Patent Grant
7634125
Cellular hematopathologies have been traditionally identified and studied by a variety of slide based techniques that include morphological analysis of May-Grunwald/Giemsa or Wright/Giemsa stained blood films and cytoenzymology. Additionally, other techniques, such as cell population analysis by flow cytometry, and molecular methods, such as polymerase chain reaction (PCR) or in situ hybridization to determine gene expression, gene mutations, chromosomal translocations and duplications, have added to the understanding of these pathologies.
Although progress has been made using such techniques in advancing diagnostic capabilities, understanding the mechanisms and the progression of disease, as well as evaluating new therapeutics, such technologies each offer challenges with regard to standardization and robustness, and to a large degree, they have not yet evolved to become routine laboratory tests.
The conventional hematology clinical laboratory includes technologies to rapidly and automatically analyze large numbers of samples of peripheral blood, with minimal human intervention. Companies such as Abbott Laboratories (Abbott Park, Ill.), Beckman Coulter Inc. (Fullerton, Calif.), and TOA Corporation (Kobe, Japan) continue to advance these technologies with regard to throughput levels, the degree of accuracy of the analysis, as well as moderately increasing the information content gathered in each sample run. However, in regard to any sample suggestive of a cellular hematopathology, i.e., falling outside the accepted degree of variance for any particular parameter, traditional slide based methodologies are largely used to determine the probable cause of the abnormality.
Diagnostic criteria in hematology are based on the morphological identification of abnormalities in cell numbers, size, shape and staining patterns. Although these have been supplemented over the past decades with cell population analysis, by staining with monoclonal antibodies to various cell surface determinants and acquiring data via flow cytometry, the most important element in the diagnostic evaluation is the visual inspection of the peripheral blood film, bone marrow and lymph node biopsy using a microscope, which enables a subjective categorization of putative abnormalities.
The manual evaluation of tissue and blood films from patients is tedious, time consuming, and subject to significant intra-laboratory and intra-observer variability. This process suffers from many sources of variability and error, including staining variability (which adversely affects longitudinal analysis), bias of the evaluator, and suboptimal sample preparation (blood films with increased “smudge” cells and atypical lymphocytes). The manual classification of a few hundred cells by morphological appearance results in poor statistical power and low confidence in evaluating observed changes over time, or as a result of treatment.
Chronic lymphocytic leukemia (CLL) is a type of cancer in which the bone marrow produces an excess of lymphocytes (a type of white blood cell) due to a malignant transformation event (e.g., chromosomal translocation). CLL is the most frequent type of leukemia in the Western world. Normally, stem cells (immature cells) develop into mature blood cells by a process of ordered differentiation, which occurs in the bone marrow. There are three types of mature blood cells: (1) red blood cells that carry oxygen to all tissues of the body; (2) white blood cells that fight infection; and, (3) platelets that help prevent bleeding by forming blood clots. Normally, the numbers and types of these blood cells are tightly regulated. In CLL, there is a chronic pathological overproduction of a type of white blood cell called lymphocytes. There are three types of lymphocytes: (1) B lymphocytes that make antibodies to help fight infection; (2) T lymphocytes that help B lymphocytes make antibodies to fight infection; and, (3) killer cells that attack cancer cells and viruses. CLL is a disease involving an increase in B lymphocyte cell numbers in the peripheral blood, usually reflective of a clonal expansion of a malignantly transformed CD5+B lymphocyte cell.
Currently, established chemotherapeutic treatments are used to treat this condition, but a number of newer therapeutics, involving monoclonal antibodies to cell surface antigens expressed on CLL cells (e.g., Rituximab), have been developed. Recent data from the National Cancer Data Base indicate that the 5-year survival for this disease condition is about 48%, with only 23% of patients surviving the disease condition after 10 years. Recently, a number of prognostic factors have been identified that allow stratification of the patient population into two subpopulations with distinct clinical outcomes. Factors that tend to correlate with decreased survival are: ZAP7O expression (a tyrosine kinase required for T lymphocyte cell signaling), increased CD38 expression, un-mutated Ig Vh genes, and chromosomal abnormalities. However, routine assessment of these factors has not evolved to a standard clinical practice, due to technical challenges with data standardization and interpretation.
Morphological evaluation remains the “gold standard” in the assessment of hematopathologies, and patients with CLL present with morphological heterogeneity. Attempts to correlate a particular morphological profile with clinical prognosis have been made, but to date, no association has been widely accepted, and the morphologic sub-classification of CLL and its correlation with clinical prognosis remains to be explored.
It would therefore be desirable to provide a method and apparatus suitable for automatically analyzing blood, including peripheral blood leukocytes, and cellular components such as bone marrow and lymph nodes (whose cells are readily amenable to being processed in suspension), to facilitate researching blood related diseases and abnormalities. It would be particularly desirable to provide a method and apparatus for rapidly collecting imagery from blood and other bodily fluids (and cellular compartments), and to provide software tools for analyzing such imagery to identify cellular abnormalities or cellular distribution abnormalities associated with a disease condition.
Aspects of the concepts disclosed herein relate to the collection of multispectral images from a population of cells, and the analysis of the collected images to measure at least one characteristic of the population, using photometric and/or morphometric markers identifiable in the collection of images, where the marker is associated with a disease condition. The term marker is intended to refer to an optical or spatial characteristic of a cell (or a group of cells) that is determined using one or more images of that cell (or that group of cells). In an exemplary application, the cells are obtained from bodily fluids and cellular compartments, and in a particularly preferred implementation, from blood, most preferably where the cellular compartments are bone marrow and lymph nodes. In a further particularly preferred implementation, both photometric and morphometric markers are used in the analysis. In a particularly preferred, but not limiting implementation, the plurality of images for each individual object are collected simultaneously.
Exemplary steps that can be used to analyze biological cells in accord with an aspect of the concepts disclosed herein includes collecting image data from a population of cells, and identifying one or more subpopulations of cells from the image data. In one implementation, a subpopulation corresponding to cells exhibiting abnormalities associated with a disease condition is identified. Such subpopulations can be identified based on empirical evidence indicating that one or more photometric and/or morphometric features are typically associated with the cellular abnormality associated with disease condition. For example, photometric and/or morphometric data from the collected images are analyzed. Such data can relate to one or more features of the cells. The term feature is intended to refer to a particular structure, region, characteristic, property, or portion of cell that can be readily discerned from one or more images of the cell. The photometric and/or morphometric data from the collected images are analyzed to enable at least one characteristic of a selected feature to be measured. Characteristics that have been empirically associated with the cellular abnormalities present during a particular disease condition can be detected in the data to determine whether a particular disease condition is present in the population of cells originally imaged.
In yet another implementation, a disease condition may be detected even when the cells themselves do not exhibit any abnormalities that can be identified by photometric and/or morphometric parameters. In such an implementation, a sample will include a plurality of different subpopulations, each of which is identified by its normal characteristic morphometric and photometric markers. Where a disease condition is not present, the ratio of the subpopulations relative to one another will vary within a determinable range across different patients. Where a malignant disease condition is present, the disease condition can alter the ratio of subpopulations, such that a change in the ratio beyond a normal range can indicate the presence of a disease condition. For example, CLL (the disease condition discussed in the Background above) alters the ratio of lymphocytes in blood. While the lymphocytes themselves may not exhibit any abnormalities, an increase in the number of lymphocytes beyond a normal range is indicative of the disease condition, which may be a consequence of a normal response to an infection, or a malignant transformation event.
Consider a population of blood cells from healthy patient. The ratio of lymphocytes to other types of blood cells can be determined by analyzing image data of the entire population of blood cells to classify the images according to blood cell type. When this same process is applied to a population of blood cells from a patient with CLL, the ratio of lymphocytes to other types of blood cells will be significantly different than the ratio identified in a patient not afflicted with CLL. Thus, a disease condition can be detected by analyzing a population of cells to identify subpopulations present in the population, and by determining changes in the ratios of the subpopulations that suggest the presence of a disease condition.
