Patent Application
20050267690
The present invention pertains to image analysis methods used to extract shape information from images of cells.
Cell shape is a recognized indicator of cell type and/or condition. Thus, it is possible to identify many cell types and to distinguish them from other cell types based on the cells' shapes. In addition, many interesting biological conditions may be correlated with cell shape. Biological “conditions” of interest to researchers include disease states, normal unperturbed states, quiescent states, states induced by exogenous biologically-active agents, etc. Valuable insight may be gained by inducing a biological condition through a genetic manipulation, exposure to a particular agent (e.g., a compound, radiation, a field, etc.), deprivation of required substance, and other perturbations. Such a condition may cause changes in a particular cell's shape, and the cell's modified shape may be indicative of that particular condition. In this regard, there known correlations between cell shape and cell condition. Agents that affect cytoskeleton, adhesion, regulatory signaling pathways, cell cycle, cause significant and specific changes in cell morphology.
In drug discovery work, valuable information can be obtained by understanding how a potential therapeutic agent affects a cell. This information may give some indication of the mechanism of action associated with the compound. As present, little or no formal effort has been made to correlate cell condition with cell shape. To the extent that cell shape analyses are performed at all, the results are typically reported in qualitative terms based on observations of cells using various microscopy techniques. Given that there are some known correlations between cell shape and cell condition, the ability to quickly determine whether a population of cells has a particular modified shape could provide a valuable tool in assessing the mechanism of action of an uncharacterized compound that has been tested on the population of cells. Therefore, it would be desirable to have improved techniques for analysis of cell shapes.
The present invention addresses this need by providing methods and apparatus for the analysis of images of cells and extraction biologically-significant shape-related features from the cell images. The extracted features may be correlated with particular conditions induced by biologically-active agents with which cells have been treated, thereby enabling the automated analysis of cells based on cell shape parameters. In particular, the invention provides methods for segmentation of cells in an image using data from a plurality of separate images of different cell components. One application of the invention involves the use of a reference cell component (preferably one that has been previously segmented and therefore one whose segmentation parameters are well understood and may be repeated) in combination with image data for a second component to perform a segmentation of another cell component or the whole cell. This application of the invention is particularly effective when the reference component has been previously segmented and is present in a single copy in the cell, such as the nucleus, centrosome, specific chromosome, Golgi complex etc. The invention further provides techniques for extraction of biologically-relevant shape-related cell features from segmented cell images.
In accordance with the present invention, image data for a reference cell component, preferably present in the cell in a single copy and/or previously segmented (for example, cell nuclei) are processed together with image data for a cell shape-indicative marker (for example, cytoskeletal components, (e.g., tubulin), one or more cytoplasmic proteins (for example lactate dehydrogenase or total cell protein), or membrane components (e.g., lipids or plasma membrane receptors)) in a watershed technique. Further, the invention provides a skeletonization and skeleton analysis technique for extracting biologically-relevant shape-related features from the segmented images.
One aspect of this invention pertains to a method of identifying boundaries of biological cells. The method involves receiving a first image of a field of one or more cells in which a reference cell component of the one or more cells is identified by a reference cell component marker image parameter, receiving a second image of the field of one or more cells in which at least one of a cell shape-indicative marker of the one or more cells is identified by a cell shape-indicative marker image parameter, and processing the first image in conjunction with the second image so that individual cell boundaries for the one or more cells in the field are identified. The image processing may involve segmenting the one or more reference cell components in the first image to generate a digital representation of the first image, called a reference cell component mask, thresholding the cell shape-indicative marker in the second image to generate a digital representation of the second image including a cell shape-indicative marker portion and a background portion, conceptually registering the nuclei mask with the digital representation of the second image, and applying a watershed algorithm to data provided by the conceptually registered reference component mask and digital representation of the second image such that individual cell boundaries for the one or more cells in the field are identified. In addition, the method may further involve extracting biologically-significant shape-related information from the field of one or more cells.
