EP1204950A1

EP1204950A1 - Image processing methods, programs and systems

Info

Publication number: EP1204950A1
Application number: EP00949811A
Authority: EP
Inventors: Geoffrey SmithKline Beecham Pharm. HEALD
Original assignee: SmithKline Beecham Ltd
Current assignee: SmithKline Beecham Ltd
Priority date: 1999-08-09
Filing date: 2000-08-08
Publication date: 2002-05-15
Also published as: GB9918681D0; WO2001011564A1; AU6307400A

Abstract

An aspect of the invention provides a method of detecting discrete features in an original image. The method includes various steps. Pixels in an original image array having an intensity with a predetermined relationship to a selected intensity threshold level are identified. The identified pixels are positioned in corresponding positions in an intermediate data array to form an intensity plane. The intermediate data array is analyzed using at least one test to determine whether any area in the intensity plane is consistent with a characteristic of the feature to be detected. Where an area in the intensity plane is accepted as being consistent with the characteristic of the feature to be detected, the pixels for that area are copied from the original image to a storage image array and a position indication of the accepted area is recorded. These steps are then repeated at least once using a lower or higher selected intensity threshold level. A stepwise segmentation approach thus involves simplification of an original image and a series of basic characteristic (for example shape) tests. An example of the application of the invention is for image analysis in proteomics.

Description

IMAGE PROCESSING METHODS, PROGRAMS AND SYSTEMS

BACKGROUND OF THE INVENTION

The present invention relates to methods, computer programs and systems for image processing. More particularly, the invention relates to a method for the segmentation (detection and isolation) of discrete features, for example areas of local high density, in 2-dimensional images.

An application of image processing is to provide qualitative and quantitative comparison of density data maps in 2-dimensional proteomic gels. A recent increase in demand for a proteomic analysis approach to studying biological systems has highlighted the need for rapid and accurate analysis of electrophoretic gels. In proteomic analysis complex pools of expressed proteins from biological systems are separated and displayed by techniques such as 2-D gel electrophoresis. In order to define changes in protein expression under differing states computerized analysis of 2-D gels is required for speed, sensitivity, reproducibility, quantisation and accuracy.

Ideally images from 2-D gels should consist of protein spots (approximately elliptical areas of high density) on a uniform ground of low density. However, images derived from 2-D gels are subject to complication by a variety of artifacts, such that in practice background levels vary considerably over each image, spots are smeared and often merge with adjacent spots. The image may have physical fractures and contaminants, all of which have densities similar to the spots to be detected.

Current commercial image processing products are available for image analysis but perform inadequately in some or all of the following respects: a) Spot detection algorithms are mostly based upon edge detection which circumscribes the transition boundary between areas of high-density values and lower background values. As the boundaries of artifacts may have similar characteristics to the protein spots, these have an equal potential to be detected resulting in a high incidence of errors which require elimination by a laborious secondary editing process. b) They are ineffective at dealing with merged spots, thus requiring significant operator intervention. c) The average time for spot detection including manual intervention currently varies between about 1-4 h / gel. d) Manual intervention leads to issues of operator subjectivity, resulting in increased variability both between operators and by the same operator over extended periods of analysis.

Thus there is a need to provide an improved imaging product for enabling users of gel image analysis to perform rapid, reliable and automated spot detection (image segmentation) and measurement.

SUMMARY OF THE INVENTION

Particular and preferred aspects of the invention are set out in the accompanying independent and dependent claims. Combinations of features from the dependent claims may be combined with features of the independent claims as appropriate and not merely as explicitly set out in the claims.

An aspect of the invention provides a method of detecting discrete features in an original image, the method comprising: a) identifying pixels in an original image array having an intensity with a predetermined relationship to a selected intensity threshold level; b) positioning the identified pixels in corresponding positions in an intermediate data array to form an intensity plane; c) analyzing the intermediate data array using at least one test to determine whether any area in the intensity plane is consistent with at least one characteristic of the feature to be detected; d) where an area in the intensity plane is accepted as being consistent with said at least one characteristic of the feature to be detected, copying the pixels for that area from the original image to a storage image array and recording a position indication of the accepted area; and e) repeating steps a)-d) at least once using a lower or higher selected intensity threshold level.