In yet another implementation, a disease condition may be detected by the presence of an uncharacteristic cell type. In such an implementation, a sample will include a plurality of different subpopulations, each of which is identified by its characteristic morphometric and photometric markers. Where a disease condition is not present, only the expected subpopulations will be evident within the sample, though they vary within a determinable range across different patients. Where a disease condition is present, an entirely atypical cell type may be evident in the sample. For example, metastatic cancer of the breast may be evidenced by the presence of distinctive epithelial cells at some level in the blood. Thus, a disease condition can be detected by analyzing a population of cells to identify subpopulations present in the population, and determining the prevalence of atypical subpopulations that suggest the presence of a disease condition. The disease condition may be further refined by analyzing the morphometric and photometric markers of the atypical cell population to determine if it includes characteristic subpopulations. For example, the presence of a large fraction of rapidly dividing cells, as evidenced by a marker defining a high nuclear to cellular size ratio, may characterize a cancer as aggressive.
In still another implementation, a disease condition may be detected by the analysis not only of the cell subpopulations and their relative abundance, but also by an analysis of free (not cell-associated) bio-molecules within the cell sample. In such an implementation, a reagent may be added to the cell sample, the reagent comprising reactive substrates, each of which indicates the abundance of a particular bio-molecule. Each reactive substrate (e.g. a microsphere) includes a unique optical signature that both identifies the species of bio-molecule to which it preferentially binds, as well as indicating the abundance of that bio-molecule in the sample. By analyzing the imagery of a co-mingled sample of reactive substrates and cells, the former may be distinguished from the latter, and both a molecular and cellular analysis can be performed on the sample in a multiplexed fashion.
Image data for the population and subpopulation(s) can be manipulated using several different techniques. An exemplary technique is referred to as gating, a manipulation of data relating to photometric or morphometric imaging. A further exemplary technique is backgating, which involves further defining a subset of the gated data. While not strictly required, signal processing is preferably performed on the collected image data to reduce crosstalk and enhance spatial resolution, particularly for image data collected using simultaneous multi-channel imaging.
In a particularly preferred implementation, image data from a population of cells exhibiting a disease condition are collected. One or more photometric or morphometric markers associated with the disease condition are identified. As noted above, such a marker may be indicative of a measurable difference of some parameter between a healthy cell and a diseased cell. Such photometric or morphometric markers used to distinguish healthy cells from diseased cells are generally associated with specific features. The identified marker can represent data present in image data collected from diseased cells, but not likely to be present in image data collected from healthy cells. The identified marker can also represent data present in image data collected from cells exhibiting the disease condition, and also likely to be present in image data collected from healthy cells, yet present to a different degree that is quantifiable and identifiable. It should also be recognized that the marker can represent a measurable change in subpopulations associated with a disease condition, as opposed to subpopulations associated with the absence of the disease condition. Using the example provided above, an increase in the number of lymphocytes in blood relative to other blood cell types is indicative of the CLL disease condition.
Once one or more identifying markers have been empirically established, a population of cells can be imaged and analyzed to determine whether the identifying marker(s) is/are present in the sample population, and to determine whether the disease condition is present. In a particularly preferred, yet not limiting implementation, the disease condition is chronic lymphocytic leukemia, and the marker relates to an increase in the size or shape of the lymphocytic cellular subpopulation.
To facilitate analysis, at least one aspect of the concepts disclosed herein is directed to labeling either diseased cells or healthy cells, and imaging a mixed population of healthy and diseased cells together, such that the identifying markers are determined from a mixed population of cells. The labels enable a subpopulation of labeled cells to be extracted from the imaged data collected from the mixed population sample. The labeling thus facilitates separating the aggregate image data into images corresponding to diseased cells and images corresponding to healthy cells, which enables the photometric and/or morphometric markers corresponding to the disease condition to be more readily identified.
Yet another aspect of the techniques disclosed herein relates to monitoring the treatment of a patient exhibiting a disease condition. Baseline data are collected by imaging a population of cells from the patient before treatment. Preferably, the population of cells is obtained from a bodily fluid, such as blood. During the course of treatment, additional data are obtained by imaging additional populations of cells collected from the patient during and after various stages of the treatment process. Such data will provide a quantitative indication of the improved condition of the patient suffering from the disease condition, as indicated by either the amount of cells expressing the disease condition versus normal cells, or by a change in a ratio of the subpopulations present in the population. Significantly, such quantification is not feasible with standard microscopy and/or conventional flow cytometry.
In a preferred implementation of the techniques disclosed herein, the imagery collected from a population of biological cells includes collection of multimodal images. That is, the images collected will include at least two of the following types of images: one or more images corresponding to light emitted from the cell, one or more images corresponding to light transmitted by the cell, and one or more images corresponding to light scattered by the cell. Such multimode imaging can encompass any of the following types of images or combinations: (1) one or more fluorescent images and at least one bright field image; (2) one or more fluorescent images and at least one dark field image; (3) one or more fluorescent images, a bright field image, and a dark field image; and (4) a bright field image. Simultaneous collection of a plurality of different fluorescent images (separated by spectrum) can also be beneficial, as well as simultaneous collection of a plurality of different bright field images (using transmitted light with two different spectral filters). Preferably, the multimode images are collected simultaneously.
This Summary has been provided to introduce a few concepts in a simplified form that are further described in detail below in the Description. However, this Summary is not intended to identify key or essential features of the claimed subject matter, nor is it intended to be used as an aid in determining the scope of the claimed subject matter.
Various aspects and attendant advantages of one or more exemplary embodiments and modifications thereto will become more readily appreciated as the same becomes better understood by reference to the following detailed description, when taken in conjunction with the accompanying drawings, wherein:
Figures and Disclosed Embodiments are not Limiting
Exemplary embodiments are illustrated in referenced Figures of the drawings. It is intended that the embodiments and Figures disclosed herein are to be considered illustrative rather than restrictive.
Overview
The present disclosure encompasses a method of using flow imaging systems that can combine the speed, sample handling, and cell sorting capabilities of flow cytometry with the imagery, sensitivity, and resolution of multiple forms of microscopy and full visible/near infrared spectral analysis to collect and analyze data relating to disease conditions in blood, particularly detecting and monitoring chronic lymphocytic leukemia.
An aspect of the concepts disclosed herein relates to a system and method for imaging and analyzing biological cells entrained in a flow of fluid. In at least one embodiment, a plurality of images of biological cells are collected simultaneously; the plurality of images including at least two of the following types of images: a bright field image, a dark field image, and a fluorescent image. Images are collected for a population of biological cells. Once the imagery has been collected, the images can be processed to identify a subpopulation of images, where the subpopulation shares photometric and/or morphometric characteristics empirically determined to be associated with a disease condition.
With respect to the following disclosure, and the claims that follow, it should be understood that the term “population of cells” refers to a group of cells including a plurality of objects. Thus, a population of cells must include more than one cell.
A preferred imaging system to be used in collecting the image data required to implement the techniques disclosed herein will incorporate the following principal characteristics:
In particular, a recently developed imaging flow cytometer technology, which is embodied in an ImageStream™ instrument (Amnis Corporation, Seattle Wash.), makes great strides in achieving each of the above-noted principle characteristics. The ImageStream™ instrument is a commercial embodiment of the flow imaging systems described below in detail with respect to
As noted above, in addition to collecting image data from a population of biological cells, an aspect of the concepts disclosed herein involves processing the image data collected to measure at least one characteristic associated with a disease condition in the imaged population. A preferred image analysis software package is IDEAS™ (Amnis Corporation, Seattle Wash.). The IDEAS™ package evaluates nearly 200 features for every cell, including multiple morphologic and fluorescence intensity measurements, which can be used to define and characterize cell populations. The IDEAS™ package enables the user to define biologically relevant cell subpopulations, and analyze subpopulations using standard cytometry analyses, such as gating and backgating. It should be understood, however, that other image analysis methods or software packages can be implemented to apply the concepts disclosed herein, and the preferred image analysis software package that is disclosed is intended to be exemplary, rather than limiting of the concepts disclosed herein.
Overview of a Preferred Imaging System
Moving objects 502 are illuminated using a light source 506. The light source may be a laser, a light emitting diode, a filament lamp, a gas discharge arc lamp, or other suitable light emitting source, and the system may include optical conditioning elements such as lenses, apertures, and filters that are employed to deliver broadband or one or more desired wavelengths or wavebands of light to the object with an intensity required for detection of the velocity and one or more other characteristics of the object. Light from the object is split into two light paths by a beam splitter 503. Light traveling along one of the light paths is directed to the velocity detector subsystem, and light traveling along the other light path is directed to TDI imaging detector 508. A plurality of lenses 507 are used to direct light along the paths in a desired direction, and to focus the light. Although not shown, a filter or a set of filters can be included to deliver to the velocity detection subsystem and/or TDI imaging detector 508, only a narrow band of wavelengths of the light corresponding to, for example, the wavelengths emitted by fluorescent or phosphorescent molecules in/on the object, or light having the wavelength(s) provided by the light source 506, so that light from undesired sources is substantially eliminated.