Another aspect of the invention pertains to a method of extracting biologically-significant shape-related information from a field of one or more cells. The method involves providing a segmented image of the field of one or more segmented cells, where the boundaries of the one or more segmented cells have been ascertained by the segmentation. For each of one or more of the cells in the segmented cell image two endpoints defining two parts of the boundary of the one or more cells are selected. For each part of the cell's boundary, computing the distance from each point on the part of the cell boundary to a line between the endpoints, determining a point dMAX on the portion of the boundary, the point dMAX being maximally distant from the line, and comparing the distance from the point dMAX to a predetermined threshold distance value, dTH. Where dMAX is greater than dTH, the line between the endpoints is discarded, and point dMAX as a new endpoint together with one of the original endpoints to separate the part into two new parts, and the process from determining a point dMAX is repeated. Where dMAX is less than dTH, the line is used as a side of a polygon approximating the cell shape until a polygon approximating the shape of the cell is complete. When the polygon is complete, it is skeletonized, and end points and/or nodes are calculated for the polygon approximation of the cell.
Still another aspect of the invention pertains to a method of correlating a cell's shape with a biological condition of the cell. The method involves providing a plurality of segmented images of fields of one or more segmented cells, where at least one of the fields has been treated with a biologically active agent and at least one of the fields is a control having not been treated with the biologically active agent. The boundaries of the one or more segmented cells having been ascertained by the segmentation. For each of one or more of the cells in the plurality of segmented cell images, two endpoints defining two parts of the boundary of the one or more cells are selected. For each part of the cell's boundary, computing the distance from each point on the part of the cell boundary to a line between the endpoints, determining a point dMAX on the portion of the boundary, the point dMAX being maximally distant from the line, and comparing the distance from the point dMAX to a predetermined threshold distance value, dTH. Where dMAX is greater than dTH, the line between the endpoints is discarded, and point dMAX as a new endpoint together with one of the original endpoints to separate the part into two new parts, and the process from determining a point dMAX is repeated. Where dMAX is less than dTH, the line is used as a side of a polygon approximating the cell shape until a polygon approximating the shape of the cell is complete. When the polygon is complete, it is skeletonized, and end points and/or nodes are calculated for the polygon approximation of the cell. The computations of end points and/or nodes for the polygon approximation of the cell are then computed to identify significant shape differences between the treated and control fields of one or more cells.
Yet another aspect of the present invention pertains to an image analysis apparatus for identifying individual biological cells in a field of cells. The apparatus includes a memory or buffer adapted to store, at least temporarily, a first image of a field of one or more cells, in which a reference cell component of the one or more cells is identified by a reference cell component marker image parameter, and a second image of the field of one or more cells, in which at least one of a cell shape-indicative marker of the one or more cells are identified by a cell shape-indicative marker image parameter. The apparatus further includes a processor configured or designed to process the first image in conjunction with the second image such that individual cell boundaries for the one or more cells in the field are identified.
Another aspect of the invention pertains to a method of identifying boundaries of biological cells. The method involves receiving a first image of the field of one or more cells in which a reference cell component of the one or more cells is identified by a reference cell component marker image parameter, and receiving a second image of the field of one or more cells in which at least one of a cell shape-indicative marker of the one or more cells is identified by a cell shape-indicative marker image parameter. Then, a thresholding techniques is applied the cell shape-indicative marker image parameter in the second image to generate a digital representation of the second image comprising a cell shape-indicative marker portion and a background portion, and the boundaries of individual cells are identified by applying a watershed algorithm to the second image using the reference cell component marker image parameter and the background portion of the digital representation of the second image as seeds.
Another aspect of the invention pertains to computer program products including a machine readable medium on which is stored program instructions for implementing any of the methods described above. Any of the methods of this invention may be represented as program instructions that can be provided on such computer readable media.
These and other features and advantages of the present invention will be described below in more detail with reference to the associated drawings.
Introduction
Generally, this invention relates to image analysis processes and apparatus configured for image analysis. It also relates to machine-readable media on which is provided instructions, data structures, etc. for performing the processes of this invention. In accordance with this invention, images of cells are manipulated and analyzed in certain ways to extract relevant cell shape-related features. Using those features, the apparatus and processes of this invention, can automatically draw certain conclusions about the biology of a cell.