The method can further include, in step (d): for each intensity threshold level, checking each newly accepted area for the presence of an already accepted area; and for newly accepted areas for which an already accepted area is identified, replacing the position indication of the already accepted area with a position indication of the newly accepted area.

Another aspect of the invention provides a computer program element operable to detect discrete features in an original image, the computer program element comprising computer code operable to carry out the steps of the above described method.

A further aspect of the invention provides a computer system or image processing system including storage and elements for carrying out the steps of the method.

An embodiment of the invention thus provides a stepwise segmentation approach involving simplification of an original image and a series of basic characteristic (for example shape) tests. The original image can be a two-dimensional array of any size, but could typically consist of about 15 million spatial point values (pixels) with 4096 or more intensity values. When the data is viewed as an image all the intensity values are presented together. In an embodiment of the present invention fractions of the image, i.e. only pixels at or above, or at or below, a fixed threshold intensity level are analyzed in isolation.

The method of the invention is particularly suited to image analysis in proteomics. In proteomics complex pools of expressed proteins from biological systems are separated and displayed by techniques such as 2-D gel electrophoresis. In order to define changes in protein expression under differing states, computerized analysis of 2-D gels is required for speed, sensitivity, reproducibility, quantitation and accuracy. The gels analyzed using the method of the invention will generally be 2-D electrophoretic polyacrylamide matrixes which are used to separate and display the protein complement of a biological fluid, cell, tissue, organ etc. Prior to image analysis a digitized representation of the gel is generally created by use of scanning, e.g. laser, densitometry, CCD camera etc, and the data stored as image files, e.g. tif. The data generated by the process of the invention is preferably provided in a format that can be displayed on screen and is exportable for further statistical analysis, e.g. comparison to highlight, for example, statistically significant changes in protein expression across gels.

BRIEF INTRODUCTION TO THE DRAWINGS

Exemplary embodiments of the present invention will be described hereinafter, by way of example only, with reference to the accompanying drawings in which like reference signs relate to like elements and in which:

Figure 1 is a schematic representation of a computer workstation for an exemplary implementation of the invention;

Figure 2 is schematic block diagram illustrating an exemplary configuration of a computer workstation as shown in Figure 1 ; Figure 3 is schematic representation of a distributed processing system; Figure 4 is a schematic block diagram of the content of the memory of the workstation of Figure 1 configured for an embodiment of the invention;

Figure 5 is a flow diagram illustrating an example of operation of an embodiment of the invention; Figure 6 is a simple representation of image density cross section, horizontal lines indicating threshold levels; and

Figures 7 - 13 show threshold images (top) and corresponding images (bottom) of the accumulation of areas identified and accepted during an example of operation of an embodiment of the invention when applied to a 2-D proteomic gel.

DESCRIPTION OF PARTICULAR EMBODIMENTS

Exemplary embodiments of the present invention are described in the following with reference to the accompanying drawings.

Figure 1 is a schematic representation of a computer workstation (for example a personal computer, or PC) on which an exemplary embodiment of the invention can be implemented. As shown in Figure 1 , a computer workstation 10 includes a system unit 12, user input devices, for example in the form of a keyboard 14 and a mouse 16, and a display 14. Removable media devices in the form, for example, of a floppy disk drive 20 and an optical and/or magneto-optical drive (e.g. a CD, a DVD ROM, a CDR drive) 20 can also be provided.

Figure 2 is schematic block diagram illustrating an exemplary configuration of a computer workstation 10 as shown in Figure 1.

As shown in Figure 2, the computer workstation 10 includes a bus 30 to which a number of units are connected. A microprocessor (CPU) 32 is connected to the bus 30. Main memory 34 for holding computer programs and data is also connected to the bus 30 and is accessible to the processor. A display adapter 36 connects the display 18 to the bus 30. A communications interface 38, for example a network interface and/or a telephonic interface such as a modem, ISDN or optical interface, enables the computer workstation 10 to be connected 40 to other computers via, for example, an intranet or the Internet. An input device interface 42 connects one or more input devices, for example the keyboard 14 and the mouse 16, to the bus 30. A floppy drive interface 44 provides access to the floppy disk drive 20. An optical drive interface 46 provides access to the optical or magneto-optical drive 46. A storage interface 48 enables access to a hard disk 50. Further interfaces, not shown, for example for connection of a printer (not shown), may also be provided. Indeed, it will be appreciated that one or more of the components illustrated in Figure 2 may be omitted and/or additional components may be provided, as required for a particular implementation.

