MXPA01013398A - System for microvolume laser scanning cytometry. - Google Patents

Abstract

The present invention provides an improved integrated system for biological marker identification. The system uses Microvolume Laser (11) Scanning Microscopy (MLSC) (16) in order to measure patterns of expression of biological markers in biological fluids (10). The system includes improved instrumentation for performing MLSC, and also includes improved particle detection and analysis methods. The system further comprises an informatics architecture for the analysis of data obtained from the MLSC in tandem with other medical information.

Description

SYSTEM FOR MICOMECURITY LASER EXPLORATION CYTOMETRY Field of the Invention The present invention relates to the analysis of biological markers using Microvolume Laser Scanning Cytometry (MLSC). The invention includes the instrumentation to perform the MLSC, a system for the analysis of the image data obtained from the instrumentation and a computer system for the coordinated analysis of the biological marker data and the medical information.

Background of the Invention As a result of recent innovations in drug discovery that include combinatorial chemistry, genetics, and high-throughput screening, the number of candidate drugs available for clinical testing exceeds the development and economic capacity of the pharmaceutical industries. In 1998, the most important companies 15 pharmaceuticals and biotechnologies worldwide spent more than $ 50 billion on research and development, more than a third of which was spent directly on clinical development. As a result of a number of factors, including increased competition and pressure from health organizations and other sponsors, the pharmaceutical industry is seeking to increase quality, including 20 the safety and efficacy of new drugs brought to the market and improve the efficiency of clinical development.

Recent innovations in drug discovery, therefore, have contributed to making clinical trials a bottleneck. The numbers of therapeutic targets being identified and leading compounds being generated exceed ¥ -; 'a lot the ability of pharmaceutical companies to conduct clinical trials as they are currently being performed. In addition, as the industry currently estimates that the average cost of developing a new drug is $ 500 million, it is prohibitively expensive to develop all potential candidate drugs. 30 The pharmaceutical industry is being forced to seek equivalent technological improvements in drug development. "Clinical trials are still very expensive and very risky, and decision-making is often based on highly subjective As a result, it is often difficult to determine the patient population for which a drug is most effective, the appropriate dose for a given drug and the potential side effects associated with its use. Not only does this lead to more flaws in clinical development, it can also lead to approved products that can be dosed, improperly prescribed or cause dangerous side effects. With an increasing number of drugs in their facilities, pharmaceutical companies require technologies to identify objective measures of the safety and efficacy profile of a candidate drug before the process of drug development.

Biological markers are characterized in that when they measure or are evaluated, they have a discrete relationship or correlation as an indicator of normal biological processes, pathogenic processes or pharmacological responses to a therapeutic intervention. Pharmacological responses to therapeutic intervention include, but are not limited to, responses to general intervention (eg, efficacy), response to intervention dose, lateral effect profiles of the intervention, and pharmacokinetic properties, such as the speed of the intervention. drug metabolism and the identity of the drug metabolites. The answers can be correlated with either beneficial or adverse (for example, toxic) changes. Biomarkers include patterns of cells or molecules that change in association with a pathological process and have a diagnostic and / or prognostic value. Biological markers can include levels of cell populations and their associated molecules, levels of soluble factors, levels of other molecules, gene expression levels, genetic mutations, and genetic parameters that can be correlated with the presence and / or progression of the disease. In contrast to such points of clinical extremes as measures of disease progression or recurrence or quality of life (which ^ typically take "a long time to determine", biological markers can provide a more rapid and quantitative measure of a clinical profile of a drug.The biological markers alone currently used in both clinical practice and the development of drugs, include cholesterol , prostate specific antigen ("PSA"), CD4 T cells and viral RNA, unlike the well known correlations between high cholesterol and "heart diseases, PSA and prostate cancer, and CD4 positive T cells decreased and viral RNA in AIDS, correlated biological markers With many other diseases, they still have to be identified. As a result, although government agencies and pharmaceutical companies are increasingly seeking to develop biomarkers for the use of clinical trials, the use of biomarkers in drug development has been limited to date.

There is a need for a system for identifying a biological marker that is capable of choosing through vast amounts of information necessary to establish the correlation of biological markers with the disease, the progression of the disease and the response to therapy. Such a biological marker identification system is described in United States Provisional Patent Application Serial No. 60 / 131,105, entitled "Identification System of a Biological Marker," filed on April 26, 1999, and in the Application of Utility of the United States, commonly possessed, presented concurrently with this application, "System of Identification of a Biological Marker and Phenotype", both of which are incorporated 15 specifically here as a reference in its entirety. This technology includes the instrumentation and tests required to measure from hundreds to thousands of biological markers, a computer system to allow this data to be accessed easily, programs to correlate the patterns of markers with clinical data and the ability to use the information resulting in the process of developing 20 a drug. This system extensively uses Microvolume Laser Scanning Cytometry (MLSC).

In the preferred embodiments of the marker identification system, a biological fluid is contacted with one or more labeled detection molecules.

Fluorescently, they can bind to specific molecules in that fluid. Typically, - _ > < "Biological fluid is a" blood sample, and the detection molecule is a fluorescent dye-labeled antibody, specific for a molecule associated with a cell that is present on, or within, one or more subtypes of blood cells. The marked sample is then placed in a capillary tube, and the tube is mounted on a 30 MLSC instrument. This instrument scans with laser light through a lens of 1 - . 1 -r? microscope the 'blood sample. The fluorescent light emitted from the sample is collected by the microscope objective and passed to a series of photomultipliers where the images of the sample in each fluorescent channel are formed. The system then processes the untreated image of each channel to identify the cells, and then determines the absolute cell counts and the relative antigen density levels for each cell type labeled with the fluorescent antibody.

The MLSC marker can also be used to quantify the soluble factors in biological fluids using a primary antibody bound to a microsphere to the factor, along with a secondary antibody fluorescently labeled to the factor. The factor, therefore, binds to the microsphere, and the binding of the secondary antibody fluorescently labels the bound factor. The system in this mode measures the fluorescent signal associated with each sphere or bead in the blood sample in order to determine the concentration of each soluble factor. It is possible to perform multiple assays on the same sample volume using multiple types of spheres or beads (each conjugated to a different primary antibody). In order to identify each type of sphere or bead, the different spheres or beads may have different sizes or may have a different internal color, 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 contemplated. For example, various types of receptor molecules can be detected through their interaction with the fluorescently labeled cognate ligand.

The raw data of the MLSC instrument is processed by a program of '* > .- image analysis to produce data on cell populations and soluble factors that were the subject of the trial. These data are then transferred to a database. Other data that can be stored together with these data from the cell population and the soluble factor, to establish the correlations between biological markers and diseases or medical conditions include: dosage of the drug and pharmacokinetics (measurement of the concentrations of a drug and its metabolites in a body); clinical parameters including, but not limited to, the age of the individual, gender, weight, height, body type, medical history (including comorbidities, medication, etc.), manifestations and categorization of the disease or medical condition (if any) and other standard clinical observations made by the physician. Also included within the parameters should be the environmental and family history factors, as well as the results of other techniques to measure the concentrations of specific molecules present in the individual's body fluids, including, without limitation, standard ELISA tests. colorimetric functions for enzymatic activity, and mass spectrometry. The data may also include images such as x-ray photographs, brain scans, or MRI, or information obtained from biopsies, EKGs, stress tests, or any other measure of the individual's condition.

A computer system then a) compares the data with the stored profiles (either from the same individual for the progression of the disease, or for purposes of therapeutic evaluation) and / or from other individuals (for the diagnosis of the disease); and b) "exploits" the data in order to derive new profiles. In this way, the diagnosis and forecast information can be obtained from and derived from the database. U.S. Patent Application Serial No. 60 / 131,105, filed on April 26, 1999, Identification System of a Biological Marker, and the commonly owned US Utility Application filed concurrently with this application, "Identification System of a Biological Marker and Phenotype", both of which are specifically incorporated herein as a reference in their entirety, describe in great detail the use of the MLSC in many different applications. The system is capable of providing consistent and robust test data, even in tests where the prior art systems are impeded by the variability between the samples of **? the donors. The applications include the use of the MLSC to measure changes in cell-type populations and changes in the soluble factor during disease progression and during therapy. For example, MKSC can be used to identify novel biomarkers for multiple sclerosis and rheumatoid arthritis.

Summary of the Invention The present invention provides an improved system for performing Microvolume Laser Scan Cytometry (MLSC). The system is called the SurroScan system. It includes an improved MLSC instrument, capable of working at variable scanning speeds and capable of simultaneously collecting data on four different fluorescent channels. The invention includes an improved method for performing image processing of the raw data obtained from the MLSC instrument, and an improved method for working with this data in a related database. The improvements described here will greatly facilitate the construction and use of a rapid, multi-factorial disease database. This database will allow users to a) compare the blood profiles obtained with cytometry by laser scanning with the stored profiles of individuals suffering from known diseases, in order to obtain diagnostic and prognostic results; and b) allow the user to quickly build new prognostic and diagnostic profiles for particular diseases, c) discover new links between the patterns of biological markers and disease in any organism.

