embodiment, other parameters evaluated in AnalyzeCell () include, but are not limited to, total flux, ellipticity, and average diameter. The total flux and average diameter are evaluated by another function call, ComputeMeanRadius () . 10 illustrates this function call in flow chart format. ComputeMeanRadius () is included in this routine because it not only automatically calculates the average diameter, but also because the total flux is automatically calculated from the same intermediate-removed image, hImbgnd . A recall driving the hImbgnd , 15 × 15 pixel median filter is applied to the source image and subsequently removed from the source image. To determine the mean diameter, the median value is first calculated (Note: this is different from the center value calculated to determine the center of the cell, which is the pixel whose center exceeds the limit for that channel in a 7x7 square). Because it is calculated from the previous center calculated from The distance of each pixel from the center is then weighted with respect to the pixel value and is calculated by the following equation.

Here, the center value, Cx and Cy are as follows,

Px _m , y _m is the pixel value at position x, y and N is 49 for the 7x7 kernel.

A method-of-moment's algorithm for calculating small particle diameters has been found to provide better performance on a two-dimensional gaussian fit routine. The Gaussian fit routine as shown in FIG. 11 is problematic due to the tendency to underestimate the actual diameter for low intensity cells. This trend is found in the algorithm at the moment, but much less is published.

The total flux is simply given by the denominator of equations (1) and (2). If the total flux that can occur in the background removed image is below zero or equal to zero, then the sum is then assigned a value of 1.0 to prevent excessive overflow, and the average diameter is fixed at zero.

Two other cellular parameters evaluated in PrintCell () include the ratio and correlation values between the channels. This ratio (see the examples in Table 1) is provided as follows.

Pearson's correlation, ρ _{m, n} coefficient is calculated by the following equation.

Here, S _m and S _n are standard deviations of the source image imSrc pixel values in the channels m and n, respectively, and bars represent average pixel values. As each cell sorts across channels, each of these cell parameters is written in a sequential manner in a * .LSM file.

The WriteLSMFile () routine continues through all cells, each time requiring a PrintCell () and a subroutine AnalyzeCell () . The total cell count is counted and written in the header of the * .LSM file. Then close the file and exit the program.

The mutual image system described herein is substantially improved over conventional systems for particle detection in the laser scanning cytometry context. One such conventional system is described in US Pat. No. 5,556,764 (hereinafter referred to as the '764 patent), which is incorporated herein by reference in its entirety. The system described in the '764 patent first sums up images from individual channels and then runs particles on the resulting composite image; The '764 system also does not perform any masking of blobs and bubbles. In addition, the '764 system is designed to detect cells within a specific size range, for example, to be highly selective for a particular type of cell of interest in an assay. On the other hand, the system is less restrictive and therefore detects more different types of cells. Independent channel analysis combined with the blob and bubble blocking techniques described herein allows for accurate identification of images with each other and collects data from cells more in fact than the '764 system. Thus, the system is more accurate and sensitive than the conventional system.

Another advantage of the mutual imaging system is that it can be easily optimized for the detection of a variety of different cells with various shapes and / or different patterns or intensities of cell-related molecules. In addition, each other imaging system can be rapidly optimized for particle detection other than cells. For example, in some embodiments of the present invention, a mutual imaging system is used to detect microbeads in capillaries, where the microbeads bind to specific reagents present in the blood. Compared to each other's imaging system, conventional systems can only detect certain cells and cannot be rearranged for the detection of other structures without significant operator intervention. The parameters of the individual subroutines of the mutual image system, for example the structure of the convolution kernel, can be changed quickly to optimize the detection of these particles. These parameters may be stored in a clinical protocol database (see below). Thus, the mutual imaging system increases the flexibility of the MLSC system, which allows to perform various assays without compromising sensitivity.

Informatics structure

The present invention includes novel informatics structures that perform a number of important functions. At the heart of the system is a relational database used to design multi-parameter assays, control measurement tools, perform image and data analysis, and coordinate all the information required to record the results. The system includes a number of connected modules that perform discrete functions. 11 shows a flowchart illustrating how this system operates in the preferred embodiment. Briefly, Instrument Control Software controls each other's scan hardware (MLSC tools), which scans samples and generates raw image files (.SM1 files). The .SM1 file is then processed and augmented by SurroImage Image Analysis Software (above). This module augments each image, determines the location and size of each cell (or fluorescent beads in some applications) in each image, and then calculates the fluorescence intensity of each cell (or beads) in each channel do. The resulting cross-image data can be stored as a text file (.LSM file) and then converted by the FCS Conversion Software to the industry standard.FCS format or to another file format suitable for subsequent analysis. The tool control software, image analysis software and FCS conversion software are all controlled by the Clinical Protocol Database, which stores the parameters for each type of assay used to execute the clinical protocol. These parameters include, but are not limited to, the scan rate of the MLSC tool, the value of the filter bandwidth used in the MLSC tool, and the kernel structure used in the mutual imaging system. For example, data in the form of .FCS and .LSM files may then be sent to a server for further processing of the data, for example using commercially available Flow Jo software. The data is also sent to an experimental data file server for archiving and for periodic sending to third-party media, and also to a central database such as an Oracle database. The central database is used to maintain the consistency of the clinical protocol database without limitations; Used as a central repository for tool results, filenames, and validation information; Used to store cell assay measurements and soluble factor measurements (whether obtained through MLSC systems or through conventional ELISA assays); Used to maintain clinical questionnaire information.