Although Figures 1 and 2 have illustrated a single computer on which an application of the present invention may be operable, an embodiment of the invention could be operated over a distributed processing system, for example a distributed processing system such as that illustrated schematically in Figure 3.

Figure 3 shows a network (for example, the Internet) 52 which provides connections between various stations 54, each of those stations communicating with each other via the network 52 by means of one or more protocols 56. The individual stations 54 may be stand-alone computers or, as illustrated at the top left of Figure 3, may in fact represent a network of computers including a gateway 55 that provides a connection via an internal network 58 to a plurality of computers 59 or other devices.

Figure 4 is a schematic block diagram of the content of the memory 34 of the workstation of Figure 1 configured for an embodiment of the invention.

Figure 4 only illustrates selected elements in the memory useful for an understanding of the invention. It will be appreciated that the memory of the workstation will typically include other elements. Figure 4 illustrates an operating system element 60, including various device drivers, that controls the basic operation of the hardware of the workstation and provides a platform on which computer applications may run. A program element 62 provides program code forming an application for implementing an embodiment of the invention. The program code may comprise on or more modules and may be implemented using any one of many programming languages, of which examples might be C, C++, Java, etc. In use, the program code is operable to define a plurality of data arrays for storing various images generated during the operation of an embodiment of the invention. The data arrays will be described in more detail later. They include an original image array 64, a working image array 66, one or more intermediate image arrays 68, an accumulating array 74 and a storage image array 76. The method of operation of an embodiment of the invention to be described later operates in a plurality of passes for different image threshold levels. A single intermediate image may be reused for each pass, or a separate intermediate image may be generated in each pass. The aim of the process is to extract features from the original image. One such feature 70 is illustrated in the intermediate image 68, and the centroid (i.e. the center of gravity) 72 of the feature is also illustrated. The accumulating array 74 is used to accumulate the centroid co-ordinates for features as they are found. The storage image array is used to store an output image of the features found by the process.

Figure 5 is a flow diagram illustrating the control flow for an exemplary embodiment of the invention. In describing Figure 5, reference will be made to Figure 4.

This embodiment of the invention provides a method for the detection of discrete features in a 2-dimensional image. This embodiment provides a stepwise segmentation approach involving simplification of an original image by sequentially selecting intermediate images with ranges of pixel intensities by a sequential thresholding approach. A range of intensity levels could, for example, be pixel image intensities equal to and/or greater than the threshold. As an alternative, a range of intensity levels could be, for example, pixel image intensities equal to and/or less than a threshold. In the former case, successive intermediate images could be identified using successively decreasing thresholds. In the latter case, successive intermediate images could be identified using successively increasing thresholds.

In step SI an original image is input to the workstation and is stored in an original image data array 64 in the computer memory. Optionally, a working image copy of the original image is stored in an original working image data array 66 in the computer memory.

In this embodiment of the invention, a working image copy of the original image is preferably made, although this is not essential. If a working copy is made, this can be subjected to a smoothing algorithm to minimize local variations and additionally to a histogram equalization algorithm to optimize the intensity range using suitable algorithms as will be known to those skilled in the art.

In step S2, a level range for intensity levels is identified for a first working image. A level range is defined with respect to a threshold intensity.

In step S3, pixels within the working image are identified within an intensity level range that is at and above a selected intensity threshold level within the image. (In the present example, it is assumed that the initial threshold is a high threshold and that successive thresholds are lower. In another example, if an initially low threshold were to be taken, with successive thresholds being higher, pixels would be identified within an intensity level range that is at or below a selected intensity threshold level). Pixels that are identified as within the intensity level range are then stored in corresponding positions in an intermediate data array 68.

Thus, in the present example pixels either at and above (alternatively, in another example at and below) the threshold intensity level in the original image are extracted and made to occupy their respective positions in a same size intermediate data array 68 otherwise uniformly set at zero. The resulting areas of contiguous pixels consist of shapes which may, or may not, be part of a feature 70 to be detected.