Brief Description of the Figures Figure 1 illustrates the optical architecture of the MLSC instrument in a preferred embodiment of the invention.

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

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

Figure 3 is a flow diagram of the Surrolmage process.

Figure 4 schematically illustrates a storage mode of a file contemplated by the present invention. No data channels are stored in - "an interleaved format in a binary file designated with the extension * .sml. The header is chosen to allow a variety of data formats.

Figure 5 is a flow diagram of the baseline analysis process.

Figure 6 is a flow diagram of the cell detection process.

Figure 7 illustrates the noise analysis process.

Figure 8 is a flowchart of the MASK generation process.

• Figure 9 is a flowchart illustrating the 8 point Connectivity Rule to find cells.

Figure 10 illustrates some possible types of cell analysis contemplated by the present invention. 15 Figure 11 is a graph comparing the calculations of the Gaussian adjustment algorithm to the diameter-momentum. Images: Each point in a value of the average diameter of those particles detected from an artificial image of a 1000 particle (cell) with an RMS noise equal to 250 counts. 20 Figure 12 is a flowchart of the computer architecture of the SurroScan system.

Detailed Description of the Invention 25 DEFINITIONS '• fi'. ^ t • As used herein, the term "biological marker" or "marker" "biomarker" means a characteristic or parameter that is measured and evaluated as an indicator of normal biological processes, pathogenic processes or pharmacological responses to a therapeutic intervention. Pharmacological responses to Therapeutic intervention includes, but is not limited to, response to the general intervention '- (eg, efficiency), response to dose to intervention, profiles of the side effects of the intervention, and pharmacokinetic properties. The answer can correlate with efficient or adverse changes (eg, toxic). Biomarkers include patterns or sets of cells or molecules that change in association with a pathological process and have a diagnostic and / or prognostic value.

Biomarkers include, but are not limited to, cell population counts and associated molecule levels, soluble factor levels, levels of other molecules, gene expression levels, genetic mutations, and clinical parameters that can be correlated with the presence and progression of the disease, normal biological processes and response to therapy. The simple biological markers currently used in clinical practice and in the development of drugs include cholesterol, PSA, CD4 T cells, and viral RNA. Unlike the well-known correlations between high cholesterol and heart disease, PSA and prostate cancer, and CD4 positive T cells and viral RNA and AIDS, biological markers correlated with at least other diseases have yet to be identified. As 15 result, although government agencies and pharmaceutical companies are increasingly seeking the development of biomarkers for use in clinical trials, the use of biomarkers in drug development has been limited to date. 20 As a non-limiting example, it is often thought that biological markers have discrete relationships with normal biological status, a disease or medical condition, for example, high cholesterol correlates with an increased risk of heart disease, Elevated PSA, correlated with an increased risk of prostate cancer, reduced CD4 T cells, and increased viral RNA, correlate with the presence / progression of AIDS. Nevertheless, : * 'l ** S it is very likely that useful markers, for a variety of diseases or medical conditions, may consist of significantly more complex patterns. For example, it can be discovered that the decreased levels of one or more specific cell surface antigens in a specific type of cell, when found in 30 together with elevated levels of one or more soluble factors - - cytokines, perhaps - - is t * indicative of a particular autoimmune disease. Therefore, for the purposes of In this invention, a biological marker can refer to a pattern of a number of indicators.

As used herein, the term "biological marker identification system" means a system for obtaining information from a patient population and assimilating the information in a manner that allows the correlation of the data and the identification of biological markers. A population of patients can comprise any organism. A biological marker identification system comprises 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 to correlate the data within the data categories, where the correlation analysis of the data categories can be done to identify the category or categories of data, where the individuals who have the disease or medical condition can be differentiated from those individuals who do not 15 have the disease or medical condition, where the category or categories identified are markers of said disease or medical condition. In addition, markers can be identified by comparing the data in several categories of data for a single individual at different time points, for example, before and after the administration of a drug. The MLSC system of the present application, termed 20 SurroScan system, is an example of a system for identifying a biological marker.

As used herein, the term "data category" means any type of measurement that can be discerned about an individual. Examples of categories of data useful in the present invention include, but are not limited to, numbers and types of data. "- v.:m ° fi: cell populations and their associated molecules in the biological fluid - of an individual numbers and types of soluble factors in the biological fluid of an individual, information associated with a clinical parameter of an individual , cell volumetric counts per ml of an individual's biological fluid, numbers and types of small molecules in the fluid Biological information of an individual, and genomic information associated with the DNA of an individual. For example, a single category of data could represent the concentration of Ljl in an individual's blood. In addition, a category of data could be the level of a drug or its metabolites in the blood or urine. An additional example of a data category would be the absolute CD4 T cell count.

As used herein, the term "biological fluid" means any biological substance, including but not limited to, blood (including whole blood, leukocytes prepared by lysis of red blood cells, peripheral blood mononuclear cells, plasma, and serum), sputum, urine, semen, cerebrospinal fluid, bronchial aspirate, sweat, feces, synovial fluid, lymphatic fluid, tears and macerated tissue obtained from any organism. The biological fluid typically contains cells and their associated molecules, soluble factors, small molecules and other substances. Blood is the preferred biological fluid in this invention for numerous reasons. First, it is readily available and can be extracted multiple times. The blood is replenished, in part, by the progenitors in the marrow, over time. Blood is responsible for antigenic challenges and has a memory of antigenic challenges. The blood is centrally located, recirculates and potentially reports changes through the body. Blood contains numerous cell populations, including surface molecules, internal molecules, and secreted molecules associated with individual cells. The blood also contains soluble factors that are both themselves, such as cytokines, antibodies, acute phase proteins, etc., as foreign, such as chemicals and products of infectious diseases.

As used herein, the term "cell population" means a set of cells with common characteristics. Characteristics may include the presence and level of one, two three or more molecules associated with cells, size, etc. One, two or more molecules associated with cells can define a cell population. In general, some -molecules associated with additional cells can be used for a subset of a cell population. A cell population is identified at the level of the population and not at the level of the protein. A cell population can be defined by one, two or more molecules. Any cell population is a potential marker.

• "As used herein, the term" associated molecule- with a 1 cell "means 1 any molecule associated with a cell, This includes, but is not limited to: 1) molecules of the intrinsic cell surface such as proteins, glycoproteins, lipids, and glycolipids; 2) extrinsic cell surface molecules, such as cytokines bound to their receptors, immunoglobulin bound to Fc receptors, foreign antigen bound to B cell or T cell receptors and self-antibodies bound to self antigens; 3) intrinsic internal molecules such as cytoplasmic proteins, carbohydrates, lipids and mRNA, and nuclear protein and DNA (including genomic and somatic nucleic acids); and 4) extrinsic internal molecules such as viral proteins and nucleic acid. The molecule associated with a preferred cell is typically a protein on the surface of the cell. As an example, there are hundreds of proteins or cell surface antigens of leukocytes, including leukocyte differentiation antigens (including CD antigens, currently up to CD166), antigen receptors (such as the B cell receptor and the T cells), and a major histocompatibility complex. Each of these classes covers a vast number of proteins.

As used herein, the term "soluble factor" means any soluble molecule that is found in a biological fluid, typically blood. The soluble factor includes, but is not limited to, soluble proteins, carbohydrates, lipids, lipoproteins, steroids, other small molecules, and complexes of any of the above components, eg, cytokines and soluble receptor; antibodies and antigens; and a drug complexed with anything. The soluble factors can be their own, such as cytokines, antibodies, acute phase proteins, etc., and foreign factors, such as chemicals, products of infectious diseases. Soluble factors can be intrinsic, that is, produced by the organism, or extrinsic factors such as a virus, drug or environmental toxin. Soluble factors can be compounds of small molecules, such as prostaglandins, vitamins, metabolites (such as iron, sugars, amino acids, etc.), drugs and drug metabolites. _ '• •--s-- - •, - _. . .- • • As used herein, the term "small molecule" or "organic molecule" or "Small organic molecule" means a soluble factor or factor associated with a cell having a molecular weight in the range of 2 to 2000. Small molecules may include, but are not limited to, prostaglandins, vitamins, metabolites (such as iron, sugars, amino acids, etc.), drugs and drug metabolites. In a modality Importantly, the MLSC system is used to measure changes in the concentration of drugs and metabolites of drugs in biological fluids in tandem with other biomarkers during a treatment regimen.