The invention provides methods and apparatus that for the analysis of images of cells and extraction biologically-significant shape-related features from the cell images. The extracted features may be correlated with particular conditions induced by biologically-active agents (e.g., drugs or drug candidates) with which cells have been treated, thereby enabling the automated analysis of cells based on cell shape parameters. In particular, the invention provides methods for segmentation of cells in an image using data from a plurality of separate images of different cell components. One application of the invention involves the use of a reference cell component (preferably one that has been previously segmented and therefore one whose segmentation parameters are well understood and may be repeated) in combination with image data for a second component to perform a segmentation of another cell component or the whole cell. This application of the invention is particularly effective when the reference component has been previously segmented and is present in a single copy in the cell, such as the nucleus, centrosome, specific chromosome, Golgi complex etc. The invention further provides techniques for extraction of biologically-relevant shape-related cell features from segmented cell images.
In accordance with the present invention, image data for a reference cell component, preferably present in the cell in a single copy and/or previously segmented (for example, cell nuclei) are processed together with image data for a cell shape-indicative marker (for example, cytoskeletal components, (e.g., tubulin), one or more cytoplasmic proteins (for example lactate dehydrogenase or total cell protein), or membrane components (e.g., lipids or plasma membrane receptors)) in a watershed technique. Further, the invention provides a skeletonization and skeleton analysis technique for extracting biologically-relevant shape-related features from the segmented images.
The invention will now be described in terms of particular specific embodiments as depicted in the drawings. However, as will be apparent to those skilled in the art, the present invention may be practiced without the employing some of the specific details disclosed herein. Some operations or features may be dispensed with. And often alternate elements or processes may be substituted. For example, in the following description, nuclei are used as the reference cell component, and tubulin as the cell shape-indicative component to provide image data useful for whole cell segmentation. As noted above, cell components other than nuclei, particularly those previously segmented and/or present in only a single copy in a cell, and other cell shape-indicative components may also be used.
Preparation of the Image
In accordance with the present invention, images may be obtained of cells that have been treated with a chemical agent to render visible (or otherwise detectable in a region of the electromagnetic spectrum) a cellular component. A common example of such agents are colored dyes specific for a particular cellular component that is indicative of cell shape. Other such agents may include fluorescent, phosphorescent or radioactive compounds that bind directly or indirectly (e.g., via antibodies or other intermediate binding agents) to a cell component. In accordance with the present invention, a plurality of cell components may be treated with different agents and imaged separately. For example, in one embodiment, cell nuclei, a previously segmented component present in only a single copy in a cell, and tubulin, a cytoskeletal protein, or other component indicative of cell shape, are treated as described further below.
Generally the images used as the starting point for the methods of this invention are obtained from cells that have been specially treated and/or imaged under conditions that contrast markers of cellular components of interest (e.g., tubulin for the cytoskeleton, and DNA for the nuclei) from other cellular components and the background of the image. In a preferred embodiment, the cells are fixed and then treated with a material that binds to a marker for the components of interest and shows up in an image. Preferably, the chosen imaging agent binds indiscriminately with the marker, regardless of its location in the cell. The agent should provide a strong contrast to other features in a given image. To this end, the agent should be luminescent, radioactive, fluorescent, etc. Various stains and fluorescent compounds may serve this purpose.
A variety of imaging agents are available depending on the particular marker, and agents appropriate for labeling cytoskeletal, cytoplasmic, plasma membrane, nuclear, and other discrete cell components are well known in the histology art. Examples of such compounds include fluorescently labeled antibodies to cytoplasmic or cytoskeletal proteins, fluorescent dyes which bind to proteins and/or lipids, labeled ligands which bind to cell surface receptors, and fluorescent DNA intercalators and fluorescently labeled antibodies to DNA or other nuclear component which bind to the nuclei. For example, a suitable label for the cytoskeletal protein tubulin is a fluorescently labeled monoclonal antibody to tubulin, rodamine-labeled Dm1alpha, produced from hybdridoma DM1A reported in the publication Blose et. al. Journal of Cell Biology, V98, 1984, 847-858. Examples of fluorescent DNA intercalators include DAPI and Hoechst 33341 available from Molecular Probes, Inc. of Eugene, Oreg. The antibodies may be fluorescently labeled either directly or indirectly.