In step S4, an analysis of the intermediate data array is effected using at least one test to determine whether any of the areas in the intensity plane are consistent with the characteristics of the feature to be detected. The analysis is effected on groups of pixels effectively masked by the thresholding process.

The analysis of the much simplified intermediate image in the intermediate data array 68 comprises at least one test, and preferably a short sequence of tests, to determine whether any of the areas in this intensity plane are consistent with the characteristics of the feature 70 to be detected. Suitable characteristics include shape criteria. Examples of suitable shape criteria for area acceptance include: area size limit, aspect ratio limit band and pixel spread band; suitable methods for applying these criteria will be known to those skilled in the art.

In the example shown in Figure 4, a first test S4A is whether a potential feature is within a given size range, and a second step S4B is whether a further parameter (for example a shape criterion) is within predetermined limits. If the potential feature does not meet the test in steps S4A and S4B, then control passes to step S6, to be described later.

For an example of the invention to be applied to the analysis of protein spots, the characteristics of those protein spots can be taken into account. Protein spots in proteomic gels are normally elliptical tending to circular but can become elongate in one direction orthogonal to the image margin, and hence suitable shape criteria can be chosen accordingly. Shapes consistent with the limits set are accepted and the equivalent pixels in the original image are copied and saved to a final storage image array 76 as will be described later. If the potential feature meets the tests in steps S4A and S4B, then a position 72 for the potential feature 70 (e.g., a protein spot) is computed. In this example, the position is computed as the coordinates of the centroid 72 (that is the center of gravity) for the potential feature. A comparison is made for each pixel of the potential feature to determine whether centroid coordinates have been stored in an accumulating array 74 for a previous intensity range, if any. If so, then the new centroid coordinates replace those previously stored. Otherwise the new centroid coordinates are simply stored in the appropriate location in the accumulating array 74.

In one example of the invention, step S4C includes checking the threshold intensity of the pixels of each accepted area for the presence of a centroid in the accumulating array 74 of centroids, areas for which at least one already stored centroid is found are accepted and its old centroid coordinates replaced in the accumulating array with the centroid coordinates calculated from the new area.

In other words, pixels of each area accepted in step S4B at each threshold are checked for the presence (match) of a centroid in the accumulating array of centroids in step S4C.

The presence of one already stored centroid in an accepted area indicates an enlarged feature which will be accepted and its old centroid coordinates replaced in the array with the new centroid coordinates calculated from the new area.

The presence of two or more centroids indicates that the accepted area includes two or more already stored areas and therefore that they merge at this threshold level. Merger can be prevented by ignoring shapes containing more than one centroid. However, this alone would mean that different criteria for limiting the spot area is applied to merging areas as opposed to that for discrete feature areas, therefore each potentially merging area pair (or more) is preferably flagged for the application of alternative algorithms to resolve merge areas after the initial pass through all the selected thresholds.

If this process is successful, then control passes to step S5. Control passes otherwise to step S6.

If control passes to step S5, this means that the potential feature has been accepted as being consistent with predetermined characteristics of a predetermined feature and the equivalent pixels of the original image are copied to a storage image array 76.

In step S6, a test is made as to whether there are any further potential features. If so, control passes back to step S4. Otherwise, control passes to step S7.

In step S7, a check is made as to whether that was the last intensity plane (i.e. range of intensity levels) to be investigated. If not, then control passes back to step S3 for processing a further intensity plane, which as described above could involve generating a new intermediate data array 68 or could involve overwriting the previous intermediate image data array 68. On the other hand, if that was the last intensity plane, then control passes to step S8.

As mentioned above, an embodiment of the invention may use intensity threshold levels of sequentially lower or sequentially higher intensities, in which case pixels are identified with intensities at and above, and at and below the selected intensity threshold level, respectively. The method preferably uses sequentially lower threshold intensities. Preferably also, all, or substantially all, intensity levels of the image are used as a threshold and examined in turn from the highest to the lowest, as analysis at every level allows extraction of the maximum amount of data. Alternatively steps through the intensity range can be made, either at regular intervals or arbitrarily, e.g. 4096, 4086, 4076 etc. Thus, this process can be applied to at least two threshold intensity levels in the working image. Every area at the intensity threshold is analyzed which is accepted as being consistent with the characteristics of the feature to be detected and has its equivalent area in the original image passed to the storage image. Progressing through the levels means that areas of contiguous pixels may enlarge through several thresholds. The areas found are allowed to overwrite the earlier areas stored (which they include anyway) and increase until the area no longer satisfies the characteristics test(s), e.g. shape criteria limits.