The velocity detector subsystem includes an optical grating 505a that amplitude modulates light from the object, a light sensitive detector 505b (such as a photomultiplier tube or a solid-state photodetector), a signal conditioning unit 505c, a velocity computation unit 505d, and a timing control unit 505e, which assures that TDI imaging detector 508 is synchronized to the flow of fluid 504 through the system. The optical grating preferably comprises a plurality of alternating transparent and opaque bars that modulate the light received from the object, producing modulated light having a frequency of modulation that corresponds to the velocity of the object from which the light was received. Preferably, the optical magnification and the ruling pitch of the optical grating are chosen such that the widths of the bars are approximately the size of the objects being illuminated. Thus, the light collected from cells or other objects is alternately blocked and transmitted through the ruling of the optical grating as the object traverses the interrogation region, i.e., the field of view. The modulated light is directed toward a light sensitive detector, producing a signal that can be analyzed by a processor to determine the velocity of the object. The velocity measurement subsystem is used to provide timing signals to TDI imaging detector 508.
Preferably, signal conditioning unit 505c comprises a programmable computing device, although an ASIC chip or a digital oscilloscope can also be used for this purpose. The frequency of the photodetector signal is measured, and the velocity of the object is computed as a function of that frequency. The velocity dependent signal is periodically delivered to a TDI detector timing control 505e to adjust the clock rate of TDI imaging detector 508. Those of ordinary skill in the art will recognize that the TDI detector clock rate is adjusted to match the velocity of the image of the object over the TDI detector to within a small tolerance selected to ensure that longitudinal image smearing in the output signal of the TDI detector is within acceptable limits. The velocity update rate must occur frequently enough to keep the clock frequency within the tolerance band as flow (object) velocity varies.
Beam splitter 503 has been employed to divert a portion of light from an object 502 to light sensitive detector 505b, and a portion of light from object 502a to TDI imaging detector 508. In the light path directed toward TDI imaging detector 508, there is a plurality of stacked dichroic filters 509, which separate light from object 502a into a plurality of wavelengths. One of lenses 507 is used to form an image of object 502a on TDI imaging detector 508.
The theory of operation of a TDI detector like that employed in system 510 is as follows. As objects travel through a flow tube 511 (
The remaining three columns 522, 524, and 526 shown in
Additional exemplary flow imaging systems are disclosed in commonly assigned U.S. Pat. No. 6,211,955 and U.S. Pat. No. 6,608,682, the complete disclosure, specification, and drawings of which are hereby specifically incorporated herein by reference as background material. The imaging systems described above and in these two patents in detail, and incorporated herein by reference, have substantial advantages over more conventional systems employed for the acquisition of images of biological cell populations. These advantages arise from the use in several of the imaging systems of an optical dispersion system, in combination with a TDI detector that produces an output signal in response to the images of cells and other objects that are directed onto the TDI detector. Significantly, multiple images of a single object can be collected at one time. The image of each object can be spectrally decomposed to discriminate object features by absorption, scatter, reflection, or emissions, using a common TDI detector for the analysis. Other systems include a plurality of detectors, each dedicated to a single spectral channel.
These imaging systems can be employed to determine morphological, photometric, and spectral characteristics of cells and other objects by measuring optical signals including light scatter, reflection, absorption, fluorescence, phosphorescence, luminescence, etc. Morphological parameters include area, perimeter, texture or spatial frequency content, centroid position, shape (i.e., round, elliptical, barbell-shaped, etc.), volume, and ratios of selected pairs (or subsets) of these parameters. Similar parameters can also be determined for the nuclei, cytoplasm, or other sub-compartments of cells with the concepts disclosed herein. Photometric measurements with the preferred imaging system enable the determination of nuclear optical density, cytoplasm optical density, background optical density, and ratios of selected pairs of these values. An object being imaged with the concepts disclosed herein can either be stimulated into fluorescence or phosphorescence to emit light, or may be luminescent, producing light without stimulation. In each case, the light from the object is imaged on the TDI detector to use the concepts disclosed herein to determine the presence and amplitude of the emitted light, the number of discrete positions in a cell or other object from which the light signal(s) originate(s), the relative placement of the signal sources, and the color (wavelength or waveband) of the light emitted at each position in the object.
Using a Multispectral Imaging System to Analyze Bodily Fluid for a Disease Condition
As noted above, aspects of the concepts disclosed herein involve both the collection of multispectral images from a population of biological cells, and the analysis of the collected images to identify at least one photometric or morphological feature that has been empirically determined to be associated with a disease condition. Thus, an aspect of the present disclosure relates to the use of both photometric and morphometric features derived from multi-mode imagery of objects (e.g., cells) in flow to discriminate cell features in populations of cells, to facilitate the detection of the presence of a disease condition. Discussed in more detail below are methods for analyzing cells in suspension or flow, which may be combined with comprehensive multispectral imaging to provide morphometric and photometric data to enable, for example, the quantization of characteristics exhibited by both normal cells and diseased cells, to facilitate the detection of diseased or abnormal cells indicative of a disease condition. Heretofore, such methods have not been feasible with standard microscopy and/or flow cytometry.
As noted above, a preferred flow imaging system (e.g., the ImageStream™ platform) can be used to simultaneously acquire multispectral images of cells in flow, to collect image data corresponding to bright field, dark field, and four channels of fluorescence. The ImageStream™ platform is a commercial embodiment based on the imaging systems described in detail above. In general, cells are hydrodynamically focused into a core stream and orthogonally illuminated for both dark field and fluorescence imaging. The cells are simultaneously trans-illuminated via a spectrally-limited source (e.g., filtered white light or a light emitting diode) for bright field imaging. Light is collected from the cells with an imaging objective lens and is projected on a CCD array. The optical system has a numeric aperture of 0.75 and the CCD pixel size in object space is 0.5 μ2, enabling high resolution imaging at event rates of approximately 100 cells per second. Each pixel is digitized with 10 bits of intensity resolution in this example, providing a minimum dynamic range of three decades per pixel. In practice, the spread of signals over multiple pixels results in an effective dynamic range that typically exceeds four decades per image. Additionally, the sensitivity of the CCD can be independently controlled for each multispectral image, resulting in a total of approximately six decades of dynamic range across all the images associated with an object. It should be understood that while the ImageStream™ platform represents a particularly preferred flow imaging system for acquiring image data in accord with the concepts disclosed herein, the ImageStream™ platform is intended to represent an exemplary imaging system, rather than limiting the concepts disclosed. Any imaging instrument capable of collecting images of a population of biological cells sufficient to enable the image analysis described in greater detail below to be achieved can be implemented in accord with the concepts presented herein.
Referring again to the preferred imaging system, the ImageStream™ platform, prior to projection on the CCD, the light is passed through a spectral decomposition optical system that directs different spectral bands to different lateral positions across the detector (such spectral decomposition is discussed in detail above in connection with the description of the various preferred embodiments of imaging systems). With this technique, an image is optically decomposed into a set of a plurality of sub-images (preferably 6 sub-images, including: bright field, dark field, and four different fluorescent images), each sub-image corresponding to a different spectral (i.e., color) component and spatially isolated from the remaining sub-images. This process facilitates identification and quantization of signals within the cell by physically separating on the detector signals that may originate from overlapping regions of the cell. Spectral decomposition also enables multimode imaging, i.e., the simultaneous detection of bright field, dark field, and multiple colors of fluorescence. The process of spectral decomposition occurs during the image formation process, rather than via digital image processing of a conventional composite image.
The CCD may be operated using TDI to preserve sensitivity and image quality even with fast relative movement between the detector and the objects being imaged. As with any CCD, image photons are converted to photo charges in an array of pixels. However, in TDI operation, the photo charges are continuously shifted from pixel to pixel down the detector, parallel to the axis of flow. If the photo charge shift rate is synchronized with the velocity of the image of the cell, the effect is similar to physically panning a camera. Image streaking is avoided despite signal integration times that are orders of magnitude longer than in conventional flow cytometry. For example, an instrument may operate at a continuous data rate of approximately 30 mega pixels per second and integrate signals from each object for 10 milliseconds, enabling the detection of even faint fluorescent probes within cell images to be acquired at relatively high speed. Careful attention to pump and fluidic system design to achieve highly laminar, non-pulsatile flow eliminates any cell rotation or lateral translation on the time scale of the imaging process (see, e.g., U.S. Pat. No. 6,532,061).