In a preferred embodiment, cells may be treated with more than one imaging agent, each imaging agent specific for a different cellular component of interest. The component(s) may then be separately imaged by separately illuminating the cells with an excitation frequency (channel) for the imaging agent of the marker for the component of interest. Thus different images of the same cells focussing on different cellular components may be obtained on different channels.
Various techniques for preparing and imaging appropriately treated cells are described in U.S. patent application Ser. Nos. 09/310,879, 09/311,996, and 09/311,890, previously incorporated by reference. In the case of cells treated with DAPI or other fluorescent material, a collection of such cells is illuminated with light at an excitation frequency. A detector is tuned to collect light at an emission frequency. The collected light is used to generate the image and highlights regions of high marker (e.g., tubulin or DNA) concentration.
In order to derive biologically meaningful cell shape information from an analysis of cell images, it is important to be able to distinguish one cell from another by establishing its boundaries and distinguishing it from other cells in the image (a process sometimes referred to as “segmentation”).
A high level process flow 201 in accordance with one embodiment of this invention is depicted in
In each case, the image obtained will represent the imaged marker as a corresponding “image parameter.” The image parameter will be an intensity value of light or radiation shown in the image. Often, the intensity value will be provided on a per pixel basis. In addition, the intensity value may be provided at a particular wavelength or narrow range of wavelengths that correspond to the emission frequency of an imaging agent that specifically associates with the imaged marker.
In the following discussion and the figures of the present application, the reference cell component and cell shape-indicative marker used to describe and illustrate the principles of the present invention are nuclei and the cytoskeletal protein, tubulin, respectively. However, as noted above, the invention is not limited to the use of nuclei and tubulin as the reference cell component and cell shape-indicative marker. Instead, it should be understood that cell components other than nuclei, particularly those previously segmented and/or present in only a single copy in a cell, may also be used, and that tubulin is just one example of an array of markers that may be correlated with and indicative of cell shape. Other such reference cell components and cell shape-indicative markers may be used in place of or in conjunction with nuclei and tubulin according to the principles of the invention described herein.
The relevant images obtained at 202, 204 are captured by an image acquisition system. In one embodiment, the image acquisition system is directly coupled with the image analysis tool of this invention. Alternatively, the image under consideration may be provided by a remote system unaffiliated with the image acquisition system. For example, the images may be acquired by a remote image analysis tool and stored in a database or other repository until they are ready for use by an image analysis tool of this invention.
Sometimes corrections must be made to the measured intensity. This is because the absolute magnitude of intensity can vary from image to image due to changes in the staining and/or image acquisition procedure and/or apparatus. Specific optical aberrations can be introduced by various image collection components such as lenses, filters, beam splitters, polarizers, etc. Other sources of variability may be introduced by an excitation light source, a broad band light source for optical microscopy, a detector's detection characteristics, etc. Even different areas of the same image may have different characteristics. For example, some optical elements do not provide a “flat field.” As a result, pixels near the center of the image have their intensities exaggerated in comparison to pixels at the edges of the image. A correction algorithm may be applied to compensate for this effect. Such algorithms can be easily developed for particular optical systems and parameter sets employed using those imaging systems. One simply needs to know the response of the systems under a given set of acquisition parameters.
Reference Cell Component Segmentation
After the nuclei image has been obtained at 202, the image analysis tool segments the image into discrete nuclei representations at 206. The goal of segmentation is to convert the image into discrete images/representations for the DNA of each nucleus to generate a “nuclei mask” to be used in conjunction with cell shape-indicative marker image data in a cell segmentation process in accordance with the present invention. In a preferred embodiment, each representation includes only those pixels where the DNA of a single cell nucleus is deemed to be present. Since the DNA is normally contained almost entirely within the nucleus of non-mitotic eucaryotic cells, the shape of each representation resulting from segmentation represents the boundaries within which a nucleus lies. The nuclei mask is a composite of the discrete nuclei representations providing intensity as a function of position for each nuclei in the image.
Individual cell nucleus representations may be extracted from the image by various image analysis procedures. Preferred approaches include edge finding routines and thresholding routines. Some edge finding algorithms identify pixels at locations where intensity is varying rapidly. For many applications of interest here, pixels contained within the edges will have a higher intensity than pixels outside the edges. Thresholding algorithms convert all pixels below a particular intensity value to zero intensity in an image subregion (or the entire image, depending upon the specific algorithm). The threshold value is chosen to discriminate between nucleus (DNA) images and background. All pixels with intensity values above threshold in a given neighborhood are deemed to belong to a particular cell nucleus.