In step S8, the features saved in step S5 can then be analyzed or measured, as required.

After completion of the method of the invention the result is a much simplified image consisting of a background set at zero upon which are the feature areas satisfying the predetermined criteria and positioned at their original coordinates. The completed image is preferably reanalyzed to (a) sum the pixels to give an area figure; and/or (b) sum the intensity values of each feature area to give a volume figure.

In Figure 5, reference is made to the features being spots and to the analysis in step S8 being the measurement of the saved areas, or spots. However, in other examples, other features and tests could be performed, as appropriate.

For gel analysis additional algorithms may be applied either before or after the segmentation process to determine local background values that will be subtracted from the area and volume figures. The resultant image is then in a suitable form for the registration and comparison processes in the detection and identification of differences between gels. The data is also available for export and further analysis by other software packages, e.g. multivariate statistical analysis. In a specific embodiment of the invention, software comprised in a computer program product accepts digitized images in a format generated from various imaging systems. The software preferably accepts a range of common gel image formats used by imaging equipment suppliers e.g. ".gel" files (Molecular Dynamics); ".lsc" files (BioRad); "..img" files (Fuji); ".im" files (MCID) as well as other common image file formats e.g. ".tif files. The software is preferably capable of handling different gel formats, sizes, orientations etc., for example it should preferably accept 8, 12, 16 bit images ranging in size from about <5 Mbytes to > 20Mbytes.

The software is able to perform accurate local background subtraction based on automated background identification and copes with variable background levels across the gel. The software preferably has the capacity to access spot characteristics (x.y values, area, volume etc.) at the operator's request, e.g. by right mouse click on the spot.

The software preferably displays the same size image of all the protein spots with their original pixel values at their original locations on a zero ground ideally with the elimination of all artifacts. Preferably the software also allows the adjustment of gray scale to aid visualization of spots and to note the variable gray settings so that it can keep the same across images for standardization purposes, if desired. The software also preferably gives a choice of visualizing detected features, e.g. as crosshairs, ellipses.

The software preferably supplies the following, selectable information in tabulated, printable and exportable format, from the spot detection:

Number of spots

X,Y coordinates of spots by centroid data

Spot alias (e.g. number)

Area of spot Volume of spot Aspect ratio Spread Factor

The software preferably displays the image with manually adjustable tonal range to optimize the display to the eye. It preferably calculates the maximum and minimum pixel values within spots and displays accordingly.

The software is preferably capable of operating in a batch mode.

Optionally, spots can be flagged which saturate i.e. data that lies above the range of the image above the maximum (i.e. at 4,096 for a 12-bit image and 65,536 for a 16-bit image).

An embodiment of the invention has advantages over known methods of image analysis since it is capable of isolating and quantifying features, i.e. the discrete local areas of high(er) density satisfying shape criteria consistent with the presence of protein in the gel. It is able to detect all features in a reproducible fashion without the need for parameter settings (which will select a cut off for features detected) i.e. it can be automated and hence eliminate operator subjectivity. However, it may be preferable to retain some operator determinable parameters in the method to aid optimization.

An embodiment of the invention can detect a large dynamic range of feature intensities and areas, and is able to detect and resolve merged features. The processing of spot detection is also fast, typically taking about 1 minute per image, e.g. per gel.

Features that exhibit "negative" staining due to abundance can be defined and noise and artifacts in the image due to dust, edge effects and gel fractures can be eliminated without defining them as background for the purposes of quantitation. Figure 6 is a simple representation of image density cross section, horizontal lines indicating threshold levels.

Figures 7-13 shows threshold images (top) and corresponding images (bottom) of the accumulation of areas identified and accepted during an example of the operation of an embodiment of the invention when applied to a 2-D proteomic gel.

Figure 7 shows spot detection planes at the 1st and 2nd thresholds of highest density levels. Spots detected are highlighted on the original image.

Figure 8 shows spot detection planes at the 3rd and 4th thresholds of highest density levels.

Figure 9 shows spot detection planes at the 5th and 6th thresholds of highest density levels.

