SE518457C2 - Segmenting method for objects in a first category in a series of images with at least one digital microscope color image that depicts biological material for analyzing peripheral blood and bone marrow - Google Patents

Abstract

The method involves identifying a first set of pixels in a color image in a series of images using a first rough measure. A first function is identified which gives an approximate description of the relationship between the respective intensities of the two color components for pixels in the first set. A second set of pixels is identified in a color image in the series of images using a second rough measure. A second function is identified. It is determined whether a given pixel in a color image in the series of images depicts an object in the first category or not, on the basis of the given pixel's respective intensities of the two color components and the first and second functions. Independent claims are included for a computer program for segmenting objects and for a digital storage medium.

Description

annu: rvnz »lO l5 20 25 30 35 518 4 s? '* 2 procedure works well. By a well-performed segmentation of a given blood cell is meant that all picture elements which are present in an image and which depict the given blood cell are identified, but no other picture elements. There can be several blood cells in an image and contiguous groups of identified pixels are handled as a segmented object. It is the identified pixels that are passed on for classification of the blood cell. If the segmentation becomes incorrect, ie if pixels that do not image the given blood cell are forwarded, or when pixels that image the given blood cell are not forwarded, the appearance of the cell is distorted, which can give an incorrect classification. Too many incorrect classifications lead to the system becoming less reliable.

A well-known method for performing segmentation is so-called thresholding. When thresholding, properties, for example grayscale intensity, of a given picture element are compared with a threshold value, in order to determine whether the picture element has such properties as are expected of the picture elements which depict the object sought. If a dark type of searched object is to be distinguished in a black and white image, for example the grayscale intensity of the given pixel can be compared with a threshold value adapted to the searched objects. If the grayscale intensity of the given pixel is darker than the desired threshold value, this pixel is judged to image a searched object, otherwise not.

When color images constitute input data for the segmentation procedure, the color information in the images can be used. A given pixel in an image then usually has an intensity component for each of the primary colors red (R), green (G) and blue (B). form a color vector. From this color vector, a so These components can be said to be called RGB projection can be calculated, such as the scalar product of the color vector and a predetermined weight vector.

As mentioned, the color information in an image can be used to segment an object of a certain type from a digital color image. Then a .nnaø:;. | » 10 15 20 25 30 35 518 457 '3 weight vector and a threshold value adapted for this type of object is used. The RGB projection of a given pixel, calculated by the weight vector, is compared with the threshold care to determine whether the pixel depicts an object of the type sought or not.

The methods shown above for performing segmentation, when analyzing a blood smear, work relatively well to correctly distinguish a white blood cell or a red blood cell against the background of an image. However, if a red blood cell and a white blood cell lie next to each other or partially obscure each other, it can be difficult to correctly segment the white blood cell. This is because the difference in color composition and intensity between red blood cells and the cytoplasm of some white blood cells is relatively small. The contrast in the transition between red and white blood cells can thus be weak. Of course, it becomes similarly difficult to segment a red blood cell, but as mentioned above, it is primarily the white blood cells that are of interest.

It is usually possible to find a working segmentation condition, consisting of a weight vector for an RGB projection and a threshold value, with which condition a red and a white blood cell can be distinguished. However, such a condition is not very robust, but only works under almost constant conditions. In reality, the smear thickness will vary both within a test tube and between samples. In addition, the intensity and hue of the dyeing may vary. Furthermore, different laboratories use many different variants of staining. A fixed segmentation condition will therefore give incorrect results in a number of cases.

SUMMARY OF THE INVENTION An object of the present invention is to completely or partially remedy the above-mentioned problems.

This object is achieved with a method according to claim 1, an arrangement according to claim 12, a computer program according to claim 10 and a digital storage medium containing according to the computer program according to claim 14.

More particularly, according to a first aspect of the invention, there is provided a method of distinguishing objects from a first category in an image series with at least one digital microscope color image, which depicts biological material and comprises a plurality of pixels having at least two color components, wherein also objects from a second category can be found in the image series. The method is characterized by the steps: to identify with a first rough measure a first set of pixels in an image in the image series, so that the pixels in the first set predominantly depict objects from the second category; identifying a first function, which gives an approximate description of the relationship between the respective intensities of the two color components for pixels in the first set; roughly identifying a second set of pixels in an image in the image series, so that the pixels in the second set predominantly depict objects from the first category; identifying a second function which provides an approximate description of the relationship between the respective intensities of the two color components for pixels in the second set; and determining whether or not a given pixel in an image in the image series depicts an object from the first category, based on the respective intensities of the given pixel for the two color components and the first and second functions.