A real-time algorithm analyzes every pixel read from the CCD to detect the presence of object images and calculate a number of basic morphometric and photometric features, which can be used as criteria for data storage. Data files encompassing 10,000-20,000 cells are typically about 100 MB in size and, therefore, can be stored and analyzed using standard personal computers. The TDI readout process operates continuously without any “dead time,” which means every cell can be imaged and the coincidental imaging of two or more cells at a time either in contact or not, presents no barrier to data acquisition.
Such an imaging system can be employed to determine morphological, photometric, and spectral characteristics of cells and other objects by measuring optical signals, including light scatter, reflection, absorption, fluorescence, phosphorescence, luminescence, etc. As used herein, morphological parameters (i.e., morphometrics) may be basic (e.g., nuclear shape) or may be complex (e.g., identifying cytoplasm size as the difference between cell size and nuclear size). For example, morphological parameters may include nuclear area, perimeter, texture or spatial frequency content, centroid position, shape (i.e., round, elliptical, barbell-shaped, etc.), volume, and ratios of selected pairs of these parameters. Morphological parameters of cells may also include cytoplasm size, texture or spatial frequency content, volume, and the like. As used herein, photometric measurements with the aforementioned imaging system can enable the determination of nuclear optical density, cytoplasm optical density, background optical density, and the ratios of selected pairs of these values. An object being imaged can be stimulated into fluorescence or phosphorescence to emit light, or may be luminescent, wherein light is produced by the object without stimulation. In each case, the light from the object may be imaged on a TDI detector of the imaging system to determine the presence and amplitude of the emitted light, the number of discrete positions in a cell or other object from which the light signal(s) originate(s), the relative placement of the signal sources, and the color (wavelength or waveband) of the light emitted at each position in the object.
The present disclosure provides methods of using both photometric and morphometric features derived from multi-mode imagery of objects in flow. Such methods can be employed as a cell analyzer to determine if a marker corresponding to a disease condition is present in the population of cells imaged. As noted above, the marker can be indicative of the cellular abnormality associated with a disease condition, or the marker can be indicative of a change in a ratio of subpopulations present in the population of the cells imaged, where the change in ratio is indicative of a disease condition. Preferably the population of cells is imaged while entrained in a fluid flowing through an imaging system. As used herein, gating refers to a subset of data relating to photometric or morphometric imaging. For example, a gate may be a numerical or graphical boundary of a subset of data that can be used to define the characteristics of particles to be further analyzed. Here, gates have been defined, for example, as a plot boundary that encompasses “in focus” cells, or sperm cells with tails, or sperm cells without tails, or cells other than sperm cells, or sperm cell aggregates, or cell debris. Further, backgating may be a subset of the subset data. For example, a forward scatter versus a side scatter plot in combination with a histogram from an additional marker may be used to backgate a subset of cells within the initial subset of cells.
In using an imaging system as described herein, it should be made clear that a separate light source is not required to produce an image of the object (cell), if the object is luminescent (i.e., if the object produces light). However, many of the applications of an imaging system as described herein will require that one or more light sources be used to provide light that is incident on the object being imaged. A person having ordinary skill in the art will know that the locations of the light sources substantially affect the interaction of the incident light with the object and the kind of information that can be obtained from the images using a detector.
In addition to imaging an object with the light that is incident on it, a light source can also be used to stimulate emission of light from the object. For example, a cell having been contacted with probe conjugated to a fluorochrome (e.g., such as FITC, PE, APC, Cy3, Cy5, or Cy5.5) will fluoresce when excited by light, producing a corresponding characteristic emission spectra from any excited fluorochrome probe that can be imaged on a TDI detector. Light sources may alternatively be used for causing the excitation of fluorochrome probes on an object, enabling a TDI detector to image fluorescent spots produced by the probes on the TDI detector at different locations as a result of the spectral dispersion of the light from the object that is provided by a prism. The disposition of these fluorescent spots on the TDI detector surface will depend upon their emission spectra and their location in the object.
Each light source may produce light that can either be coherent, non-coherent, broadband, or narrowband light, depending upon the application of the imaging system desired. Thus, a tungsten filament light source can be used for applications in which a narrowband light source is not required. For applications such as stimulating the emission of fluorescence from probes, narrowband laser light is preferred, since it also enables a spectrally decomposed, non-distorted image of the object to be produced from light scattered by the object. This scattered light image will be separately resolved from the fluorescent spots produced on a TDI detector, so long as the emission spectra of any of the spots are at different wavelengths than the wavelength of the laser light. The light source can be either of the continuous wave (CW) or pulsed type, such as a pulsed laser. If a pulsed type illumination source is employed, the extended integration period associated with TDI detection can enable the integration of signals from multiple pulses. Furthermore, it is not necessary for the light to be pulsed in synchronization with the TDI detector.
Particularly for use in collecting image data for cell populations found in bodily fluids such as blood, it can be desirable to employ a 360 nm UV laser as a light source, and to optimize the optical system of the imaging system for diffraction-limited imaging performance in the 400-460 nm (DAPI emission) spectral band.
In embodiments consistent with the disclosure herein, it is to be understood that relative movement exists between the object being imaged and the imaging system. In most cases, it will be more convenient to move the object than to move the imaging system. It is also contemplated that in some cases, the object may remain stationary and the imaging system move relative to it. As a further alternative, both the imaging system and the object may be in motion, which movement may be in different directions and/or at different rates.
Exemplary High Level Method Steps
Another type of marker relates to identifying some difference between subpopulations present in a cellular population absent the disease condition, and subpopulations present in a cellular population during the disease condition. For example, CLL is a disease condition where the number of lymphocytes in blood increases relative to the numbers of other blood cell types. Thus, a change in the ratio of lymphocytes to other blood cell types can be indicative of a disease condition.
Once a photometric and/or morphometric marker associated with the disease condition is identified, image data are collected from a second population of cells, in which it is not known whether the disease condition exists or not. In a block 408 image data are collected for the second population of cells, and then the image data are analyzed for the presence of the previously identified marker, to determine whether the disease condition is present in the second population of cells.
Significantly, where the imaging systems described above are used to collect the image data from a population of cells, the image data can be collected quite rapidly. In general, the analysis (i.e., analyzing the collected image data to either initially identify a marker or to determine the presence of a previously identified marker in a population of cells) will be performed off-line, i.e., after the collection of the image data. Current implementations of imaging processing software are capable of analyzing a relatively large population of cells (i.e., tens of thousands of cells) within tens of minutes using readily available personal computers. However, it should be recognized that as more powerful computing systems are developed and become readily available, it may become possible to analyze the image data in real-time. Thus, off-line processing of the image data is intended to be exemplary, rather than limiting, and it is contemplated that real-time processing of the image data is an alternative.
Where the marker relates to some photometric and/or morphometric difference between a healthy cell and a diseased cell, before using an imaging instrument to collect image data on the first population of cells (the population known to be associated with the disease condition), it can be desirable to label either the diseased cells or the healthy cells, particularly where the first population includes a mixture of both diseased and healthy cells. This approach facilitates separating the collected image data into images corresponding to diseased cells and images corresponding to healthy cells, to facilitate identification of photometric and/or morphometric markers that can be used to distinguish the two. It should be recognized however, that the first population could include only diseased cells, and that if the image data of the first population is compared with image data of a cell population known to include only healthy cells, the photometric and/or morphometric markers that can be used to distinguish the diseased cells from the healthy cells can readily be identified.
Where the marker relates to some photometric and/or morphometric difference between subpopulations present in a cellular population absent the disease condition, and subpopulations present in a cellular population associated with disease condition, image data corresponding to the subpopulations present in a healthy cellular population must be provided before the image data corresponding to the first population of cells (the population known to be associated with the disease condition) can be analyzed to identify some photometric and/or morphometric difference between the subpopulations present in the healthy cellular population, and the subpopulations present in the cellular population having the disease condition.