Figure 10 shows spot detection planes at the 7th and 8th thresholds of highest density levels.

Figure 11 shows lower levels passing down through the density range.

Figure 12 shows how image objects merge as the threshold is lowered further while detectability of potential spot areas at the lower levels remains the same.

Figure 13 shows spot areas extracted from the original image array and transferred to a "zero ground" array. A preferred embodiment of a method according to the invention is implemented using computer readable instructions, which may be in the form of software, wherever and by whatever means such software is stored and/or accessed.

Thus a preferred embodiment of the invention provides a computer program product directly loadable into the internal memory of a digital computer such as the work station of Figure 1 and comprising software coded portions for performing the method as described above when said product is run on the computer. The computer program product is preferably stored on a computer usable medium.

It will be apparent to those skilled in the art that programs defining the method of the present invention can be delivered to a computer in many forms, including but not limited to: information permanently stored on non-writable storage media, e.g. read only memory devices within a computer such as ROM or CD-ROM disks readable by a computer I/O attachment; information alterably stored on writable storage media, e.g. floppy discs and hard drives; and information conveyed to a computer through communication media such as networks and telephone networks via modem.

Thus, a computer program product, or element, for implementing the invention could comprise program code on a carrier medium. The carrier medium could be a storage medium, such as solid state magnetic optical, magneto-optical or indeed any other storage medium. The carrier medium could be a transmission medium such as broadcast, telephonic, computer network, wired, wireless, electrical, electromagnetic, optical or indeed any other transmission medium.

An embodiment of the invention may be implemented using any general-purpose computer or personal computer. An embodiment of the invention may be implemented by program code in the form of one or more software modules for controlling a general-purpose processor or group of processors. Any appropriate programming language may be used. However, other structures and implementations could also be employed. For example, an embodiment of the invention could be embodied in a special purpose integrated circuit such as an application specific integrated circuit (ASIC).

There has been described a method for the detection of discrete features in a two- dimensional image which comprises the steps of: (i) identifying pixels either at and above, or at and below, a selected intensity threshold level within the image;

(ii) positioning the identified pixels into the corresponding positions in an intermediate data array;

(iii) analysis of the intermediate data array using at least one test to determine whether any of the areas in the intensity plane are consistent with the characteristics of the feature to be detected;

(iv) acceptance of areas in the intensity plane which are consistent with characteristics of the feature to be detected and copying of the equivalent pixels in the original image to a storage image array; (v) calculation of the centroid coordinates of each accepted area;

(vi) storage of the centroid coordinates in an accumulating array;

(vii) repeating steps (i)-(vi) at least once using a lower or higher selected intensity threshold level;

(viii) for each threshold intensity the pixels of each accepted area are checked for the presence of a centroid in the accumulating array of centroids; and

(ix) areas for which at least one already stored centroid is found are accepted and its old centroid coordinates replaced in the accumulating array with the centroid coordinates calculated from the new area. Although particular embodiments of the invention have been described, it will be appreciated that many modifications/additions and/or substitutions may be made within the scope of the invention.

Claims

1. A method of detecting discrete features in an original image, the method comprising:

a) identifying pixels in an original image array having an intensity with a predetermined relationship to a selected intensity threshold level; b) positioning the identified pixels in corresponding positions in an intermediate data array to form an intensity plane; c) analyzing the intermediate data array using at least one test to determine whether any area in the intensity plane is consistent with at least one characteristic of a feature to be detected; d) where an area in the intensity plane is accepted as being consistent with said at least one characteristic of the feature to be detected, copying the pixels for that area from the original image to a storage image array and recording a position indication of the accepted area; e) repeating steps a)-d) at least once using a lower or higher selected intensity threshold level.

2. The method of claim 1, wherein recording a position indication for an accepted area comprises calculating centroid coordinates of the accepted area and storage of the centroid coordinates in an accumulating array.

3. The method of claim 1 or 2, comprising, in step (d):

for each intensity threshold level, checking each newly accepted area for the presence of an already accepted area; and for each newly accepted area for which an already accepted area is identified, replacing the position indication of the already accepted area with a position indication of the newly accepted area.