The concepts underlying thresholding are well known. The technique is exemplified in
An appropriate threshold may be calculated by various techniques. In a specific embodiment, the threshold value is chosen as the mode (highest value) of a contrast histogram. In this technique, a contrast is computed for every pixel in the image. The contrast may be the intensity difference between a pixel and its neighbors. Next, for each intensity value (0-255 in an eight byte image), the average contrast is computed. The contrast histogram provides average contrast as a function of intensity. The threshold is chosen as the intensity value having the largest contrast. See “The Image Processing Handbook,” Third Edition, John C. Russ 1999 CRC Press LLC IEEE Press, and “A Survey of Thresholding Techniques,” P. K. Sahoo, S. Soltani and A. K. C. Wong, Computer Vision, Graphics, and Image Processing 41, 233-260 (1988), both of which are incorporated herein by reference for all purposes. In a specific embodiment, edge detection may involve convolving images with the Laplacian of a Guassian filter. The zero-crossings are detected as edge points. The edge points are linked to form closed contours, thereby segmenting the relevant image objects. See The Image Processing Handbook, referenced above. Further details regarding the segmentation of nuclei in accordance with the present invention and associated apparatus and techniques are described in co-pending patent applications Nos. 09/729,754 and ______ [Attorney Client Docket No. CYTOP012] by Vaisberg et al., filed concurrently herewith and titled IMAGE ANALYSIS OF GOLGI COMPLEX, the disclosures of which have been previously incorporated by reference herein.
Cell Shape-Indicative Marker Thresholding
After the tubulin image has been obtained at 206, the image analysis tool invokes a thresholding algorithm to convert the image to a binary representation of the image. See 208. As noted above, the concepts underlying thresholding in general are well known. However, in a preferred embodiment of the present invention, a particular technique for calculating the tubulin (or other cell shape-indicative marker) threshold is used. In this case, the threshold value is chosen to discriminate between tubulin and background. All pixels with intensity values above a particular (threshold) intensity value zero in an image subregion (or the entire image, depending upon the specific algorithm) are deemed to represent tubulin and are set to a non-zero value. All pixels the threshold value are set to zero and are deemed to represent background. Of course, the values assigned the marker (tubulin) and background are relative and may be reversed so that the marker is non-zero and the background zero. A cell shape-indicative marker thresholding algorithm in accordance with the invention is illustrated in the process flow 401 of
The intensity is then analyzed for every pixel in the image. See 402 A histogram of the number of pixels having a given range of intensities over the range of intensities analyzed is computed. The histogram may be visualized as graphically depicted in
ITH═IMAX+c*ISTD
where ITH is the threshold, IMAX is the intensity of the greatest number of pixels (assumed to be the background), c is a constant selected on empirical evidence suggesting its suitability for the purpose, and ISTD is computed as follows: A symmetrical curve 424 of the left part of the histogram 422 up to the vertical line denoting Imax in
Application of Watershed Algorithm
Referring now to
These digital images are now processed in conjunction with each other using a watershed technique in order to achieve segmentation of the cells in the original image. See 210. The concepts underlying watershed algorithms are well known The technique is illustrated by way of a geographic analogy in
The parameters required for application of a watershed algorithm are an image and seeds. According to the present invention, a watershed algorithm is applied to the digital image data contained in the nuclei mask and the digital tubulin (or other cell-shape indicative marker) image in order to elucidate cell boundaries for segmentation of the cells. Referring again to
Appropriate watershed algorithms suitable for use in accordance with the present invention are described in detail in L. Vincent and P. Soille, Watersheds in digital spaces: an efficient algorithm based on immersion simulations, IEEE Transactions on Patter Analysis and Machine Intelligence, 13:583-589, 1991, incorporated by reference herein for all purposes.