While not strictly required, in a working embodiment of the techniques described herein, additional processing was implemented to reduce crosstalk and spatial resolution for the multi-channel imaging. The crosstalk reduction processing implemented is described in commonly assigned U.S. Pat. No. 6,763,149, the specification, disclosure and the drawings of which are hereby specifically incorporated herein by reference as background material. Those of ordinary skill in the art will recognize that other types of crosstalk reduction techniques could alternatively be implemented.
Identification of Exemplary Photometric and Morphometric Disease Condition Markers
In the context of the present disclosure, the multi-spectral imaging flow cytometer described above employs UV excitation capabilities and algorithms to quantitate DNA content and nuclear morphology, for the purpose of detecting and monitoring disease conditions, such as chronic lymphocytic leukemia. In addition to employing a flow imaging instrument including a 360 nm UV laser and an optical system optimized for diffraction-limited imaging performance in the 400-460 nm (DAPI emission) spectral band, an imaging processing system is employed to process the image data. A personal computer executing image processing software represents an exemplary imaging processing system. The imaging processing software incorporates algorithms enabling photometric and/or morphometric properties of cells to be determined based on images of the cells. Exemplary algorithms include masking algorithms, algorithms that define nuclear morphology, algorithms for the quantization of cell cycle histograms, algorithms for analyzing DNA content, algorithms for analyzing heterochromaticity, algorithms for analyzing N/C ratio, algorithms for analyzing granularity, algorithms for analyzing CD45 expression, and algorithms for analyzing other parameters. In addition, the imaging processing software incorporates an algorithm referred to as a classifier, a software based analysis tool that is configured to evaluate a sample population of cells to determine if any disease condition markers are present. For determining the presence of cancer cells, the classifier will analyze the images of the sample population for images having photometric and/or morphometric properties corresponding to previously identified photometric and/or morphometric properties associated with cancer cells.
For samples of cell populations being analyzed to detect CLL, the classifier will analyze the images of the sample population to separate the images into different cellular subpopulations (based on different types of blood cells), and determine if the ratios of the subpopulations indicates the presence of CLL (for example, because of a higher than normal amount of lymphocytes). Preferably, the classifier configured to detect CLL will separate blood cells into the following subpopulations: lymphocytes, monocytes, basophils, neutrophils, and eosinophils. The classifier configured to detect CLL will be based on empirical data from healthy patients and from patients with CLL. Classifier profiles for CLL can be improved by collecting and comparing classifier data for a variety of patients with the same diagnosis. Preferably, large (10,000 to 20,000-cell) data sets from each patient will be collected to assess the existence and diagnostic significance of CLL cell subpopulations for classifier optimization. Such an optimized classifier can then be used to monitor patient treatment response and assess residual disease after treatment.
Significantly, for detection of epithelial cell carcinomas, high rates of data acquisition is required. Such cells have been reported to range from 1 cell in 100,000 peripheral blood leukocytes to 1 cell in 1,000,000 peripheral blood leukocytes. The ImageStream™ cytometer and IDEAS™ analytical software package discussed above are ideally suited for this application. Imagery from peripheral blood leukocytes can be obtained in the absence of artifacts typical of preparing blood films. Large cell numbers (in the tens and hundreds of thousands) can be accumulated per sample, providing greater confidence in the analysis of subpopulations. Immunofluorescent staining with accepted markers (CD5, CDI9, etc.) can easily be correlated with morphology. The quantitative cell classifiers eliminate the subjectivity of human evaluation, giving comparisons between patients a degree of confidence previously unattainable. Longitudinal studies will also benefit greatly by the quantitative analysis, and the ability to digitally store and retrieve large numbers of cellular image files, particularly as compared to prior art techniques for the retrieval of microscope slides and/or digital photographs of relatively small numbers of cells.
Discrimination of Morphological Features Using Fluorescence-Based Methodologies
A technology employed in detection of cancer cells in a bodily fluid based on image data of a population of cells from the bodily fluid was the development of preliminary absorbance and fluorescence staining protocols for simultaneous morphological analysis of bright field and fluorescence imagery.
Initially, investigations considered the simultaneous use of chromogenic stains and fluorescent dyes. The ability of the imaging system discussed above to produce bright field imagery, as well as multiple colors of fluorescence imagery of each cell, raised the possibility of simultaneously employing both traditional chromogenic stains and fluorescent dyes for analysis. However, because chromogenic stains do not normally penetrate cell membranes of viable cells, and because the optical systems discussed above are able to collect laser side scatter imagery, it was determined that much of the information on cell granularity that was traditionally acquired via stains, such as Eosin, could be obtained using laser side scatter imagery, without the need for cell staining. Numerous cell-permeant fluorescent dyes offer nuclear morphology without the need for fixing and chromogenic staining. Based on these considerations, it was determined that fluorescence-based alternatives for discrimination of morphological features provide a better approach than traditional staining methodologies.
The primary fluorescence-based alternatives to chromogenic stains useful in conjunction with the optical systems discussed above are fluorescent DNA binding dyes. A wide variety of such dyes are excitable at 488 nm, including several SYTO dyes (Molecular Probes), DRAQ5 (BioStatus), 7-AAD, Propidium Iodide (PI), and others. These dyes are alternatives to chromogenic nuclear stains such as Toluidine Blue, Methyl Green, Crystal Violet, Nuclear Fast Red, Carmalum, Celestine Blue, and Hematoxylin. A fluorescent DNA binding dye is generally included in assay protocols developed for use with the optical systems described above, for the purposes of defining the shape and boundaries of the nucleus, its area, its texture (analogous to heterochromaticity), as well as to provide DNA content information.
IDEAS™, the software image analysis program discussed above, enables evaluation of combinations of features from different images of the same cell, in order to expand the utility of the fluorescence nuclear image. For example, the nuclear image mask can be subtracted from the bright field image mask (which covers the entire cell) as a means for generating a mask that includes only the cytoplasmic region. Once defined, the cytoplasmic mask can be used to calculate the cytoplasmic area, the N/C ratio, the relative fluorescence intensity of probes in the cytoplasm and nucleus, etc., via an intuitive “Feature Manager.” An example of a Feature Manager session for the definition of the N/C ratio is shown in
Measurement of Photometric and Morphometric Parameters
In an exemplary implementation of the concepts disclosed herein, ImageStream™ data analysis and cell classification are performed post-acquisition using the IDEAS™ software package. An annotated IDEAS™ software screen capture of an analysis of human peripheral blood is shown in
The exemplary screen shot of
The software also enables one- and two-dimensional plotting of features calculated from the imagery. Dots 55 that represent cells in the two-dimensional plots can be “clicked” to view the associated imagery in the gallery. The reverse is true as well. Cell imagery can be selected to highlight the corresponding dot in every plot in which that cell appears. In addition, gates 57 can be drawn on the plots to define subpopulations, which can then be inspected in the gallery using a “virtual cell sort” functionality. Any feature calculated from the imagery or defined by the user (i.e., selected from a list of basic and automatically combined algebraically using a simple expression builder) can be plotted. A dot plot 59a (displayed at the center left of
Area
Area of mask in pixels
Aspect Ratio
Aspect ratio of mask
Aspect Ratio Intensity
Intensity-weighted aspect ratio of mask
Centroid X
Centroid of mask in horizontal axis
Centmid X Intensity
Intensity-weighted centroid of mask in horizontal
axis
Centroid Y
Centroid of mask in vertical axis
Centmid Y Intensity
Intensity-weighted centroid of mask in vertical
axis
Frequency
Variance of intensity of pixels within mask
Major Axis
Major axis of mask in pixels
Major Axis Intensity
Intensity-weighted major axis of mask in pixels
Minor Axis
Minor axis of mask in pixels
Minor Axis Intensity
Intensity-weighted minor axis of mask in pixels
Perimeter
Number of edge pixels in mask
Spot Count
Number of spots detected in image
Similarity
Pixel intensity correlation between two images
of the same object
User-Defined Features
Any algebraic combination of imagery and masks
User-Defined Masks
Erode, dilate, threshold, Boolean combinations
Features that quantitate morphology are shown in italics in Table 1. Each feature is automatically calculated for all six types of images (dark field, bright field, and four fluorescent images, that are simultaneously captured) for each cell, when an image data set is loaded into the software.
Over 35 features are calculated per image, which amounts to over 200 features per cell in assays that employ all six images, not including user-defined features. Each cell is also assigned a unique serial number and time stamp, enabling kinetic studies over cell populations.