As used herein, the term "disease or medical condition" means an interruption, cessation, disorder or change of the functions, systems or organs of the body in any organism. Examples of the medical condition or disease include, but are not limited to, immune and inflammatory conditions, cancer, cardiovascular disease, infectious diseases, psychiatric conditions, obesity, and other diseases. By way of illustration, immune and inflammatory conditions include autoimmune diseases, which also include rheumatoid arthritis (RA), multiple sclerosis (MS), diabetes, -etc.

As used herein, the term "clinical parameter" means information that is obtained that may be relevant to a disease or medical condition. Examples of clinical parameters include, but are not limited to, age, gender, weight, height, body type, medical history, ethnicity, family history, genetic factors, environmental factors, manifestation and categorization of the disease or medical condition, and any results of a clinical laboratory test, such as blood pressure, MRI, or x-rays, etc.

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

As used herein, the term "Laser scanning cytometry of. ': Microvolume" or "MLSC": or "MLSC system" means a method for detecting the presence of a component in a small volume of a shows, using a fluorescently labeled detection molecule, and subjecting the sample to an optical scan, where the fluorescent emission is recorded.The MLSC system has several key characteristics that distinguish it from other technologies: 1) only small amounts of blood are required (5-50 μl) for many trials, 2) Absolute cell counts (cells / μl) are obtained, and 3) the assay can be run either directly in the whole blood or in purified white blood cells. The implementation of this technology will facilitate the measurement of several hundred different cell populations from a single blood collection. MLSC technology is described in U.S. Patent Nos. 5,547,849 and 5,556,764 and in Dietz et al. (Cytometry 23: 177-186 (1996)), and the provisional patent application entitled "Confocal Time-Resolved Fluorescence Spectroscopy System Laser-Scanner," and U.S. Patent Application Serial No. 09 / 378,259, filed on August 20, 1999, entitled "Novel Optical Architectures for Microvolume Laser Scanning Cytometry", each of which is incorporated here in its entirety. Laser scanning cytometry with microvolume capillaries provides a powerful method for verifying fluorescently labeled cells in whole blood, processed blood, and other fluids, including biological fluids. The present invention further improves the MLSC technology to improve the ability of the MLSC instrument to make simultaneous measurements of multiple biological markers of a small amount of blood. The improved MLSC system of the present invention is called the "SurroScan system".

As used herein, the term "detection molecule" means any molecule capable of binding to a molecule of interest, particularly a protein. The preferred detection molecules are antibodies. The antibodies can be monoclonal or polyclonal.

As used herein, the terms "dye", "fluorophore", "fluorescent dye", "fluorescent label", or "fluorescent group" are used interchangeably to mean a molecule capable of fluorescing under excitation by a laser. The dye is typically linked directly to a detection molecule in the present invention, although indirect binding is also encompassed here. Many dyes are well known in the art. In certain preferred embodiments, fluorophores are used, which can be excited in the red region (> 600 nm) of the spectrum. Two red dyes, Cy5 and Cy5.5, are typically used. They have emission peaks of 665 and 695 nanometers, respectively, and can be easily coupled to antibodies. Both can be excited at 633 nm with a helium-neon laser. The -sets of 3 red tints that can be used include, Cy5, Cy5.5 and Cy7 or Cy5, Cy5.5 and Cy7 APC. See also, United States Provisional Patent Application No. 60 / 142,477, filed July 6, 1999, entitled "Bridged Fluorescent Dyes, Their Preparation and Their Use in Assays".

As used herein, the term "particle" means any, macromolecular structure, which is detected by MLSC, in order to obtain information on a biological marker. In some embodiments, the particle to be detected is a cell, in other embodiments, the particle to be detected is a bead marked with a f. antibody. The present invention provides an improved system of cytometry by Microvolume Laser Scanning ("MLSC"), called the SurroScan system, or simply SurroScan. The above systems are described in the Patents of the United States Numbers 5,547,849 and 5,556,764, Provisional Patent Application of the 15 United States Serial No. 60 / 131,105, entitled "Biological Marker Identification P System ", filed on April 26, 1999, Provisional Patent Application of the States United Series No. 60 / 097,506, entitled "Confocal Laser-Scanner Tíme-Resolved Fluorescence Spectroscopy System ", presented on August 21, 1998, Dietz et al.

(Citometry 23: 177-186 (1996)), and U.S. Patent Application No. 2 or Series 09 / 378,259, filed on August 20, 1999, entitled "Novel Optical Architectures for Microvolume Laser Scanning Cytometry ", each of which is incorporated here in its entirety The Imagn 2000 system, commercially available from Biometric Imaging Inc., is an example of an MLSC system.

The improved MLSC system of the present invention comprises the following components: '"• x" -' A * 1. - - ^ '* •::, -; .v :. r' (a) an MLSC instrument, Including an electronic control system, to obtain untreated data from analyte samples, (b) an image analysis system to collect and improve data without 3 or to deal with the MLSC instrument, and (c) an integrated computer architecture. for a multi-parameter assay design, instrument control, final data analysis, and data archiving.

The present invention provides significant improvements in several key aspects of the operation of the MLSC system: a) MLSC opticians; b) the electronic components of the MLSC system control, c) the representation of the image and the analysis algorithms; and d) the computer architecture. The present invention also provides improved methods for image representation and for data conversion to a Flow Cytometry Standard, industrial standard (.FCS file format).

MLSC INSTRUMENTATION The SurroScan system provides significant improvements in the optical architecture of the MLSC instruments. Previous MLSC instruments have typically been able to detect fluorescent signals in two channels, thus limiting the number of analyzes that can be detected simultaneously in a single experiment. In some applications, it is necessary to detect more than two different fluorescent signals to identify a particular cell. For example, simultaneous measurements of three or more antigens are needed to identify some cell populations, such as cells T Cándidas that express CD4, CD45RA and CD62L. The improved SurroScan instruments of the present invention are capable of detecting at least four separate fluorescent signals, thus allowing the use of at least four separate fluorescent reagents in a single experiment. One embodiment of the improved optical configurations is shown in Figure 1. A capillary array 10 contains samples for analysis. In the preferred embodiment, collimated excitation light is provided by one or more lasers. In particularly preferred embodiments, the excitation light of 633nm is provided by a He-Ne 11 laser. This wavelength avoids the problems associated with the self-fluorescence of the biological materials. The laser power is increased from 3 to 17 mW. The power of the larger laser has two potential advantages, increased sensitivity and increased scanning speed. The collimated laser light is deflected by a dichroic filter of excitation 12. With reflection, the light hits a scanning mirror driven by a galvanometer 13. The scanning mirror can oscillate rapidly in a fixed range of angles by the galvanometer, for example +/- 2.5 degrees. The scanning mirror reflects the incident light in two transmitting lenses 14 and 15 which form the image of the scanning mirror in the target entrance pupil of the microscope 16. This optical configuration converts a specific scanned angle in the mirror to a specific field position in the focus of the microscope objective. The angular sweep of +/- degrees results in a 1mm wide scan at the target's focus. The relationship between the angle of exploration and the position of the field is essentially linear in its configuration and over this range of angles. In addition, the objective of the microscope focuses the incoming collimated beam at a point in the focus plane of the objective. The diameter of the point, which adjusts the optical resolution, is determined by the diameter of the collimated beam and the focal length of the objective.

The fluorescent samples placed in the trajectory of the swept excitation beam emit light displaced by stokes (fluorescence unit). This light is collected by the target and collimated. This collimated light emerges from the two transmission lenses 14 and 15 until it collimates and collides with the scanning mirror that reflects it and does not explore it. The light shifted by stokesios (fluorescence unit), then passes through a dichroic excitation filter (which reflects a light of shorter wavelength and allows light of longer wavelength to pass through) and then through a first long-pass filter 17 which also serves to filter any reflected excitation light.

The improved instrument of the present invention then uses a series of additional dichroic filters to separate the light shifted by stokesios (fluorescence unit) in four different emission bands. A first fluorescent dichroic filter 18 divides the two bluer fluorescent colors of the two redder colors. The two bluer colors are then focused on a first aperture 19 via a first focusing lens 20, in order to significantly reduce any fluorescent signal. out of focus. After passing through the opening, a second fluorescent dichroic filter 21 further separates the individual blue colors from each other. The individual blue colors are brought to two separate photomultipliers 22 and 23. The two reddest colors are focused on a second aperture 24 via a second long pass filter 25, a mirror 26, and a second focusing lens 27, after be divided from the two bluer colors by the first fluorescent dichroic 28. • After passing through aperture 24, the two reddest colors are separated from each other by a third fluorescent dichroic 28. The individual red colors are then taken to photomultipliers 29 and 30. In this way, four separate fluorescent signals can be transmitted simultaneously from the sample held in the capillary to individual photomultipliers. This improvement, for the first time, allows four separate analyzes to be verified simultaneously. Each photomultiplier generates an electronic current in response to the incoming fluorescent photon flux. These individual currents are converted to voltages separated by one or more preamplifiers in the electronic detection elements. The voltages are sampled at regular intervals by an analog-to-digital converter, in order to determine the intensity values of the pixels for the scanned image. The four channels of the present invention are called channels 0, 1, 2, and 3.