In such a way, the discrimination is adapted to the appearance of the sample, so that effective segmentation can take place even if variations in, for example, staining and layer thickness vary greatly within or between images in the image series. The segmentation therefore becomes so reliable that correct classification can be performed.

The first and second functions may preferably be substantially linear, providing a simple procedure. . . c .u I .uu-n - -a-v 11,1- 10 l5 20 25 30 35 51 s 457 šïï šïïí Im? Preferably, the step of determining whether the given pixel depicts an object from the first category or not includes the sub-steps: identifying for the given pixel an element line which in the intensity plane of the two color components intersects both the intensities of the given pixel for the two colors as an intersection point between the first and the second function; assigning the given pixel an angular value based on the angle of the element line relative to lines corresponding to said first and second functions; and determining whether the given pixel images an object from the first category or not based on the angular value.

The use of such angular values has proven very reliable in segmenting biological material.

According to a preferred embodiment, the angular value of the given pixel assumes a first extreme value if the element line coincides with the line of the first function or has an angle relative to the line of the second function which is greater than the angle between the lines of the first and second functions; a second extreme value if the element line coincides with the line of the second function or has an angle relative to the line of the first function which is greater than the angle between the lines of the first and the second function; and a value between the first and the second extreme value if the element line has smaller angles in relation to the lines of the first and second functions than the mutual angle between the lines of the first and the second function.

According to a preferred embodiment, the distance between the intensities of the given pixel and the intersection of the first and second functions in the plane of the two color components can also be used to determine whether the given pixel depicts an object from the first category or not. This makes segmentation even more secure. wvaz. 10 15 20 25 30 35. uuuø s 1 s 457 2112? Preferably, the objects in the first category are white blood cells, and the objects in the second category are red blood cells. The method of the invention has been found to be particularly suitable for segmenting white blood cells in the presence of red blood cells.

Preferably, the two color components are red and blue. These primary color components are particularly suitable for segmenting white blood cells in the presence of red blood cells, at least when using so-called MGG - (May-Grünwald-Giemsa) ~ staining or so-called Wright staining.

Preferably, the first coarse measure includes by threshold identifying and excluding pixels that image background. This provides a simple and often sufficiently reliable way to roughly identify a set of pixels that depict red blood cells.

The first coarse measure may further comprise identifying and excluding areas of pixels, which areas can be assumed to contain white blood cells, which assumption is based on the intensity of the green color component. This provides a "cleaner" set of pixels that depict red blood cells.

Preferably, the second coarse measure comprises identifying at least one low intensity pixel cluster of green color component. This is a simple way to identify pixels that image nuclei in white blood cells.

According to one embodiment, the second coarse measure may further comprise performing the sub-steps: selecting an area associated with the pixel cluster is selected, which area has a size larger than the expected size of a white blood cell; selecting at least one pixel in the area, which pixel has an appearance different from the expected appearance of pixels depicting white blood cells, said at least one pixel forming an initial amount; that object similarity values are determined for pixels in the area; | n »-. lO l5 20 25 30 35 s 1 s 4s7 ä? 7 and that the second set of picture elements is identified on the basis of the initial set and the object similarity values. This coarse measure gives more pixels from the cytoplasm and thus a better selection.

According to a second aspect of the invention, there is provided an arrangement for distinguishing objects from a first category in a series of images with at least one digital color image, which depicts biological material and comprises a plurality of picture elements, which have at least two color components, wherein also objects from a second category can be found in the image series. The arrangement is characterized by: means for identifying with a first rough measure a first set of pixels in an image in the image series, so that the pixels in the first set predominantly depict objects from the second category; means for identifying a first function, which gives an approximate description of the relationship between the respective color components' intensities of pixels in the first set; means for roughly identifying a second set of pixels in an image in the image series, so that the pixels in the second set predominantly depict objects from the first category; means for identifying a second function, which gives an approximate description of the relationship between the respective intensities of the two color components for pixels in the second set; and means for determining whether or not a given pixel in an image in the image series depicts an object of the first category, based on the respective intensities of the given pixel for the two color components and the first and second functions.

The arrangement entails corresponding advantages as the method and can be varied accordingly.

According to a third aspect of the invention, there is provided a computer program for distinguishing objects from a first category in a series of images with at least one digital color image, which depicts biological material and comprises a plurality of> |. || 10 l5 20 25 30 35 51 s 457 šï * ik 8 picture elements, which have at least two color components, whereby objects from a second category are also found in the picture series. The computer program is characterized by instructions corresponding to the steps: to identify with a first rough measure a first set of pixels in an image in the image series, so that the pixels in the first set predominantly depict objects from the second category; identifying a first function which gives an approximate description of the relationship between the respective intensities of the two color components for pixels in the first set; identifying with a second coarse measure a second set of pixels in an image in the image series, so that the pixels in the second set predominantly depict objects from the first category; identifying a second function which gives an approximate description of the relationship between the respective intensities of the two color components for pixels in the second set; and determining whether or not a given pixel in an image in the image series depicts an object from the first category, based on the respective intensities of the given pixel for the two color components and the first and second functions.