Selection of a Photometric/Morphometric Marker for Carcinoma Cells
It was initially proposed that bladder epithelial cells would be used to investigate morphometric differences between normal and epithelial carcinoma cells. However, the initial samples of bladder washings that were analyzed revealed that the cell number per sample was highly variable, and generally too low to be practical for use in the ImageStream™ instrument. Mammary epithelial cells were therefore used in place of bladder cells. Mammary cells were chosen because normal, primary cells of this kind are commercially available (Clonetics/InVitrogen) and will expand as adherent cells in short-term tissue culture with specialized growth media. In addition, mammary epithelial carcinoma cells derived from breast cancer metastases are available from the American Type Tissue Culture Collection (ATCC). In order to better control for tumor to tumor variability, three different mammary epithelial carcinoma cell lines were studied: HCC-1 500, HCC-1 569, and HCC-1428. These lines were established from metastases in three separate patients and were purchased from ATCC as frozen stocks. The cell lines grew adherent to plastic, were expanded by routine tissue culture methods, and used experimentally.
Normal and cancerous mammary epithelial cells were harvested separately by brief incubation with trypsin/EDTA at 37 degrees Celsius. The cells were washed once in cold phosphate buffer solution (PBS) containing 1% FCS, counted, and used experimentally. The three separate mammary epithelial carcinoma cell lines were pooled in equal proportions for the experiments described below.
Normal mammary epithelial cells were stained with a fluorescein-conjugated monoclonal antibody to the HLA Class I MHC cell surface protein by incubating the cells with the appropriate, predetermined dilution of the mAb for 30 minutes at 4 degrees C. Despite the fact that mammary carcinomas are known to down-regulate Class I MHC expression, as a precaution, the normal cells were fixed in 1% paraformaldehyde to limit passive transfer to the carcinoma cells. The combined mammary carcinoma cells lines were also fixed in 1% paraformaldehyde and added to the normal mammary cell population. DRAQ5 (BioStatus, Ltd, Leicestershire, UK), a DNA binding dye that can be excited with a 488 nm laser and emits in the red waveband, was added to the sample prior to running on the ImageStream™ instrument. The labeling of normal mammary epithelial cells with anti-Class I MHC mAb enabled the normal cells to be identified in mixes of normal and carcinoma cells, thereby providing an objective “truth” to facilitate the identification of image features distinguishing normal epithelial cell from epithelial carcinoma cells.
Normal peripheral blood was obtained from AIICells (San Diego, Calif.). Whole blood was incubated with FITC conjugated anti-CD45 mAb, which is expressed at some level on all peripheral white blood cells. Red blood cells were then lysed by incubation of the whole blood in a Becton Dickinson FACSLyse™ for 3 minutes at room temperature. The cells were washed in PBS, counted and fixed with 1% paraformaldehyde. Mammary epithelial carcinoma cells were prepared as above, fixed in 1% paraformaldehyde and added to the peripheral blood cells. DRAQ5 was then added as a nuclear stain, and the cells were run on the ImageStream™ instrument.
Image files containing image data of the cell mixes described above (normal mammary epithelial cells mixed with mammary carcinoma cells, and normal peripheral blood cells mixed with mammary carcinoma cells) were analyzed using the IDEAS™ software package with the results described below.
After performing spectral compensation on the data file, an initial visual inspection was performed to compare normal mammary epithelial cells (positive for anti-HLA-FITC) to the carcinoma cells (unstained for anti-HLA-FITC). Representative images of normal cells are shown in
When visually comparing full-color images of
The analysis shown in
Next, the normal mammary cells were distinguished from the mammary carcinoma cells using the anti-HLA-FITC marker that was applied only to the normal cells. A solid yellow histogram 85a of FITC intensity was generated and is shown in
The remaining ten histograms (i.e., histograms 85b-85k) shown in
Differential histograms 85e and 85f quantitated scatter texture using either an intensity profile gradient metric (“scatter gradient RMS”; histogram 85e) or the variance of pixel intensities (“scatter frequency”; histogram 85f), which proved more selective.
Differential histograms 85g, 85h and 85i plotted the cellular area (bright field area, histogram 85g), nuclear area (from the DNA fluorescence imagery, histogram 85h), and cytoplasmic area (cellular/nuclear area, histogram 85i). The carcinoma cell lines were generally smaller in bright field area, confirming the qualitative observations from cell imagery. While the nuclear area of the carcinoma cell lines was proportionately smaller than that of the normal cells (e.g. the Nuclear/Cellular area ratio was not discriminatory), the cytoplasmic area was significantly lower in the carcinoma cells.
Finally, differential histograms 85j and 85k plotted the nuclear mean intensity (histogram 85j) and nuclear frequency (heterochromaticity, histogram 85k), respectively. As in the case of scatter, both of these features provided some discriminatory power.
The multispectral/multimodal imagery collected by the ImageStream™ instrument and analyzed using the IDEAS™ software package in this engineered experiment revealed a number of significant differences in dark field scatter, morphology, and nuclear staining between normal epithelial and epithelial carcinoma cells. While it is well-recognized that cells adapted to tissue culture have undergone a selection process that may have altered their cellular characteristics, these data demonstrate that it is feasible to build an automated classifier that uses the morphometric and photometric features identified and described above to separate normal from transformed epithelial cells, and possibly other cell types.
A further experimental investigation analyzed image data collected from a mixture of normal peripheral blood cells and mammary carcinoma cells. As shown in
In order to determine whether the techniques disclosed herein (utilizing the flow imaging instrument system described above, which is exemplified by the ImageStream™ instrument, and imaging analysis software, which is exemplified by the IDEAS™ software package) could discriminate epithelial carcinoma cells from normal PBMC, an artificial mixture of tumor cells and normal PBMC was produced as described above. The cell mixture was labeled with an anti-CD45-FITC mAb and a fluorescent DNA binding dye in order to differentiate PBMC subpopulations, generally as described above. A comparison of the scatter vs. CD45 bivariate plots for normal peripheral blood mononuclear cells and the PBMC sample spiked with the carcinoma cells is shown in
As shown in
It should be recognized that
The results noted above were verified by visual inspection of the segregated images (i.e., the images separated into subpopulations corresponding to carcinoma cells and healthy cells using one or more of the above identified photometric and/or morphometric parameters). Image gallery data were produced from the spiked PBMC data described above.
Significantly, the above studies demonstrate the feasibility of optically discriminating a subpopulation of normal epithelial cells from a subpopulation of transformed cells by analyzing multi-spectral/multimodal image data from a mixed population of such cells, where the image data are simultaneously collected. The above studies also demonstrate the feasibility of detecting epithelial carcinoma cells in blood by analyzing multi-spectral/multimodal image data from a mixed population of such cells, where the image data are simultaneously collected.
With respect to applying the concepts described herein to a specific disease condition concept, because of the relatively high operating speed of the exemplary imaging system (˜100 cells/second or ˜350,000 cells/hour), and because of the relatively large amount of image information collected for each cell (high resolution bright field image, dark field image, and four fluorescence images), it is believed that the concept disclosed herein is particularly suitable for the detection and monitoring of chronic lymphocytic leukemia.
In such an application, a 360 nm UV laser will be incorporated into the simultaneous multispectral/multimodal imaging system, and the optics of the imaging system will be optimized for diffraction-limited imaging performance in the 400-460 nm (DAPI emission) spectral band. The exemplary imaging system used in the empirical studies detailed above (i.e., the ImageStream™ instrument) employs a solid state, 200 mW, 488 nm laser for fluorescence excitation. While such a laser wavelength excites a broad range of fluorochromes, it is not optimal for cell cycle analysis due to its inability to excite DAPI, which binds stoichiometrically to DNA. In addition, the beam is configured to have a narrow width, which improves overall sensitivity in exchange for increased measurement variation from cell to cell. Feasibility studies employing propidium iodide as a DNA stain indicate that the imaging system employing the 488 nm laser can generate cell cycle histograms having GO/GI peak coefficients of variation of ˜5%.
In order to generate high resolution cell cycle histograms for the detection of changes in DNA content associated with CLL, the DAPI optimized 360 nm UV laser will instead be used. The beam will be configured to have a relatively wide illumination cross-section (˜100 microns), so that under typical operating conditions, DAPI excitation consistency will be within 1% from cell to cell. Overall, cell cycle histogram CV is expected to be about 2-3%. In addition, the optics in the exemplary instrument used in the empirical studies discussed above are diffraction-limited from 460-750 nm, which does not cover the DAPI spectral emission band. Thus, such optics will be replaced with optics that are configured to achieve diffraction-limited imaging performance in the 400-460 nm spectral band, in order to measure detailed nuclear characteristics of diagnostic value, such as notched morphology and heterochromaticity.