Feature Extraction
At some point, an image analysis process must obtain image parameters relevant to a biological condition of interest. Typically, the parameters of interest relate to the size, shape, contour, and/or intensity of the cell images. Examples of some specific parameters for analyzing cell shape include the following:
Image analysis routines for extracting these various parameters and others can be designed using well known principles. See The Image Processing Handbook, referenced above. In addition, various commercially available tools provide suitable extraction routines. Examples of some of these products include the MetaMorph Imaging System, provided by Universal Imaging Corporation, a company with headquarters in West Chester, Pa. and NIH Image, provided by Scion Corporation, a company with headquarters in Frederick, Md.
The present invention also provides for the extraction of some novel features associated with cell shape and thus may be used to correlate cell shape information with particular cell conditions, for example a condition resulting from the treatment of cells with a putative toxic or therapeutic agent. See 212. The particular features of interest are skeletal end points and nodes of object masks.
Skeletonization techniques, and techniques for the computation of end points and nodes, from an object's skeleton are well known. See, for example, J. C. Russ, The Image Processing Handbook, CRC press, 1998, previously incorporated by reference herein. However, these techniques have not been used to extract features from biological cells because standard skeletonization algorithms have been found to produce poor results biological context. In particular, standard algorithms tend to be more sensitive to branching patterns common along cell membranes that is desirable to produce biologically significant results.
The present invention provides techniques for extracting biologically significant end point and node parameters from images of cells. This is achieved by essentially simplifying the cell shape to a point where the end point and node branching pattern obtained from analysis of the cell image by application of known skeletonization techniques is reduced to a basic level that characterizes the cell in a biologically significant manner. Preferably, this requires that the cell shape be simplified to eliminate “noise” without significant loss of cell shape features that convey information characteristic of the cell condition.
In a preferred embodiment, the invention accomplishes this by a process of adjustable polygon cell shape approximation. An algorithm illustrating a preferred embodiment of this process is depicted in the process flow 610 of
The distance from the point dMAX to a predetermined threshold distance value, dTH, is then compared (618). The threshold distance value, dTH, is computed according the following function:
dTH=c*A
where A is the area of the cell and c is a constant selected on empirical evidence suggesting its suitability for the purpose. In a preferred embodiment of the present invention, c is between about 1/50 and 1/100, most preferably, 1/80.
If dMAX is greater than dTH, the line between the endpoints is discarded, point dMAX is used as a new endpoint together with one of the original endpoints, this part of boundary is separated into two new parts and the process for determining dMAX is repeated iteratively for each of the two new parts (620). When dMAX is less than dTH, the line between the endpoints is used as a side of a polygon approximating the cell shape (622). The process is repeated until no dMAX is greater than dTH and all the dMAX points selected in this iterative approach and their link lines complete a polygon approximating the shape of the cell (624, 626).
Upon completion of the polygon approximating the shape of the cell, known skeletonization techniques may be applied to the polygon to generate its skeleton and extract biologically significant end point and node features for the particular cell shape (628).
Software/Hardware
Generally, embodiments of the present invention employ various processes involving data stored in or transferred through one or more computer systems. Embodiments of the present invention also relate to an apparatus for performing these operations. This apparatus may be specially constructed for the required purposes, or it may be a general-purpose computer selectively activated or reconfigured by a computer program and/or data structure stored in the computer. The processes presented herein are not inherently related to any particular computer or other apparatus. In particular, various general-purpose machines may be used with programs written in accordance with the teachings herein, or it may be more convenient to construct a more specialized apparatus to perform the required method steps. A particular structure for a variety of these machines will appear from the description given below.
In addition, embodiments of the present invention relate to computer readable media or computer program products that include program instructions and/or data (including data structures) for performing various computer-implemented operations. Examples of computer-readable media include, but are not limited to, magnetic media such as hard disks, floppy disks, and magnetic tape; optical media such as CD-ROM disks; magneto-optical media; semiconductor memory devices, and hardware devices that are specially configured to store and perform program instructions, such as read-only memory devices (ROM) and random access memory (RAM). The data and program instructions of this invention may also be embodied on a carrier wave or other transport medium. Examples of program instructions include both machine code, such as produced by a compiler, and files containing higher level code that may be executed by the computer using an interpreter.