In a preferred embodiment, the mutual scan informatics system is used for clinical research in the following methods (assuming the usefulness of the previous design of the appropriate relational database schema and the validated tools). First, the user defines a clinical research protocol that includes information such as the number and identity of patients, the number of samples per patient, and the like. The clinical study may include 10 to 100 patients and may last from 1 week to 1 month. The user also defines an assay protocol, which defines in detail each of the assays that can be performed on each particular patient sample. Each assay includes, but is not limited to, a detailed definition and description of each of the reagents, including the fluorophore, target molecule, dilution and fluorescence supplemental parameters used. Sample preparation methods and sample dilution are also included. The protocol also includes the information required to automatically control each other's scanning tools and data analysis software. After patient samples have been processed for each assay (which can be automated under the control of a database) and loaded onto measurement cartridges on each other, the user enters the protocol ID and sample ID parameters into the scanner software, It then searches the database to determine detailed scan parameters such as scan rate, filter bandwidth setting, stage recognition rate, and the like. After the scan is complete, the tool again scans the database to learn the appropriate analysis parameters, automatically performs the correct type of assay with each other's images and with each other's FCS software modules, generating an FCS output file. FCS output files are further analyzed using commercially available FCS analysis software. A summary of the FCS output data for each patient sample is then generated by the FCS software and processed for further storage in a relational database. Measurement results and patient clinical information are then processed to identify patterns and correlations that can represent candidate biological markers with various statistical and visualization methods. Sample and assay information is associated with data from analysis from raw images in a list mode format for a relational database.

The present invention also contemplates the use of an image system to graphically represent the enriched data. The system, named SuroView, represents individual cells identified by the image software of each other; A box can be placed around each identified cell to distinguish the true cell from the plausible signal of another cell type in the image. MutualView software is particularly useful for quickly diagnosing various types of system failure modes. It should be noted that during normal operation of each other's scanning tools, the operator does not need to see anything like the image of the cell.

Example of list mode output. The data corresponds to a two-channel scan. Cell number Channel: 0 Total flux Flux width Cell diameter Cell ellipticity Pixel brightness Correlation 0-1 Ratio Correlation 0-2 Ratio X Y One 191 23 4141 427 3.2 0.90 3798 0.91 0.56 0.48 0.08 2 196 26 2282 501 1.3 0.95 4124 0.917 0.36 0.71 0.11 3 ... Cell number Channel: 1 Total flux Flux width Cell diameter Cell ellipticity Pixel brightness Correlation 1-2 Ratio Event Source X Y One 191 23 1824 241 2.5 0.88 1962 0.41 0.14 3 2 196 26 1294 181 1.9 0.93 2027 0.77 0.29 One 3 ...

ImBlobSrc , the line kernel used to produce the filtered image. imBlobSrc is used both for cell detection and for evaluation of the weighted flux of cells. -3.54 -3.54 -3.54 -3.54 -3.54 -3.54 -3.54 -3.54 2.00 2.00 3.00 2.00 2.00 -3.54 -3.54 2.00 4.00 6.00 4.00 2.00 -3.54 -3.54 3.00 6.00 9.00 6.00 3.00 -3.54 -3.54 2.00 4.00 6.00 4.00 2.00 -3.54 -3.54 2.00 2.00 3.00 2.00 2.00 -3.54 -3.54 -3.54 -3.54 -3.54 -3.54 -3.54 -3.54

Claims (6)