Particularly for use with applying the concept disclosed herein for the detection of changes in DNA content associated with CLL, it would be desirable to provide image processing software incorporating additional masking algorithms and features that define nuclear morphology in normal samples, beyond those described above.
The morphometric feature set available in the exemplary image processing software discussed above does not include boundary contour features that quantitate nuclear lobicity, number of invaginations, and similar parameters. Because such features capture many of the qualitative observations of nuclear morphology traditionally used by hematopathologists, they would be of extremely high utility in the analysis of leukocytes. Incorporation of such algorithms and features would enable improved automated classification of normal cells, precursors, and transformed cells.
The boundary contour masking algorithm and associated features employed in the empirical studies discussed above improve cell classification between eosinophils, neutrophils, monocytes, basophils, and lymphocytes in about ⅓ of cells of each type, as a function of their orientation with respect to the imaging plane. Cells that are not in one of two preferred orientations (out of six possible orientations) do not benefit from the previously employed algorithm and features. To improve the cell classification, the boundary contour algorithm and features can be extended to consistently classify normal leukocytes, independent of their rotational orientation, which will lead to a first-pass classifier between normal and transformed cells, by increasing the statistical resolution between the expected locations of normal cell distributions, thereby improving the ability to flag abnormal cells that fall outside the expected positions. Such a classifier will also enable the features to be characterized for the morphologic differences observed between normal and transformed lymphocytes, to further improve discrimination, using the techniques generally discussed above.
To configure the imaging analysis software for the detection of changes in DNA content associated with CLL, an automated classifier will be incorporated into the software package. The automated classifier will incorporate at least one or more of the following photometric and/or morphometric parameters: DNA content, nuclear morphology, heterochromaticity, N/C ratio, granularity, CD45 expression, and other parameters. As discussed above, the classifier will be configured to analyze image data corresponding to a population of blood cells, to classify the population into the following subpopulations: lymphocytes, monocytes, basophils, neutrophils, and eosinophils.
Automated differential analysis of PBMC based on multimodal imagery simultaneously collected from cells in flow will be performed using imaging systems consistent with those described above, and imaging processing software consistent with those described above. PBMC will be stained with FITC conjugated anti-CD45 and the DNA binding dye, DAPI. Peripheral blood leukocytes will be classified in a five-part differential analysis into lymphocytes, monocytes, basophils, neutrophils, and eosinophils, generally as indicated in
Data sets from peripheral blood leukocytes from CLL patients will be acquired and analyzed, as discussed above. The classification scheme developed for normal peripheral blood leukocytes will be applied to these data sets, and the identification of CLL cells will be determined by comparison with normal profiles. Various classifiers will be evaluated to determine which segments CLL cells best exemplify, generally as described above with respect to the histograms of
In addition to the normal staining protocol utilizing anti-CD45 as a marker, peripheral blood leukocytes will be stained with monoclonal antibodies to CD5 and CD2O, plus DAPI, before image data are collected. This approach will enable the identification of the CLL cells according to accepted flow cytometric criteria. In this way, morphologic criteria can be correlated with the immunophenotype.
Analyzing large (10,000 to 20,000 white blood cell) data sets from multiple CLL patients will facilitate the optimization and selection of photometric and morphometric markers that can be used classify blood cells by subpopulation (i.e., lymphocytes, monocytes, basophils, neutrophils, and eosinophils).
Morphological heterogeneity has been observed in CLL cells; however, an accurate objective appreciation of the degree of this has not been achieved due to the technical difficulty of preparing and assessing peripheral blood films from patients consistently. Acquisition of large data sets from CLL patients using the multimodal imaging systems discussed above will enable the objective analysis of the degree of morphological heterogeneity by the imaging processing software package. The classifier(s) developed above will be applied to these data sets, and morphological heterogeneity assessed by analyzing the degree to which the particular classifier (e.g., nuclear size, N/C ratio, etc.) applies across the large populations of CLL cells. Based on this analysis, the classifier that most accurately identifies the greatest percentage of CLL cells will be optimized, so that the entire population is included by the classifier.
As noted above, when applied to CLL, the techniques disclosed herein are not being used to separate a population of cells into a subpopulation corresponding to healthy cells, and a subpopulation corresponding to diseased cells. Instead, image data collected from a population of blood cells will be used to separate the population of blood cells into subpopulations based on blood cell type (i.e., lymphocytes, monocytes, basophils, neutrophils, and eosinophils). Because CLL is associated with an increase in the amount of lymphocytes present in the blood cell population (i.e., an increase in the lymphocytes subpopulation), detecting an increase in lymphocytes provides an indication of the existence of the disease condition (i.e., CLL). While the preferred method described herein involves separating the blood cell population into a plurality of different subpopulations, it should be recognized that a CLL detection technique could be implemented simply by separating the blood cell population into a lymphocyte subpopulation and a non-lymphocyte subpopulation. Using empirical data representing average lymphocyte subpopulations in healthy patients, detection of a higher-than-average lymphocyte subpopulation provides an indication of a CLL disease condition.
In addition to initially detecting the CLL disease condition, the imaging and analysis techniques discussed in detail above can be applied to follow patients with CLL longitudinally to determine their response to treatment, stability of the clinical response, and disease relapse. Changes in peripheral blood populations, including both normal and any residual CLL, can be followed and correlated with clinical outcome.
Exemplary Computing Environment
As noted above, an aspect of the present invention involves image analysis of a plurality of images simultaneously collected from members of the population of cells. Reference has been made to an exemplary image analysis software package.
An exemplary computing system 150 suitable for implementing the image processing required in the present invention includes a processing unit 154 that is functionally coupled to an input device 152, and an output device 162, e.g., a display. Processing unit 154 include a central processing unit (CPU 158) that executes machine instructions comprising an image processing/image analysis program for implementing the functions of the present invention (analyzing a plurality of images simultaneously collected for members of a population of objects to enable at least one characteristic exhibited by members of the population to be measured). In at least one embodiment, the machine instructions implement functions generally consistent with those described above, with reference to the flowchart of
Also included in processing unit 154 are a random access memory 156 (RAM) and non-volatile memory 160, which typically includes read only memory (ROM) and some form of memory storage, such as a hard drive, optical drive, etc. These memory devices are bi-directionally coupled to CPU 158. Such storage devices are well known in the art. Machine instructions and data are temporarily loaded into RAM 156 from non-volatile memory 160. Also stored in memory are the operating system software and ancillary software. While not separately shown, it should be understood that a power supply is required to provide the electrical power needed to energize computing system 150.
Input device 152 can be any device or mechanism that facilitates input into the operating environment, including, but not limited to, a mouse, a keyboard, a microphone, a modem, a pointing device, or other input devices. While not specifically shown in
Although the concepts disclosed herein has been described in connection with the preferred form of practicing it and modifications thereto, those of ordinary skill in the art will understand that many other modifications can be made to the concepts disclosed herein within the scope of the claim that follows. Accordingly, it is not intended that the scope of the concepts disclosed herein in any way be limited by the above description, but instead be determined entirely by reference to the claims that follow.
This application is a divisional application based on application Ser. No. 11/344,941, filed on Feb. 1, 2006, now U.S. Pat. No. 7,522,758, which itself is based on a prior provisional application, Ser. No. 60/649,373, filed on Feb. 1, 2005, the benefit of the filing dates of which is hereby claimed under 35 U.S.C. §120 and 35 U.S.C. §119(e). Patent application Ser. No. 11/344,941, noted above, is also a continuation-in-part application of application Ser. No. 11/123,610, filed on May 4, 2005, now U.S. Pat. No. 7,450,229, which itself is based on a prior provisional application, Ser. No. 60/567,911, filed on May 4, 2004, and which is also a continuation-in-part of prior patent application Ser. No. 10/628,662, filed on Jul. 28, 2003, which issued as U.S. Pat. No. 6,975,400 on Dec. 13, 2005, which itself is a continuation-in-part application of prior patent application Ser. No. 09/976,257, filed on Oct. 12, 2001, which issued as U.S. Pat. No. 6,608,682 on Aug. 19, 2003, which itself is a continuation-in-part application of prior patent application Ser. No. 09/820,434, filed on Mar. 29, 2001, which issued as U.S. Pat. No. 6,473,176 on Oct. 29, 2002, which itself is a continuation-in-part application of prior patent application Ser. No. 09/538,604, filed on Mar. 29, 2000, which issued as U.S. Pat. No. 6,211,955 on Apr. 3, 2001, which itself is a continuation-in-part application of prior patent application Ser. No. 09/490,478, filed on Jan. 24, 2000, which issued as U.S. Pat. No. 6,249,341 on Jun. 19, 2001, which itself is based on prior provisional patent application Ser. No. 60/117,203, filed on Jan. 25, 1999, the benefit of the filing dates of which is hereby claimed under 35 U.S.C. §120 and 35 U.S.C. §119(e). Patent application Ser. No. 09/976,257, noted above, is also based on prior provisional application Ser. No. 60/240,125, filed on Oct. 12, 2000, the benefit of the filing date of which is hereby claimed under 35 U.S.C. § 119(e).