In order to obtain meaningful data using a single excitation wavelength - for example, 33nm of the He-Ne laser - dyes are needed, which can be excited from a single excitation wavelength and which emit lengths of different waveforms, which overlap in a minimal way. For a three-channel detection system using a He-Ne laser, other suitable triple combinations of the dyes are Cy5 (emission peak at 670nm), Cy 5.5 (emission peak at 694nm) and Cy7 (emission peak at 767nm) ). In alternate modalities, allophycocyanin (APC) is replaced by Cy5. Because the absorption peak for the Cy7 (743nm) is far from the wavelength of the He-Ne excitation laser (633nm), the Cy7 would not normally be considered by those skilled in the art, to be useful in a He-Ne excitation system. However, the present inventors have found that Cy7 can be adequately excited at 633 nm to enumerate specific cells in whole blood. This excitation probably results from the presence of a long excitatory tail, as described in Mujumdar, RB, LA Ernst, SR Mujumdar, CJ Lewis, and AS Wagoner, 1993, Cyanine dye labeling reagents: sulfoindocyanine succinimidyl esters, Bioconjug Chem. : 105-11, incorporated herein by reference in its entirety. The excitation and detection of Cy7 can be improved by increasing the laser power and using detectors that are more sensitive in the red region of the spectrum. - ... », In other embodiments, the Cy7 is coupled to the APC to make a tandem dye that can be excited at the excitation wavelength of the APC, but which emits at the emission wavelength of the Cy7. This tandem dye uses donor energy transfer (APC) to excite the receptor (Cy7) as described in Beavis, AJ, and KJ Pennline, 1996, Allo-7: a new fluorescent tandem dye for use in flow cytometry, Cytometry , 24: 390-5; and Roederer, M., A. B. Cantor, D. R. Parks and L. A. Herzenberg, 1996, Cy7PE and Cy7APC: bright new probes for immunofluorescence, Cytometry, 24: 191-7, both of which are incorporated herein by reference in their entirety.

In some embodiments of the present invention, more than one excitation wavelength is used. By using more than one excitation wavelength, it is possible to use a wider variety of fluorescent dyes, since each dye does not need to 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 of 488nm, 568nm and 647nm, and can therefore be used to triple excitations of three different fluorescent groups (e.g., fluorescein, rhodamine, and Texas Red), (2) using more than one laser source, each providing a different wavelength of collimated excitation light; (3) using a laser capable of generating femtosecond pulses, such as the Ti-S laser (excitation light of ~ 700 nm) or a Nd: YLF laser (excitation light of 1047 nm), for the fluorescence excitation of the multiphoton .

Although the embodiment of the present invention described above uses four separate channels, the optical architecture described here allows the design of instruments with an even greater number of channels.

In the preferred embodiments, the sample to be scanned is mounted on a platform that is automatically translated in the X, Y and Z planes. The galvanometer-driven mirror scans the excitation beam on the Y axis, the platform moves the sample on the axis X at a constant speed. The sample interval of each analog-to-digital converter, multiplied by the speed of the sweep beam, determines the separations of the pixels on the Y axis of the image. The scanning speed of the X platform divided by the speed of the line, determines the separation of the pixels on the X axis of the image.

The platform not only scans an individual sample on the X axis, but can also transport many samples to the microscope target. In this way, many individual samples can be scanned sequentially by a computer control without any operator intervention. This greatly increases the instrument's performance, and will make the instrument even more adaptable to high-speed automated analyzes of blood samples in a clinical facility.

In the preferred embodiments of the invention, the MLSC platform of the SurroScan supports one or more arrays of capillaries, each of which has the imprint of a 96-well plate. Each capillary holds a sample to be analyzed. The disposable capillary arrangements, which have 32 fixed capillaries each and with a separation that is 15 compatible with multi-channel pipettes, are described in the United States Provisional Patent Applications, Attorney File No. 032517-005, 005 and 006, entitled, "Disposable Optical Cuvette Cartridge," "Spectrophotometric Analysis System Employing a Disposable Optical Cuvette Cartridge, "and" Vacuum Chuck for Thin Film Optical Cuvette Cartridge, "filed on April 23, 1999, and the utility application 20 commonly owned filed on April 20, 2000, entitled "Disposal Optical Cuvette Cartridge", which are hereby incorporated by reference in their entirety. Each arrangement is constructed of 2 layers of Mylar sandwiched together in a layer of double-sided adhesive, which is die cut to define the internal dimensions of the capillary. The f = resulting cartridge, called Flex-32, can be manufactured at low cost in high volumes. The cartridge is flexible, which allows it to be held on a wool base plate optically by pressure-vacuum, removing the requirements for planarity in the manufacturing process. The capillary spacing was designed to retain compatibility with multi-channel microplate pipettes and robotic devices.

In the preferred embodiments, the operator is capable of loading two plates of 32 capillaries at a time. No operator intervention is required while the plates are scanned and the images are processed. As an alternative, 16 individual capillaries designed for Imagn 2000 (VC120) are loaded into alternative fasteners.

The movement in Z of the platform provides a means to place each sample in the plane of focus of the objective. The movement in Z can also be scanned to allow the acquisition of a stack of images of the focal plane for each individual sample. The position of the optimum focus for each sample can be determined from this scanned Z-image, preferably by the computer control system, in order to avoid the need for operator intervention. While the sample is scanned on the X axis, the platform moves at a constant speed through the focus difference between the two ends, thus correcting any inclination in the sample or device.

The scanning speed of the laser beam determines the amount of time spent 15 in the integration of the optical signal in each pixel; the longer the integration time, the better the signal-to-noise ratio. The scanning speed is also proportional to the speed of system performance. The previous MLSC instruments have explored the sample at a single speed. Although this is suitable for many applications, the present invention contemplates the use of a speed system of 20 variable scan. Such a variable scan speed system allows the sensitivity of the system to be optimized for each individual sample. For example, some assays may involve the detection of analytes that are present at a very low concentration in the sample. The fluorescent signal relative to the background noise of such low analyte concentrations may be low, correspondingly. In this case, the sensitivity of the system can be increased by slowly exploring, f *. 'Allowing more time-to integrate the optical signal in each pixel, resulting in a much improved signal-to-noise ratio. may involve the detection of brighter fluorescent signals, possibly due to the relatively high concentration of the particular analyte to be detected in the sample.

In this case, a higher scanning speed would be desirable: it takes less time to integrate the signal 'in each pixel to achieve a satisfactory signal-to-noise ratio. Higher scan speeds also result in performance of the major sample. Thus, the variable scan speed system contemplated herein is a significant improvement over the fixed scan speed systems of the prior art, because a) it allows the signal to noise ratio for each analyte to be optimized, collecting at both the data with the highest possible quality for each analyte, and b) allows the system to operate at the most efficient performance speed possible. In all cases, the scanning speed can be varied by adjusting the scanning speed of the mirror mounted on the galvanometer, and by adjusting the speed at which the platform moves on the X axis during the imaging of the sample.

To optimize the sensitivity of the system at each scanning speed, the SurroScan system also provides a novel, switchable filter scheme that is incorporated into the analog processing circuit. Low pass filters are commonly used to pass the signal of interest, and to reject high noise 15 unnecessary frequency that is created by the measurement process. In the SurroScan system, the optimum filter bandwidth for each scanning speed is different, and is usually proportional to the scanning speed. In the preferred embodiments, at least 2 bandwidths are provided for each channel by the switchable filters. In the especially preferred embodiments, 4 widths of 2 0 band. Figures 2A and 2B show a circuit diagram for a switchable filter scheme that provides bandwidths of 4, 8, 12 and 16 kHz (corresponding to the optimum bandwidths for scanning speeds of 64, 128, 192 and 256 Hz respectively). In the preferred embodiments, such a bandwidth switching scheme is associated with each photomultiplier channel.

SH - '' 'So, the preSéfíte-inven ón e? A significant improvement over the MLSC systems of the prior art, because the system is optimized in two separate ways: 1) the scanning speed of the system is variable to optimize the signal-to-noise ratio, 2) the bandwidth of each analog filter in each signal channel also varies 30 to further optimize the signal-to-noise ratio. This innovative combination synergistically improves the sensitivity and efficiency of the * MLSC system and instrument.

In the preferred embodiments of the present invention, the optimal scanning speed and filter bandwidth of the SurroScan system are determined for each particular test that is performed. These variables are stored in a clinical protocol database (see below) that can then automatically select these settings when an operator subsequently chooses to run the same trial again. In this way, it is possible to have many different tests present in the same stage, the computer can automatically select the optimal scanning speed predetermined and the filter settings for each sample. This advance will contribute enormously 'with the flexibility of the SurroScan system.