(a) scanning a fixed volume capillary comprising a sample to generate a plurality of data channels, each data channel comprising a distinct detectable characteristic and a distinct background characteristic; (b) sampling each data channel to produce a corresponding set of pixel values; (c) generating a set of enhanced pixel values by independently modifying each set of pixel values to selectively enhance the displayed spatial features of the target particle; (d) removing distinct background characteristics for the corresponding channel from the set of one or more enhanced pixel values; (e) independently constructing a noise limit value for detection of the particles for each set of enhanced pixel values; (g) independently identifying, in each set of enhanced pixel values, the group of limit pixels located in the pattern required for diagnosing the particle; (h) independently identifying, for each group of said-limit pixels located in the diagnostic pattern in a particular set of enhanced pixel values, a corresponding lower or lower limit pixel in the remaining set of enhanced pixel values; (i) characterizing the target particles in the sample by analyzing the pixels independently identified in steps (g) and (h); Here, a particle is initially identified and analyzed in the channel with the limit pixel located in the pattern diagnosis of the particle, and the particle then places the pixel at the same position as in the limit pixel initially identified in all remaining channels. And a particle for detecting and characterizing a target particle having a plurality of detectable properties in a fixed volume capillary comprising a fluorescent background and showing background properties. (a) one or more sources for excitation light; (b) means for scanning the excitation light at a plurality of predetermined scan rates for a sample; (c) means for collecting and descanning light rays emitted by the sample in response to the excitation light; (d) a plurality of dichroic filter means for separating the emitted light rays into a plurality of separate component wavelengths and for inducing each of the individual component wavelengths alongside an independent optical path; (e) a plurality of photo detectors, each responsive to emitted light rays, positioned to collect separate component wavelengths, each operatively connected to an analog filter means having a variable bandwidth; (f) digital sampling means operatively connected to each analog filter means for providing a digital signal output from each photo detector and capable of operating at a plurality of predetermined digital sampling rates; Wherein the scan rate, the bandwidth of each analog filter means, and the digital sampling rate include particles that emit light into multiple wavelength components in accordance with illumination of excitation light, which can be coordinated to optimize detection of the particles. A tool for performing microvolume laser scanning cytometry on a sample. (a) one or more sources for excitation light; (b) means for scanning the excitation light against a sample; (c) means for collecting and descanning light rays emitted by the sample in response to the excitation light; (d) a plurality of dichroic filter means for separating the emitted light beam into four or more separate component wavelengths and for inducing each respective component wavelength alongside a separate optical path; And (e) four or more photo detectors, each positioned to collect separate component wavelengths, each of which is responsive to emitted light rays operatively connected to a digital sampling means; A tool for performing microvolume laser scanning cytometry (MLSC) on a sample comprising particles that emit light into separate wavelength components. (a) one or more sources for excitation light; (b) means for scanning the excitation light at a plurality of predetermined scan rates for a sample; (c) means for collecting and descanning light rays emitted by the sample in response to the excitation light; (d) a plurality of dichroic filter means for separating the emitted light rays into a plurality of discrete component wavelengths and for inducing each respective component wavelength alongside a separate optical path; (e) a plurality of photo detectors responsive to emitted light rays, each positioned to collect separate component wavelengths and operatively connected to analog filter means each having a variable bandwidth; (f) digital sampling means operatively connected to each analog filter means and operable at a plurality of predetermined digital sampling rates to provide a digital signal output from each photo detector; And (g) a control and data analysis system, comprising one or more computers, capable of the following (i) to (i), comprising: obtaining from an organism and emitting light to a multi-wavelength component under illumination of excitation light; Integrated system for microvolume laser scanning cytometry (MLSC) on samples containing particles: (Iii) storage of each clinical protocol including information on: Scan rate required for optimal detection of the particles; A bandwidth of the analog filter means required for optimal detection of the particles; Digital sampling rate required for optimal detection of the particles; Assay conditions for the sample; A history of the organism; And Algorithm for Analysis of the Digital Signal Output (Ii) control of the scanning means, the analog filter means and the digital sampling means using information from the clinical protocol; (Iii) accepting and storing the digital signal output; (Iii) analysis of the digital signal output using information obtained from the clinical protocol to provide data regarding the particle; (Iii) storing information about medical symptoms about data relating to the particles; (Iii) determining the medical symptoms of the organism using the information obtained from (5); And (Iv) the generation of new information relating to medical symptoms for the data relating to the particles. A method of analyzing a sample comprising a particle that detects and characterizes a target particle having a plurality of detectable properties in a fixed volume capillary that includes a fluorescent background and exhibits the background property. (a) scanning a fixed volume capillary comprising a sample to generate a plurality of data channels, each data channel comprising a distinct detectable characteristic and a distinct background characteristic; (b) sampling each of the data channels to produce a corresponding set of source pixel values; (c) totaling a set of source pixel values to generate a composite image; (d) calculating a limit in the composite image for particle detection; (e) performing particle detection using the limit in the composite image; (f) identifying the corresponding pixel in the set of source pixel values for each particle identified in the composite image; And (g) analyzing the pixels identified in step (f); wherein improvements are: (Iii) independently calculating a limit for particle detection in each set of source pixel values; (Ii) independently performing particle detection on each set of source pixel values using a corresponding limit; (Iii) identifying the corresponding pixel in the remaining set of source pixel values for each particle identified in the particular set of source pixel values in step (2); And (Iii) analyzing the pixels identified in steps (2) and (3). A method of analyzing a sample comprising a blotter, which detects a target particle having a plurality of detectable features in a fixed volume capillary that includes a fluorescent background and exhibits the background characteristics. (a) scanning a fixed volume capillary comprising a sample to generate a plurality of data channels, each data channel comprising a distinct detectable characteristic and a distinct background characteristic; ; (b) sampling each of the data channels to produce a corresponding set of source pixel values; (c) totaling a set of source pixel values to generate a composite image; (d) calculating a limit for particle detection in the composite image; (e) performing particle detection using the limits in the composite image, wherein the improvements include: (Iii) calculating a limit for particle detection independently in each set of source pixel values without having to first sum up the source images; (Ii) performing particle detection on each set of source pixel values using a corresponding limit.