4. The method of claim 3, wherein:

checking each newly accepted area for the presence of an already accepted area comprises checking the pixels of each accepted area for the presence of a centroid in the accumulating array of centroids; and for newly accepted areas for which the presence of a centroid is identified, the replacing of the position indication of the already accepted area with a position indication of the newly accepted area comprises replacing the old centroid coordinates in the accumulating array with the centroid coordinates calculated from the new area.

5. The method of any preceding claim, wherein all, or substantially all, intensity levels of the image are used as a threshold and examined in turn from the highest to the lowest.

6. The method of any preceding claim, wherein prior to step a) a working copy of the original image is made which is subjected to at least one of a smoothing algorithm to minimize local variations and a histogram equalization algorithm to optimize the intensity range.

7. The method of any preceding claim, wherein the test(s) to determine whether any of the areas in an intensity plane are consistent with at least one characteristic of the feature to be detected is based on at least one shape criterion.

8. The method of claim 6, wherein at least one shape criterion is selected from area size limit, aspect ratio limit band and pixel spread band.

9. The method according to any preceding claim, wherein a newly accepted area for which more than one position indication of an already accepted area is found is flagged for the application of alternative algorithms to resolve merged areas.

10. The method of any preceding claim additionally comprising a step of reanalysis of the completed image to perform at least one of a sum of the pixels to give an area figure and a sum of the intensity values of each spot area to give a volume figure.

1 1. The method of any preceding claim, wherein the original image is a two- dimensional image of a proteomic gel.

12. A method for the detection of discrete features in a two-dimensional image which comprises the steps of:

i) identifying pixels either at and above, or at and below, a selected intensity threshold level within the image; ii) positioning the identified pixels into the corresponding positions in an intermediate data array; iii) analysis of the intermediate data array using at least one test to determine whether any of the areas in the intensity plane are consistent with the characteristics of the feature to be detected; iv) acceptance of areas in the intensity plane which are consistent with characteristics of the feature to be detected and copying of the equivalent pixels in the original image to a storage image array; v) calculation of the centroid coordinates of each accepted area; vi) storage of the centroid coordinates in an accumulating array; vii) repeating steps (i)-(vi) at least once using a lower or higher selected intensity threshold level; viii) for each threshold intensity the pixels of each accepted area checking for the presence of a centroid in the accumulating array of centroids; and ix) areas for which at least one already stored centroid is found being accepted and its old centroid coordinates being replaced in the accumulating array with the centroid coordinates calculated from the new area.

13. A computer program element operable to detect discrete features in an original image, the computer program element comprising computer code operable:

a) to identify pixels in an original image array having an intensity with a predetermined relationship to a selected intensity threshold level; b) to position the identified pixels in corresponding positions in an intermediate data array to form an intensity plane; c) to analyze the intermediate data array using at least one test to determine whether any area in the intensity plane is consistent with at least one characteristic of the feature to be detected; d) where an area in the intensity plane is accepted as being consistent with said at least one characteristic of the feature to be detected, to copy the pixels for that area from the original image to a storage image array and to record a position of the accepted area; and e) to repeat steps a)-d) at least once using a lower or higher selected intensity threshold level.

14. A computer program element operable to detect discrete features in an original image, the computer program element comprising computer code operable to perform all the steps of a method according to any one of claims 1 to 12.

15. The computer program element of claim 13 or claim 14 on a carrier medium.

16. A computer system comprising a processor and storage and the computer program element of any of claims 13 to 15.

17. A computer system comprising storage and processing elements, the processing elements being operable:

a) to identify pixels in an original image array having an intensity with a predetermined relationship to a selected intensity threshold level; b) to position the identified pixels in corresponding positions in an intermediate data array to form an intensity plane; c) to analyze the intermediate data array using at least one test to determine whether any area in the intensity plane is consistent with at least one characteristic of the feature to be detected; d) where an area in the intensity plane is accepted as being consistent with said at least one characteristic of the feature to be detected, to copy the pixels for that area from the original image to a storage image array and to record a position of the accepted area; e) to repeat steps a)-d) at least once using a lower or higher selected intensity threshold level.

18. The computer system of claim 17, wherein the processing elements are further operable:

for each intensity threshold level, to check each newly accepted area for the presence of an already accepted area; and for each newly accepted area for which an already accepted area is identified, to replace the position indication of the already accepted area with a position indication of the newly accepted area.