Note that all the modalities described above use laser excitation of fluorophores that emit in the visible or near infrared part of the electromagnetic spectrum in order to detect particles. However, the present invention also contemplates the use of other types of electromagnetic radiation and probes of 15 emission, such as infrared radiation. In addition, the present invention contemplates the use of probe assemblies, instead of only a single probe. The present invention also contemplates the use of light scattering modes other than fluorescence, including, but not limited to, Raman scattering, Wed dispersion, luminescence and phosphorescence. 20 SURROIMAGE IMAGE ANALYSIS PROGRAM Image processing is a critical requirement for laser scanning cytometry. An image processing program needs to handle multiple binary images, which represent different regions of the spectrum of a cell or other fluorescent particles (channels), needs to determine the level of > "! • background fluorescence in each channel, the total noise in each channel, so that can enumerate the cells or other noise particles, need to ignore strange signals such as bubbles, dust particles, and other sources of" spots " or "grunge", and you need to characterize every cell or particle recognized to report parameters 30 including, but not limited to, weighted flow, size, ellipticity and relationships and correlations between the signal in other channels in the same location. The SurroScan system includes an image processing and particle detection system, called the Surrolmage system, which meets the above criteria and produces the results of the analysis in a form of text list mode.

The following description of the Surrolmage system is presented in a functional format, starting with the input file of the binary image (.sml) to the text produced in text list mode (.lsm) with the descriptions and discussions of the various algorithms involved. . Figure 3 describes a flow diagram of the operations executed by the Surrolmage system. Note also that by allowing the description that follows, the system "Surrolmage is able to detect any structure with predefined physical parameters, such as beads or spheres labeled with antibodies. The Surrolmage system is contemplated for use in any MLSC embodiment described in the prior art, including, but not limited to, the modalities described in United States Provisional Patent Application Serial No. 60 / 131,105, entitled "Identification System. of Biological Marker ", and the United States Utility Application commonly had filed concurrently with this application, entitled" Identification System of Biological Marker and Phenotype ".

Introduction In the embodiments of the invention, a binary, interlaced format is used to store the image data. Any number of 16-bit data channels (images) can be interleaved in the format illustrated in Figure 4. A channel image array is stored along each row, (Row 0; Column 0, Column 1, Column 2 , ..., Column nColumn, Row 1: ... to RowNile), where nColumn is typically 250 pixels, and nHolera is typically 10000 pixels. The SMl header, as shown in Figure 4, has 28 bytes in the header with four bytes per descriptor. • Each descriptor file is arranged in a low-high-form form. The "4-character descriptor" can be any four characters that describe a single image type, such as "SM01".

In one embodiment of the invention, the system uses two bytes or 16 bits per pixel, thus, each pixel can have any of 65536 values. However, the field descriptor, "Bytes per pixel", allows flexibility to extend the WORD image type a float, or any other data format. In addition, the variable field, "Bytes in the Header", allows the ability to add additional field descriptors. For example, a four-byte float image using this format would set BytesPerPyxel = 4, and then perhaps an additional descriptor field would be added to describe the type of format as float. The "interleaved" field gives one the option to write channels in a sequential mode. For example, in some embodiments of the invention, the scanning system gathers the channel information sequentially, rather than concurrently, for example, storing all data on channel 0 first, followed by channel 1, etc. Figure 4 shows a graphic representation of the preferred file format.

In the preferred modes, the * .SM1 file is read in the Surrolmage and each channel is stored in the memory with the manipulation descriptors. The information on each data channel is stored in a class designated Smlmagelnfo with the 15 property of image manipulation, hlm is the number of that structure.

Execution: Optional Parameters In the preferred modalities, the SurroScan is an executable command line. The following format can be used to run the program. If parameters are not given, 20 current default parameters are displayed. C: > Surrolmage. { input file SMl} . { Optional LSM output file} . { optional parameter list} where, . { input file SMl} : Full path for * .sml file "25. {Optional LSM output file.}.: Optional full path that designates the location (Output ti * .lsm. If this parameter is omitted, then the same path is used as in * .sml, including the base name, *.. {Optional parameter list.}: Multiple parameters can be assigned, separated by a space An example of format is: 30 Surrolmage C: /SMl_Files/Imagenl.sml C: /LSM_Files/Imagel.lsm ThreshRatio_1.2 Write RAWFiles. '' Optional parameters include, but are not limited to, the following: Th resh Ratio Noise multiplicative factor used to determine the threshold level of detection of the cell InumCorrelations Provides correlations outside the number. of channels NumCorrelations UseBandPassForBlob 1 = Use a filtered image to detect cells (must be mutually exclusive to use UsePeacksForBIobs UsePeacksForBIobs 1 = Use the difference between the center of the 5x5 core and the external pixels to detect cells UseFullPerimDetec 1 = Use all the pixels of the outer perimeter in conjunction with the center to locate the cells Block it Minimum cell diameter to be detected MaxCellSize Adjust the diameter of the cell so that the MaxCellSize is the diameter > MaxCellSize RowsPerNoiseBlock Number of rows to be used for peak-peak noise calculation SampleRowaPerNoiseBlock Number of rows to sample in each block for noise calculation MaxBlobPix Number of contiguous pixels on which a source image subtracted from the middle threshold would designate that particular segment as a "stain" to be added to the mask of the image MaxBubblePix Number of contiguous pixels on which a source image subtracted from the average threshold negatively 'i' - * 'would designate that particular segment as a "bubble" to be added to the mask of the image BubbleThreshFactor - Threshold * noise factor to be applied to the source image subtracted from the medium for bubble detection. Alternatively, the Noise Factor can be replaced with a baseline value (see text) BlobThreshFactor - Threshold * noise factor to be applied to the image of source stolen from the medium for spot detection. In alternative mode, the Noise Factor can be replaced with a baseline value (see text) MaskDilationPix The final mask image is dilated pixels PixDilationMask WriteRAWFIes Diagnosis: Boolean variable that indicates whether all intermediate image files should be written to the C: \ A SameCelIRadius directory Cells in alternate channels are considered the same cell if the distance between their centroids (in the float format) is less than or equal to SameCelIRadius NomCellMicrons The following three parameters determine the size of the kernel used for all calculations of the cells: Beam Microns NomCellPix = (NomCeilMicrons, BeamMicrons) / hypothetical MicronsPerPix MicronsPerPix INomCellPix = (¡nt) (NomCellPix + 1.); NomCellPix is (KernelSize - 1) / 2 PrintMode Variables listed to determine the output format of the text of the LSM file: 0 = Human readable, 1 = delimited by Tab, 2 = delimited by comma Processing of the source images of each channel: The central routine in the Surrolmage is designated SMProcessImagesQ. In the preferred modalities, the Surrolmage system performs a number of functions on each source image, ie, the image of each channel - including, but not limited to, filtering, masking, spotting and bubbling, and establishing a initial cell list. The central feature of the Surrolmage system is that each channel is analyzed independently, without any addition of individual channels taking place. Briefly, the Surrolmage system performs a number of manipulations independently of each source image, in order to remove noise and background characteristics (such as bubbles and dirt), and improve the characteristics with the spatial characteristics of the particles to be identified. The system also determines a threshold for the determination of particles in each channel, and independently identifies and analyzes the particles in each channel based on their threshold and on the parameters of the particle. The system then finds the same pixels in the remaining channels - where the particle was not detected because it was below the threshold for that channel - and measures the parameters of the particle also in those channels. In this way, the Surrolmage system collects data for each identified particle even in those channels where the particle was not originally identified.

In the preferred modalities, the Surrolmage system starts by opening the manipulators to a number of floating-point images, used to store 1) filtered source images (application of the convolution core) 2) source images subtracted from the medium and 3) images of work, used for temporary storage. In addition, a number of BYTE images are created to store the threshold versions of the previous floating point images, including a MASK image that will be discussed later.

For each channel, the preference routine begins by performing a baseline analysis. This subroutine calls the return statistics on the total variation of the baseline with respect to and (Note: Pata a future reference, x is the long capillary direction, typically 40 mm or, n Ranges = 10000 pixels, and y is the address of galvoexploration, typically of lmm or nColumns = 250 pixels). Statistical values can be stored globally, including a Boolean value, BaselineErrorFlag, which designates that the baseline has varied over a predefined limit (generally, max -min >; 0.3 medium. Figure 5 describes this process in a flowchart format. In the preferred embodiments, an average core of 15x15 is then applied to each source image using a medium high speed algorithm, designated TurboMedianQ. The kernel operates by replacing the center pixel in the 15x15 kernel with the average value of all the pixels within the kernel. The application of this half-core to each pixel acts to "uniformize" the gradual variations in the intensity of the pixel that arise along the image on the y-axis. The main role of the operation of uniformed is to eliminate the intensity contributions due to the cells, and in effect, a background representation of the image is obtained. The average image can then be subtracted from the source image and stored in a global manipulator designated hlmbgnd. This image may subsequently be used after the list of cells has been generated to determine the parameters of the cell, including, but not limited to, total flow, ellipticity, and cell diameter (also called adjustment area).