This invention was made with U.S. government support under grant No. R43 CA 94590-01 awarded by the National Cancer Institute. The U.S. government has certain rights in the invention.
| Number | Name | Date | Kind |
|---|---|---|---|
| 3922069 | Kishikawa et al. | Nov 1975 | A |
| 4635293 | Watanabe | Jan 1987 | A |
| 4662742 | Chupp | May 1987 | A |
| 4677680 | Harima et al. | Jun 1987 | A |
| 4770992 | Van den Engh et al. | Sep 1988 | A |
| 4777525 | Preston, Jr. | Oct 1988 | A |
| 4786165 | Yamamoto et al. | Nov 1988 | A |
| 5096807 | Leaback | Mar 1992 | A |
| 5141609 | Sweedler et al. | Aug 1992 | A |
| 5153916 | Inagaki et al. | Oct 1992 | A |
| 5159397 | Kosaka et al. | Oct 1992 | A |
| 5159398 | Maekawa et al. | Oct 1992 | A |
| 5159642 | Kosaka | Oct 1992 | A |
| 5247339 | Ogino | Sep 1993 | A |
| 5247340 | Ogino | Sep 1993 | A |
| 5257182 | Luck et al. | Oct 1993 | A |
| 5272354 | Kosaka | Dec 1993 | A |
| 5351311 | Rogers | Sep 1994 | A |
| 5422712 | Ogino | Jun 1995 | A |
| 5444527 | Kosaka | Aug 1995 | A |
| 5471294 | Ogino | Nov 1995 | A |
| 5548349 | Mizuguchi et al. | Aug 1996 | A |
| 5548395 | Kosaka | Aug 1996 | A |
| 5568315 | Shuman | Oct 1996 | A |
| 5596401 | Kusuzawa | Jan 1997 | A |
| 5621460 | Hatlestad et al. | Apr 1997 | A |
| 5633503 | Kosaka | May 1997 | A |
| 5644388 | Maekawa et al. | Jul 1997 | A |
| 5674743 | Ulmer | Oct 1997 | A |
| 5695934 | Brenner | Dec 1997 | A |
| 5733721 | Hemstreet, III et al. | Mar 1998 | A |
| 5754291 | Kain | May 1998 | A |
| 5760899 | Eismann | Jun 1998 | A |
| 5764792 | Kennealy | Jun 1998 | A |
| RE35868 | Kosaka | Aug 1998 | E |
| 5831723 | Kubota et al. | Nov 1998 | A |
| 5848123 | Strommer | Dec 1998 | A |
| 5855753 | Trau et al. | Jan 1999 | A |
| 5900942 | Spiering | May 1999 | A |
| 5929986 | Slater et al. | Jul 1999 | A |
| 5959953 | Alon | Sep 1999 | A |
| 6007994 | Ward et al. | Dec 1999 | A |
| 6007996 | McNamara et al. | Dec 1999 | A |
| 6014468 | McCarthy et al. | Jan 2000 | A |
| 6066459 | Garini et al. | May 2000 | A |
| 6116739 | Ishihara et al. | Sep 2000 | A |
| 6156465 | Cao et al. | Dec 2000 | A |
| 6210973 | Pettit | Apr 2001 | B1 |
| 6211955 | Basiji et al. | Apr 2001 | B1 |
| 6249341 | Basiji et al. | Jun 2001 | B1 |
| 6256096 | Johnson | Jul 2001 | B1 |
| 6259807 | Ravkin | Jul 2001 | B1 |
| 6330081 | Scholten | Dec 2001 | B1 |
| 6330361 | Mitchell et al. | Dec 2001 | B1 |
| 6381363 | Murching et al. | Apr 2002 | B1 |
| 6473176 | Basiji et al. | Oct 2002 | B2 |
| 6507391 | Riley et al. | Jan 2003 | B2 |
| 6522781 | Norikane et al. | Feb 2003 | B1 |
| 6532061 | Ortyn et al. | Mar 2003 | B2 |
| 6549664 | Daiber et al. | Apr 2003 | B1 |
| 6563583 | Ortyn et al. | May 2003 | B2 |
| 6580504 | Basiji et al. | Jun 2003 | B1 |
| 6583865 | Basiji et al. | Jun 2003 | B2 |
| 6608680 | Basiji et al. | Aug 2003 | B2 |
| 6608682 | Ortyn et al. | Aug 2003 | B2 |
| 6618140 | Frost et al. | Sep 2003 | B2 |
| 6620591 | Dunlay et al. | Sep 2003 | B1 |
| 6671044 | Ortyn et al. | Dec 2003 | B2 |
| 6707551 | Ortyn et al. | Mar 2004 | B2 |
| 6727066 | Kaser | Apr 2004 | B2 |
| 6763149 | Riley et al. | Jul 2004 | B2 |
| 6778263 | Ortyn et al. | Aug 2004 | B2 |
| 6875973 | Ortyn et al. | Apr 2005 | B2 |
| 6906792 | Ortyn et al. | Jun 2005 | B2 |
| 6934408 | Frost et al. | Aug 2005 | B2 |
| 6947128 | Basiji et al. | Sep 2005 | B2 |
| 6947136 | Ortyn et al. | Sep 2005 | B2 |
| 6975400 | Ortyn et al. | Dec 2005 | B2 |
| 7006710 | Riley et al. | Feb 2006 | B2 |
| 7057732 | Jorgenson et al. | Jun 2006 | B2 |
| 7079708 | Riley et al. | Jul 2006 | B2 |
| 7087877 | Ortyn et al. | Aug 2006 | B2 |
| 7190832 | Frost et al. | Mar 2007 | B2 |
| 7221457 | Jorgenson et al. | May 2007 | B2 |
| 20010006416 | Johnson | Jul 2001 | A1 |
| 20010011018 | Baum et al. | Aug 2001 | A1 |
| 20020030812 | Ortyn et al. | Mar 2002 | A1 |
| 20020126275 | Johnson | Sep 2002 | A1 |
| Number | Date | Country |
|---|---|---|
| WO 9853093 | Nov 1998 | WO |
| WO 9853300 | Nov 1998 | WO |
| WO 9924458 | May 1999 | WO |
| WO 0014545 | Jul 2000 | WO |
| WO 0042412 | Jul 2000 | WO |
| WO 0146675 | Jun 2001 | WO |
| WO 0217622 | Feb 2002 | WO |
| Number | Date | Country | |
|---|---|---|---|
| 20090003681 A1 | Jan 2009 | US |
| Number | Date | Country | |
|---|---|---|---|
| 60649373 | Feb 2005 | US | |
| 60567911 | May 2004 | US | |
| 60117203 | Jan 1999 | US | |
| 60240125 | Oct 2000 | US |
| Number | Date | Country | |
|---|---|---|---|
| Parent | 11344941 | Feb 2006 | US |
| Child | 12194842 | US |
| Number | Date | Country | |
|---|---|---|---|
| Parent | 11123610 | May 2005 | US |
| Child | 11344941 | US | |
| Parent | 10628662 | Jul 2003 | US |
| Child | 11123610 | US | |
| Parent | 09976257 | Oct 2001 | US |
| Child | 10628662 | US | |
| Parent | 09820434 | Mar 2001 | US |
| Child | 09976257 | US | |
| Parent | 09538604 | Mar 2000 | US |
| Child | 09820434 | US | |
| Parent | 09490478 | Jan 2000 | US |
| Child | 09538604 | US |