According to a fourth aspect of the invention, it relates to a digital storage medium containing such a program.

The computer program and thus the storage medium entails corresponding advantages as the method and can be varied accordingly.

Brief Description of the Figures Fig. 1 shows an example of a microscope image in which the present invention can be applied.

Fig. 2 shows a diagram in which the pixels in the image in Fig. 1 have been plotted based on their respective intensities for red and blue. ,.; || Fig. 3 shows a red-blue plane in which two approximate functions have been identified according to an embodiment of the invention.

Fig. 4 shows an example where angular values have been determined for the picture elements included in the image in Fig. 1.

Figures 5a and 5b show an arrangement according to an embodiment of the invention.

Fig. 6 shows a flow chart of a method according to an embodiment of the invention.

Description of Preferred Embodiments Fig. 1 shows an example of a microscope image in which the present invention can be applied. In the picture again, in a stained blood smear (stained according to the MGG method), there is a set of blood cells of various kinds.

There are four white blood cells V1- V4, one of which (V3) probably indicates a disease state. The image also contains a large number of red blood cells, R1, R2, etc. Furthermore, a number of platelets T can be seen in the image, which appear as small, dark spots. The blood cells appear against a background B. The image shown in Fig. 1 is a black and white version of a color image. As those skilled in the art know, white blood cells appear in such a color image as slightly more blue than the background and the red blood cells. The contrast between the cytoplasm of the white blood cells and the background therefore appears more clearly in the color image in which a method according to the invention is used than in the black-and-white image reproduced here.

As mentioned, it is particularly difficult to segment certain classes of white blood cells when their cytoplasm is adjacent to red blood cells. This is the case with the white blood cell V1, which is adjacent to the red blood cell R1. The present invention relates to a reliable and efficient way of performing such segmentation.

Fig. 2 shows a diagram in which all the picture elements in the picture in Fig. 1 have been plotted based on their respective intensities for red and blue. The red intensity ||| »»,., - 1 10 l5 20 25 30 35 518 457. 1 ~ '. . . . . . . 10 for a pixel is deposited on the vertical axis and the blue intensity on the horizontal. In the diagram, the plotted pixels can be compared to two elongated "clouds" 11, 13, which meet in an upside-down V-shape. It turns out that the first cloud II, which has greater variation in red intensity, contains almost exclusively pixels that image white blood cells and background. The second cloud 13, which is more collected, contains pixels that depict red blood cells and background. The pixels in the two clouds depicting the background are gathered in a small area around a background center of gravity 15. A white cell center of gravity 16 for the pixels that depict white blood cells can also be determined, as well as a corresponding red cell center of gravity 17 for those depicts red blood cells. It can be said that the pixels of the red and white blood cells deviate from the background center of gravity in two different directions. This is because the red and white blood cells during staining are affected in different ways.

According to an embodiment of the present invention, a first function, a white function 12 in the red-blue plane, which function approximately describes the relationship between the intensities of red and blue in pixels depicting white blood cells, is to be determined. Likewise, a corresponding second function, a red function 14 for the pixels that image red blood cells, must be determined. White blood cells are then said to be objects of a first category, and red blood cells objects of a second category. On the basis of these red and white functions, it can then be determined, with great certainty, whether a given pixel depicts a red or a white blood cell.

These red and white functions can be determined from smaller sets of pixels that image red and white blood cells, respectively. These sets can be identified by means of rough dimensions, so that a predominant part of the picture elements in a set depict objects from the desired category. faan »lO 15 20 25 30 35 518 457 ll IDENTIFICATION OF PICTURES THAT IMAGE RED BLOOD CELLS Unlike white blood cells, it can be assumed that red blood cells are present in each image in a series of images with microscope images of a blood smear. The most convenient way to find a set of pixels that image red blood cells in an image is to exclude the pixels that can be assumed to image the background. Of the remaining amount of pixels in an image, a clear majority will then image red blood cells. Occasional small platelets and possibly some white blood cell may be present, but only a small portion of the remaining pixels depict such objects.

The pixels that depict the background are suitably excluded by thresholding. It is then assumed that the pixels that are brighter than a certain threshold image depict the background. As will be shown later, it may also be appropriate to record the properties of the pixels excluded by the threshold. For example, it may be advantageous to calculate the background center of gravity 15, as will be shown below.