In the preferred embodiments, the multiple images are then co-voluted with u? predefined core and stored in a global manipulator designated imBlobSrc. Such convolutional nuclei are well known in the art. The structure of the chosen nucleus (the size of the nucleus and the weighted values within the nucleus) depend on the particle to be detected. For example, for the determination of cells in the blood, a 7x7 nucleus is typically used, since this nucleus is about the size of a blood cell. For the purposes of this description, it will be assumed that the convolution nucleus is a 7x7 nucleus, but it should be appreciated that other nuclei will be useful in other modalities. The result of this convolution is a filtered image that improves those characteristics with predefined spatial components, which correspond to the cell types to be detected. A threshold version of this image can be used for cellular detection and also for weighted flow calculations.

In some modalities, a "perimeter" method, instead of the convolution method described above, is used for the initial improvement of those features with a predefined spatial component that corresponds to the cell types to be detected. The perimeter method creates a differential source image - a "difference" image - and can be done in two different ways. In some modalities of the perimeter method, each pixel is adjusted to the smallest difference between it and the four external pixels of a 7x7 kernel. In other modalities, each pixel is adjusted to the smallest difference between its value and all the outer pixels of a 7x7 core. The use of these "difference" images, instead of the convolutional images, can be designated through a Boolean order line argument designated UsePeaksForBIobs. Again, the enhanced image is stored in the global manipulator mBlobSrc. Figure 6 illustrates the use of the perimeter method and the convolutional filter method in a flowchart format.

Whichever method is used for the initial improvement, the resulting image is assigned a threshold and the segmentation analysis is done to determine the locations of the cells. To set a threshold for cell detection, the noise in each source image must be determined. In the preferred embodiments, an algorithm is used that calculates peak-peak noise in segments or blocks of an image. Figure 7 illustrates this process in the flowchart format. Each block is nColumns wide (the total width of the image) and RowsPerNoiseBlock (an order line argument) long. Each noise value for each block is stored in an array with elements (int) (nRows (RowsPerNoiseBlock).) This arrangement is then multiplied by threshratio (an order line argument) and interpolated in an array of n rows? E. is used to assign a threshold The threshold assignment subroutine uses either the convolved image or the "difference" image to generate the BYTE image with threshold, imBlobSeg.

In the preferred modes, a subroutine, called MaskgrungeAndBubbles (), is called before performing segmentation or cellular detection in imBlobSeg, if the source image is that associated with channel 0. Figure 8 illustrates this subroutine in a format of Flowchart. Preferably, channel 0 is used to find bubbles and spots whose regions are added to a MASK image. This is because the dirt in the sample tends to be consistently emitted in this channel, which corresponds to the shortest emission wavelength of the sample. However, in other modes, other channels (one or more) can be used for the MASK image. : • '! "& * • ^ l '¡- X ~'. '.". '- JCl' The MASK byte image is added through three different conditions: MaskGrungeAndBubblesQ tests these conditions.Use the hlmbgnd image, the average subtracted source image, to apply the bubble and spot thresholds, BubbleThreshFactor and BlobThreshFactor (multiplied by the peak-peak noise value), "respectively, for example, with respect to bubbles, if any portion of hlmbgn is below -1 * BubbleThreshFactor * p-pNoise (bubbles means absence background fluorescence) for a particular block of the source image and if the total number of contiguous pixels exceeds MaxBubblePix, then those corresponding pixels are adjusted in a mask image to a particular value that indicates "bubbles". Similarly, spot detection is done using BlobThreshFactor * p-pNoise and MaxBlobPix. In another preferred embodiment, the threshold allocation for bubbles and spots is based on a percentage of an average baseline value instead of a peak-peak noise level factor. Thus, the threshold levels of bubbles and spots are given by BubbleThreshFactor * BaseLine (y) and BlobThreshFactor * BaseLine (y) respectively, where BaseLine (y) is the average value of the baseline evaluated over the x range of pixels for a given value of y (that is, over the width of the capillary). The final addition to the mask is done based on the filtered, segmented imBlobSeg image. It also uses the same threshold relationship as given in the order line, although it only adds to the mask if MaxBubblePix is exceeded. Rally, a pixel dilation n = MaskDilationPix (a binary dilation adjusts any background pixels to "on" if that pixel touches another pixel that is already part of a region), is done in the mask, just to ensure that the cells do not they are identified at the edges of the bubbles. An artifact of the convolution filter is that the edge of the bubble tends to convolve in a ring that can be mistakenly identified as a cell. Dilation tends to suppress this error.

In the preferred embodiments, the cells in the imBlobSeg image are then square using an 8 point connectivity rule. Figure 9 illustrates this process in a flowchart format. Any number of contiguous pixels is added to the cell list and the basic parameters are determined for each one. This includes, but is not limited to, an index, pixel values x and y maximums, total number - of pixels, a value of centroid xy based on the region of the cell with a uniform threshold, and a weighted centroid using the same pixels that exceed the threshold, but ponders those positions with a pixel value in the source image. This centroid value is a floating point value used for all future calculations. If a centroid value falls in a nonzero region in the mask (remember that each of the additions to the mask mark those pixels with a different "dentifier", so that those added due to the bubbles can be discerned from those aggregates due to spots), then that cell is deleted from the cell list. The last part of the calculation done in SMProcessImages is a histogram of the image of the mask to determine the percentage of the image that are obscured due to each of the factors mentioned above (spots, bubbles and filter artifacts). A masked parameter of the overall, global image is also calculated. This allows one to recalculate the volume of the capillary if a significant fraction is masked.

As mentioned above, the MLSC system also stores parameters in the clinical protocol database for the operation of the MLSC instrument, for example, scanning speed, filter bandwidth value, etc. The ability to fine-tune the operational parameters of the MLSC instrument with the Surrolmage system allows each test to be performed in the most efficient and sensitive manner possible.

Cellular analysis and lsm output In the preferred modalities, most cellular analysis and file output in the Surrolmage system occurs in the routine, WriteLsmFileQ. The purpose of this routine is to produce a text-based list file of all cellular events detected on any channel. In addition, the header portion of the * .LSM file contains image statistics (measured noise levels, average, median, and standard deviation statistics of the baseline level, percentages of the masked image due to bubbles and spots, and data of image creation), as well as total cellular statistics (number of cells detected in each channel, and minimum and maximum sizes). Even if only one channel has a "spot" that exceeds the threshold of ^ detection for that given channel, information on cell characteristics occurs on all channels. For example, if a "spot" was detected in channel 1 and that spot has a weighted centroid value of (x = 22.4, y = 2342.3), the center of the core 7x7 would be (22.2342) and the cellular statistics calculated on the 7x7 array would be determined in all the channel images, regardless of which channel the cell that exceeded the threshold actually has. This coordinated analysis of each channel greatly improves the accuracy of the MLSC system by ensuring that all fluorescence data for each cell is collect. In this way, many weak fluorescent signals that can, however, provide significant information - for example, if the detected molecule is present in very low concentrations - are not ignored. An example part of an * .LSM file for a 2-channel scan is shown in Table 1. The example in Table 1 lists the cellular data for two independent cellular events. In this particular example, the first cell was detected in both channels, as seen by the Event Source parameter (Note that 1 = CH0, 2 = CH1, 4 = CH2, etc., and the multiple channel detections are indicated by the sums of the values). However, the second cell was only detected on channel 0, although the parameters were still calculated for the same location on channel 1. Although it is not obvious for this example, the data output on the * .LSM is completely classified by the centroid value and. A description of how these data are generated in the preferred modalities of a cellular list of the individual channel follows.

The routine begins by sorting the cell lists in each channel. Since the "FindCell" routine adds to the list of cells any cell perimeter that locates first "walking" in the direction y, is not necessarily classified by the centroid value y.

Therefore, a bubble classification is used to generate this list (bubble classifications are the best classification algorithms when a small number of rearrangements need to take place).

The next step is to create a general cell list, which fuses the cells in the channels and is also classified by the centroid and. The details of this routine are as follows. An index of the next available cell to be processed is created for each channel, called Ce / IFirstAvailIndexfChanneIJ. The routine cycles the channels to locate the cell with the lowest centroid value, which must still be printed. This cell index and its corresponding channel number are then stored in a temporary set of variables. A list is created, Ce / IPrintListlndex Channe / sMax], which contains the indices of the cells in the alternating channels whose centroids are within SameCelIRadius of the previously located cell. To fill the nChan elements of this list, the routine cycles through the cells on all channels. However, if a cell in an alternate channel has already been "marked" as it has already been analyzed, it jumps and moves to the rest of the cells in the specified channel. (Note that with the entry to this cycle, the originating cellular index is first added to the CellPrintListIndex element [source_channel] (ie, it is marked as "to be" analyzed). Any cells whose centroid is smaller than the SameCelIRadius distance of the original cell, has its index added to the CellPrintListlndex array.