CPU 702 is also coupled to an interface 710 that connects to one or more input/output devices such as such as video monitors, track balls, mice, keyboards, microphones, touch-sensitive displays, transducer card readers, magnetic or paper tape readers, tablets, styluses, voice or handwriting recognizers, or other well-known input devices such as, of course, other computers. Finally, CPU 702 optionally may be coupled to an external device such as a database or a computer or telecommunications network using an external connection as shown generally at 712. With such a connection, it is contemplated that the CPU might receive information from the network, or might output information to the network in the course of performing the method steps described herein.
In one embodiment, the computer system 700 is directly coupled to an image acquisition system such as an optical imaging system that captures images of cells. Digital images from the image generating system are provided via interface 712 for image analysis by system 700. Alternatively, the images processed by system 700 are provided from an image storage source such as a database or other repository of cell images. Again, the images are provided via interface 712. Once in the image analysis apparatus 700, a memory device such as primary storage 706 or mass storage 708 buffers or stores, at least temporarily, digital images of the cell. Typically, the cell images will show locations where a cell component marker of interest (e.g., DNA, tubulin, etc.) exists within the cells. In these images, local values of a marker image parameter (e.g., radiation intensity) correspond to amounts of the marker at the locations within the cell shown on the image. With this data, the image analysis apparatus 700 can perform various image analysis operations such as distinguishing between individual cells in an image of multiple cells and deciphering cell shape. To this end, the processor may perform various operations on the stored digital image. For example, it may obtain and process multiple images of the same original image using to different channels (e.g., excitation/detection wavelengths) to analyze the image in manner that extracts values of one or more cell shape-indicative parameters that correspond to a cellular condition.
The following example provides the results of an experiment showing the effectiveness of techniques in accordance with the present invention for cell segmentation and the extraction of cell features evidencing differences in cell condition. It should be understood the following is representative only, and that the invention is not limited by the detail set forth in this example.
Two fields of HUVEC cells were stained for nuclei with DAPI and for tubulin with rodamine-labeled Dm1alpha antibody, as described above. One of the fields was treated with the anti-actin drug Cytochalasin D in a solution in DMSO. The second field was untreated with any drug, and had only DMSO applied to act as a control. Each field was imaged for nuclei and tubulin and the image data obtained was processed using the thresholding and watershed techniques described above to segment the HUVEC cells in each field. End point features were then extracted from the segmented HUVEC cells in each field using the processes of adjustable polygon cell shape approximation, skeletonization and feature extraction described above.
The results are depicted in
Although the foregoing invention has been described in some detail for purposes of clarity of understanding, those skilled in the art will appreciate that various adaptations and modifications of the just-described preferred embodiments can be configured without departing from the scope and spirit of the invention. For example, a variety of reference components and cell shape-indicative markers other than nuclei and tubulin may be used, and the cell segmentation and feature extraction techniques described may be used alone or in combination. Therefore, the described embodiments should be taken as illustrative and not restrictive, and the invention should not be limited to the details given herein but should be defined by the following claims and their full scope of equivalents.
This application is related to the following co-pending U.S. patent applications, U.S. patent application Ser. No. 09/310,879 by Vaisberg et al., filed May 14, 1999 and titled DATABASE METHOD FOR PREDICTIVE CELLULAR BIOINFORMATICS; U.S. patent application Ser. No. 09/311,996 by Vaisberg et al., filed May 14, 1999 and titled DATABASE SYSTEM INCLUDING COMPUTER FOR PREDICTIVE CELLULAR BIOINFORMATICS; U.S. patent application Ser. No. 09/311,890 by Vaisberg et al., filed May 14, 1999 and titled DATABASE SYSTEM FOR PREDICTIVE CELLULAR BIOINFORMATICS; U.S. patent application Ser. No. 09/729,754 by Vaisberg et al., filed Dec. 4, 2000 and titled CLASSIFYING CELLS BASED ON INFORMATION CONTAINED IN CELL IMAGES and; U.S. patent application Ser. No. ______ [Attorney Client Docket No. CYTOP012] by Vaisberg et al., filed concurrently herewith and titled IMAGE ANALYSIS OF THE GOLGI COMPLEX; each of which is incorporated herein by reference for all purposes.
| Number | Date | Country | |
|---|---|---|---|
| Parent | 09792013 | Feb 2001 | US |
| Child | 11165914 | Jun 2005 | US |