The non-excluded pixels thus depict mainly red blood cells and constitute a roughly selected set of pixels according to an embodiment of the invention.

It is also possible to exclude pixels that can image white blood cells or platelets to further refine the selected set. Platelets can usually be threshed away, as they are darker in the green component than red blood cells.

Nuclei in white blood cells also usually have a markedly low or dark green component and can thus usually be found. Thereafter, it is possible to exclude an area around such nuclei to remove the entire white blood cell. The size of the area is then preferably of the order of magnitude that white blood cells usually have at the current magnification, or larger. Identification of a set of red blood cells does not normally need to be done for each image in a series of images.

DETERMINATION OF APPROXIMATIVE FUNCTION FOR RED BLOOD CELLS Based on the roughly identified set of pixels depicting red blood cells, an approximate function describing the ratio of red to blue intensity for pixels in the set is then determined. This is called red cell function 14.

The function can be substantially linear, ie in the form R = kfB + mr, where R and B describe red and blue intensities respectively. The parameters SEK and mr must then be determined.

This can be done for the set of pixels that image red blood cells by ordinary linear regression, which is an operation well known to those skilled in the art.

Alternatively, the function R = kfB + mr can be determined as the function that intersects both the background center of gravity 15 and the center of red cell 17.

The background center of gravity is calculated as the averages of the blue and red intensities for the set of pixels that depict the background. The red cell center of gravity is calculated as the mean values of the blue and red intensities, respectively, for the set of pixels that image red blood cells. The centers of gravity thus correspond to averages.

Alternatively, points in the red-blue plane based on median values or type values can be used. Both of these concepts are well known to those skilled in the art.

Regardless of whether the above-mentioned function is determined by means of linear regression or using the above-mentioned centers of gravity, it is advantageous in connection with this to determine a measure of the uncertainty in the estimation of the function. This provides a measure of reliability for the entire operation. Such a measure of uncertainty can be based on the mean error of the estimate of the red cell center of gravity or on the mean error of the respective regression parameters. unß: lO 15 20 25 30 35 518 457 l3 IDENTIFICATION OF PICTURES THAT IMAGE WHITE BLOOD CELLS Rough identification of pixels that image white blood cells and determination of an approximate function associated with these pixels can preferably be done for each white blood cell that appears in a series of pictures.

This is preferable because the group of white blood cells has a large mutual variation.

A first rough method to identify picture elements that image white blood cells is to use the previously mentioned low green component, which is found in the nucleus of the white blood cell. Thus, in the set of pixels that are considered to image white blood cells, pixels with a green intensity below a certain threshold value are included. This essentially provides pixels that image the nucleus of a white blood cell.

A second more accurate method for identifying pixels that image white blood cells is to use an active contour model that is designed to originate from an area outside the white blood cell. This can be done as follows.

First, as in the first method, contiguous areas, with a certain minimum size, are identified by picture elements in the digital picture that have a green component below a certain threshold value. These can be assumed to image the nucleus of a white blood cell. Then an area in the digital image is selected which has the nucleus in its center and which is large enough to enclose the entire white blood cell with great certainty. The picture elements in the selected area are used for further processing.

Then, for the picture elements in the selected area, object similarity values are determined with respect to red and blue intensity. The image similarity value of a pixel is a measure of how similar this pixel is to the expected pixels in the white blood cell.

The determination of this object-like value can be made on the basis of the function l4 already determined for red blood cells and the center of gravity 15 determined for the background pixels »nßnu 10 15 20 25 30 35 518 457 14 Each given pixel in the selected quantity has a corresponding position in the diagram in Fig. 2. The given pixel can be assigned an object similarity value which is a (not necessarily linear) function of the angle between a line drawn through this position and the background center of gravity 15, and the line of the red cell function 14. This is done so that small angles give small object similarity values, and large angles large object similarity values.

Then a starting amount is selected from the pixels in the area. The starting amount is selected from the pixels that have the lowest object similarity values, so that the starting amount with great certainty does not include pixels in the white blood cell. The pixels in the initial set are assigned a first object confidence value, for example zero. The object probability value is a measure of how likely it is that a picture element, set in its context, depicts a searched object.

The starting amount is then allowed to grow by including pixels that are adjacent to the starting amount. When a new pixel is included in the set, it is assigned an object probability value which is a function (for example, the sum) of its object similarity value and an adjacent, already included pixel object probability value. The object confidence values then become higher and higher, the later a picture element is included.