Once a single cellular event has been matched with the associated channels, it is ready to be produced in the text-based .LSM file. This PrintCellQ subroutine is called from the WriteLsmFlieQ routine and takes two arguments, the CellPrintListlndex array, which contains the indexes in the cell channel lists, and the current cell event account. The routine cycles through all the channels and accesses the centroid value of those cells indexed in the CellPríntListlndex array. Next, the routine calculates the average centroid value in x and y between the channels for the particular cell that is being evaluated. The result is rounded to the nearest complete pixel in X, Y, and is used to call another routine called AnalyzeCell () that calculates the cellular parameters in the region of the 7x7 pixel centered in X, Y. This routine is called in a cycle over the channel number The C ++ AnalyzeCellQ cellular structure is filled as follows: typedef struct. { double x, y, Area, TotalFlow, WeightedFlow, Diameter, • '! : > *? - ' ' 'Ellipticity, Brilliance; Nt Printed; / * TRUE i f printed already * /} CELLINFO; AnalyzeCellQ starts by obtaining an indicator in the imBlobSrc image, and relocating that indicator in the X, Y location of the cell. One of the parameters passed to AnalyzeCellO, besides the location X, Y and the called channel number, is a Boolean indicator that indicates if this particular channel is a "source" channel (that is, if the cell was actually detected in this channel). If it is a source channel, then the location of the maximum value found in the region of interest 7x7 (ROI) in the ImBlobSrc image is returned. If this does not match the location of the center X, Y of the kernel, then a global parameter nBlobsOfísetFromPeak, for this particular channel is incremented. In this way, the methods used to determine the location of the center of the cell could be evaluated. Furthermore, it is possible that this parameter could be added to the cell structure itself as a means to elucidate the doublets.

Regardless of whether the cell was detected or not detected in the channel that called AnalyzeCellQ, the weighted flow is calculated simply by evaluating the value of the pixel in the 15 location X, Y in the imBlobSrc image. This value of the pixel represents a weighted weighted sum of all the values of the source image pixel in the 7x7 region, weighted by a predefined 7x7 core given in Table 2 below. In another modality Other parameters evaluated in AnalyzeCellQ include, but are not limited to, flow 20 total, ellipticity and average diameter. The total flow and the total diameter are evaluated by another functional call, ComputeMeanRadiusQ. FIGURE 10 illustrates this functional call in a flowchart format. ComputeMeanRadiusQ not only calculates the average diameter, but since the total flow is calculated from the same image subtracted from the median, hlmbgnd, it is also included in this routine. Recalling, to derive: hlmbgnd, a medium filter of 15x15 pixels was applied to the source image, and the • - result was subtracted from the source image. To determine the average diameter, the centroid value is calculated first (Note: this is different from the centroid value calculated to determine the cell center, since this centroid is calculated from the pixels in a 7x7 square versus the previous centroid calculated from those pixels that exceed 30 the threshold for that channel). Next, the distance of each pixel from the centroid is - • ponders against the value of the pixel, as mathematically shown by, '• - where the centroid values Cx and Cy, are given by, P 'Wl is the value of the pixel at the location x, y, and N is 49 for a 7x7 kernel.

. This method of the moment algorithm for calculating the diameter of small particles was found to provide better performance on the two-dimensional Gaussian fit routine. The Gaussian adjustment routine, as shown in FIGURE 11, suffers from a tendency to underestimate the actual diameter for low intensity cells. This deviation, which although it is in the algorithm of the moment, is much less pronounced.

The total flow is simply given by the denominators of Equations (1) and (2). If the total flow is less or equal to zero, which can happen in images subtracted from the background, then the sum is assigned with a value of 1.0 to avoid overflows, and the average diameter is set to 0. ¡. . . . . . . . JgM; . . - Item-_ '*; '.' f '' -F¡r- Two other cellular parameters evaluated in the PrintCe / IQ routine include radio and correlation values between the channels. The radius (see example in Table 1) is given by, Wtd.Fluxm *. R-mtn = - where m > n. (3) m / n Wtd.Flux.

The Pearson correlation coefficient pm n, is calculated by where Sm and Sn, are the standard deviations of the pixel values of the source image (imSrc) in the m and n channels, respectively, and the bar represents the value of the average pixel. Each of these cell parameters are written to the * .LSM file in a sequential manner as each cell is grouped through the channels.

The WriteLSMFileQ routine is sequenced through all the cells, each time calling PrintCellQ and the AnalyzeCellQ subroutine. The total cell count is squared and written to the header portion of the * .LSM file. The file is then closed and the program is exited.

The Surrolmage system described herein is a substantial advance over the prior art systems for the detection of particles in the context of laser scanning cytometry. Such a system of the prior art is described in U.S. Patent No. 5,556,764 (the 764 patent), incorporated herein by reference in its entirety. The system described in the 764 patent first adds the images of the individual channels and then performs the particle in the resulting composite image; the 764 system also does not perform any masking of the spots and bubbles. In addition, system 764 is designed to be very selective for particular types of cells of interest in the assay, for example, by detecting cells within a certain range of size. In contrast, the present system is less restrictive, and therefore detects more different cell types. The independent channel analysis coupled with the spot and bubble masking techniques described herein allows Surrolmage to accurately identify, and collect data from, more true cells than system 764. Thus, the present invention is more accurate and sensitive than prior art systems.

Another advantage of the Surrolmage system is that it can be easily optimized for the detection of a variety of different cells with different morphologies and / or different patterns or intensities of molecular fluorescence associated with the cell. Additionally, the Surrolmage system can be rapidly optimized for the detection of different cell particles. For example, in some embodiments of the invention, the Surrolmage system is used to detect microspheres in capillaries, microspheres which bind to a particular reagent present in the blood. In contrast to the Surrolmage system, the prior art systems are capable of detecting only certain cells, they can not be reconfigured for the detection of other structures without significant operator intervention. The parameters of the individual subroutines of the Surrolmage system, such as the structure of the convolution core, can be changed quickly to optimize the detection of these particles. These parameters can be stored in a database of a clinical protocol (see below). Thus, the Surrolmage system increases the flexibility of 15 MLSC system, allowing you to perform various tests without making compromises in sensitivity.

ARCHITECTURE OF THE COMPUTERS The present invention includes a novel computer architecture that 20 performs a number of critical functions. The heart of the system is a related database that is used to coordinate all the information required to design multiple parameter tests, control measurement instrumentation, perform image and data analysis and archive results. The system comprises a number of inter-linked modules that perform discrete functions. FIGURE 11 shows a flowchart representation of the way this system operates in the :: _-.- - r: preferred modalities. Briefly, the Control-Instruments Program controls the physical elements of the SurroScan (the MLSC instrument), thereby scanning the sample and producing the raw image files (.SMl files). The files .SMl are then processed and improved by the Image Analysis Program 30 Surrolmage (previous). This module improves each image, determines the position and size of each cell (or fluorescent sphere in some applications) in each image, and then calculates the fluorescent intensity in each cell (or sphere) in each channel.

The resulting Surrolmage data is stored in a text file (.LSM file) and can then be converted to a standard format in the .FCS Industry using the FCS Conversion Program, or in any other appropriate file format for subsequent analyzes. The Instrument Control Program, the Image Analysis Program and the FCS Conversion Program are all controlled by a Clinical Protocol Database, which stores the parameters for each type of assay used in the execution of a clinical protocol. Such parameters include, but are not limited to, the scanning speed of the MLSC instrument, the value of the filter bandwidth used in the MLSC instrument, and the core structures used in the Surrolmage system. The data in the form, for example, of .FCS and .LSM file can then be exported to a server for the purpose of further data processing using, for example, a commercially available Flow Jo program. The data is also sent to a file server for experimental data to be archived and exported periodically to a tertiary environment, and also to a central database, such as an Oracle database. The central database is used, without limitations: to maintain the consistency of the clinical protocol database; as a central repository for instrument results, file names, calibration information, to store cell assay measurements and soluble factor measurements (either obtained through the MLSC system or through conventional ELISA assays); and to maintain information from clinical questionnaires.

In the preferred embodiments, the SurroScan computer system is used in the following manner for clinical studies (assuming the previous design of a scheme of an appropriate related database, and the availability of a calibrated instrument). First, the user defines the protocol of the clinical study, including information such as the number and identity of the patients, number of samples per patient, etc. The clinical study can involve tens to hundreds of patients, and can last from weeks to months. The user also defines the test protocol, which defines in detail each of the tests that will be performed on each sample of the particular patient. Each assay includes identification and detailed description for each of the reagents, including, but not limited to, the fluorophore used, target molecules, dilution compensation and fluorescence parameters. The method of sample preparation and Sample dilution are also included. The protocol also includes the information required to automatically control the SurroScan instrument and the data analysis program. After the patient samples have been processed for each test (which can be automated under the control of the database), and loaded into measuring cartridges in the SurroScan, the user enters a Protocol ID and ID parameters of the Sample in the browser program, which then interrogates the database to determine the parameters of the detailed scan, for example, scanning speed, filter bandwidth settings, platform translation speed, etc. After the scan is complete, the instrument interrogates the database again to learn the appropriate analysis parameters, and automatically performs the correct type of analysis with the Surrolmage and SurroFCS program modules, generating FCS output files. The FCS output files are further analyzed using a commercially available FCS analysis program. A summary of the FCS output data for each patient sample is generated by the FCS program, and further processed to allow storage in a related database. The results of the measurements and the patient's clinical information are then processed with various statistical and visualization methods to identify patterns and correlations that may indicate candidate biomarkers. The information of the sample and the trial is associated with the data through the analysis, from an untreated image to a format in list mode with the related database.

The present invention also contemplates the use of an image forming system to graphically represent the improved data. This system, called SurroView, represents the individual cells identified by the Surrolmage program; A box can be placed around each cell identified in order to distinguish bona fide cells from other spurious signals shaped like cells in the image. The SurroView program is particularly useful for quickly diagnosing various types of system failure modes. It should be noted that during the normal operation of the SurroScan instrument, it is not necessary for the operator to see such images of the cells.

Table i: Example of outputs in the list mode. The data corresponds to a 2-channel scan.

Table 2 Convolution kernel used to create a filtered image, imBlobSrc. ImBlobSrc is used both for cell detection and for the evaluation of a weighted flow of the cell.

Claims (6)

1. A method for analyzing a sample containing particles to detect and characterize target particles having a plurality of detectable characteristics in a capillary of fixed volume containing a fluorescent background, and which exhibits background characteristics, the method is characterized in that it comprises: a) scanning the fixed volume capillary containing the sample to generate a plurality of data channels, wherein each data channel comprises a distinctive detectable feature and a distinctive background feature: (b) sampling each of the data channels to produce corresponding sets of pixel values; (c) generating sets of improved pixel values by independently modifying each set of pixel values to selectively improve the spatial characteristics that are indicative of a target particle; (d) removing one or more sets of the improved pixel values, the distinctive background characteristic for the corresponding channel; (e) independently setting the noise threshold values for the detection of the particles for each set of improved pixel values; (g) independently identifying, in each set of improved pixel values, the groups of pixels above the threshold located in patterns that are diagnostic of the particles; (h) independently identifying, for each group of pixels above the threshold located in a diagnostic pattern in a particular set of improved pixel values, the pixels below the threshold or at the corresponding threshold in the remaining sets of improved pixel values; and '' * (i) characterizing the target particles in the sample by analyzing the pixels independently identified in steps (g) and (h); so that the particles are identified and analyzed initially in the channels with the pixels above the threshold located in diagnostic patterns of the particles, and the particles are then analyzed independently in all the Remaining channels locating the pixels in the same positions as the pixels above the threshold and initially identified.

2. An instrument for performing microvolume laser scanning (MLSC) cytometry in a sample containing particles that emit light with multiple wavelength components with illumination with an excitation light, the instrument is characterized in that it comprises: (a) a or more sources of excitation light; (b) means for scanning the excitation light at a plurality of predetermined scanning rates on the sample; (c) means for collecting and not exploring the light emitted by the sample in response to the excitation light; (d) a plurality of dichroic filter means for separating the emitted light into a plurality of wavelengths of distinctive components, and for directing each of the wavelengths of the distinctive component along a separate optical path; (e) a plurality of light detectors sensitive to emitted light, each light detector is positioned to collect a wavelength of the distinctive component, each light detector is operatively coupled to analog filter means with a variable bandwidth; and (f) digital sampling means operatively coupled to each analog filter means and capable of operating at a predetermined number of digital sampling rates to provide an output of a digital signal from each light detector; wherein the speed of the scan, the bandwidth of each analog filter medium, and the digital sampling rate can be adjusted in a coordinated manner in order to optimize the detection of the particles. '*' - 's "!" ^

3. An instrument for performing microvolume laser scanning (MLSC) cytometry in a sample containing particles that emit light with at least four distinctive wavelength components, with illumination with an excitation light, the instrument is characterized in that it comprises: (a) one or more sources of excitation light; (b) means for scanning the excitation light on the sample; (c) means for collecting and not exploring the light emitted by the sample in response to the excitation light; (d) a plurality of dichroic filter means for separating the emitted light into at least four wavelengths of distinctive components, and for directing each wavelength of the distinctive component along a separate optical path; and (e) at least four light detectors sensitive to emitted light, each light detector is positioned to collect a wavelength of the distinctive component, each light detector being operatively coupled to digital sampling means.

4. An integrated system for conducting microcycling laser scanning (MLSC) cytometry in a sample obtained from an organism, the sample contains particles that emit light with multiple wavelength components with illumination with an excitation light, the system is characterized because it comprises: (a) one or more sources of excitation light; (b) means for scanning the excitation light at a plurality of predetermined scanning rates on the sample; (c) means for collecting and not exploring the light emitted by the sample in response to the excitation light; (d) a plurality of dichroic filter means for separating the emitted light into a plurality of distinctive component wavelengths, and for directing each wavelength of the distinctive component along a separate optical path; (e) a plurality of light detectors responsive to the emitted light, each light detector is positioned to collect a wavelength of the distinctive component, each light detector is operatively coupled to analogous filter means with variable bandwidths; : . . , * .. *. * - *. . *. ,,. "... (f) digital sampling means operatively coupled to each analog filter means and capable of operating at a predetermined number of digital sampling rates to provide a digital signal output for each light detector; Y (g) a data control and analysis system comprising one or more computers and capable of: • • • (i) storing clinical protocols, wherein each clinical protocol comprises information about: the scanning speed required for the optimal detection of the particles; the bandwidth of each analog filter medium required for optimal detection of the particles; the digital sampling rate required for optimal detection of the particles; the test conditions of the sample; the medical history of the organisms; and algorithms for the analysis of the digital signal output; (ii) controlling the scanning means, the analogous filter means, and the digital sampling means using the information of the clinical protocols; (iii) receiving and storing the digital signal output; (iv) analyze the output of the digital signal using information from the clinical protocols to provide data related to the particles; (v) store the information relating the medical conditions to the data with respect to the particles; (vi) determine the medical condition of the organism using the information of (5); and (vii) generate new information relating the medical conditions to the data with respect to the particles.

5. In a method for analyzing a sample containing particles to detect and characterize target particles having a plurality of detectable characteristics in a capillary of fixed volume, which contains a fluorescent background, and which exhibits background characteristics; the method is characterized in that it comprises: (a) scanning the fixed volume capillary containing the sample to generate a plurality of data channels, wherein each data channel comprises a distinctive detectable feature and a distinctive background feature; (b) sampling each of the data channels to produce corresponding sets of source pixel values; (c) adding the sets of the source pixel values to generate a composite image; (d) calculating a threshold for the detection of the particle in the composite image; (e) performing the detection of the particle in the composite image using the threshold; (f) identifying, for each particle identified in the composite image, the corresponding pixels in the sets of the source pixel values; and (g) analyzing the pixels identified in step (f); the improvement comprises: (i) calculating the threshold for the detection of the particle independently in each set of pixel values of origin; (ii) performing the detection of the particle independently in each set of the pixel values of origin using the corresponding threshold; (iíi) identifying, for each particle identified in a particular set of source pixel values in step (2), the corresponding pixels in the remaining sets of the source pixel values; and (v) analyze the pixels identified in steps (2) and (3).

6. In a method for analyzing a sample containing particles to detect target particles having a plurality of detectable characteristics in a capillary of fixed volume, containing a fluorescent background, and which exhibits background characteristics, the method is characterized in that it comprises: ( a) scanning the fixed volume capillary containing the sample to generate a plurality of data channels, wherein each data channel comprises a distinctive detective characteristic and a distinctive background feature; (b) sampling each of the data channels to produce corresponding sets of source pixel values; (c) adding the sets of the source pixel values to generate a composite image; (d) calculating a threshold for the detection of the particle in the composite image; (e) performing the detection of the particle in the composite image using the threshold; the improvement comprises: (i) calculating the threshold for particle detection independently in each set of source pixel values without first adding the source images; and (? i) performing the detection of the particle independently in each set of the pixel values of origin using the corresponding threshold.