This is preferably done until all pixels in the image have been assigned object fidelity values. The object confidence value of a given pixel then depends on the distance to the starting amount before it has started to grow, on its object similarity value and on any object similarity values belonging to any picture element that lies between the given picture element and the starting amount before it has started to grow.

It is also possible to perform the above-mentioned determination of object trust values approximately by means of a modified distance transform. The spacing transform must then be modified so that for a given pixel it determines an object trust value, which depends on the object similarity value of the given object, the Euclidean distance to the starting set and object similarity values of picture elements lying between the given the object and the starting quantity.

Based on the object confidence values that pixels are assigned in the selected area, the pixels that are then to form the basis for the white cell function are selected.

DETERMINATION OF APPROXIMATIVE FUNCTION FOR WHITE BLOOD CELLS Based on the roughly identified set of pixels depicting white blood cells, an approximate function describing the ratio of red to blue intensity of pixels in the set must then be determined. This is called white function. This function can also be substantially linear, ie in the form R = kWB + etc., where R and B describe red and blue intensities, respectively, for pixels that image white blood cells. The parameters kv and nn must then be determined. This can be done for the set of pixels that image white blood cells, preferably by ordinary linear regression, which is a well-known operation to those skilled in the art.

Alternatively, the white function R = kfB + nn can be determined as the function that intersects both the background center of gravity 15 and the white cell center of gravity 16.

The white cell center of gravity 16 is calculated in the same way as the red cell center of gravity 17.

Regardless of whether the above-mentioned white function is determined by means of linear regression or using the above-mentioned centers of gravity, it is advantageous to determine in connection with this a measure of the uncertainty in the estimation of the function. Such a measure can be based on the mean error of the estimate for the white cell center of gravity 16 or on the mean error for the respective regression parameters.

USE OF RED AND WHITE CELL FUNCTIONS Fig. 3 shows a red-blue plane in which two approximate functions, a red 14 and a white cell function 12 have been identified according to an embodiment of the invention. .

According to the invention, for pixels in a digital microscope image, it must be determined whether this depicts objects from the first category, ie in this case white blood cells, or not. This is done on the basis of the pixel's respective intensities for the two color components (in this case red and blue) and the first and second functions (in this case red function and white function respectively).

Fig. 3 shows as an example four different picture elements, 30, 32, 33, 34, which have been plotted in the diagram in accordance with their respective red and blue intensities. With reference to Fig. 2, it can be assumed that the picture elements 33 and 34 most likely depict a white and a red blood cell, respectively. This cannot be said with equal certainty for the pixels 30 and 32. It is likely that the pixels 30 and 32 depict a white and a red cell, respectively.

Preferably, angular values for the pixels in a digital image are determined. For the picture element 30, a corresponding element line 31 has been drawn. The element line 31 of the picture element 30 is such that in the intensity plane of the two color components (red, blue) it intersects both the intensities of the picture element 30 for the two colors and an intersection point 15 between the red and white function.

An angular value for a pixel 30 is based on the angles (ß and a) between the pixel element line 31 and the red and white function 12, respectively. Based on the amount of this angular value, it can be determined whether the pixel depicts an object from the first category or not.

The angular value can preferably be standardized. The angular value of a given pixel then assumes a first extreme angular value, if the element of the pixel - for example 0, ment line coincides with the white function. The same preferably applies even if the angle ß of the element line in relation to the red function is greater than the angle between: anno nous: 10 15 20 25 30 35, .. ~ '". Ü. .I HI;"': 2, ". Uv : 'j' 2 'nno I o I guo. nu ,, ,, uu __,.. v _: HH' '_ .'Z 2 f ~' j, J. '. »' 2 Ä i - u u The pixel 33 thus has the angular value O.

The angular value of a given pixel assumes a second extreme angular value, for example 1, if the element line of the pixel coincides with the red function. The same preferably applies if the angle d of the element line in relation to the white function is greater than the angle between the red and white functions. The pixel 34 thus has the angular value 1.

If the element line of a pixel has smaller angles (ß and d, respectively) in relation to the lines of the red 14 and white functions than the angle between the functions, the angular value assumes a value between the first and the second extreme value, ie between 0 and 1. Preferably this depends on the proportions between the two angles a and ß.

The pixel 30 could therefore have the angular value 0.3 while the pixel 32 could have the angular value 0.9. The distance x to the intersection 15 of the red and white functions 12 in the plane of the two color components can also be used to determine whether the given pixel 30 depicts an object from the first category or not.

Fig. 4 shows an example where angular values have been determined for the picture elements included in the image in Fig. 1.

Bright picture elements have a low angular value there. As can be seen from the picture, the contrast between the cytoplasm of the white blood cells and the adjacent red cells is now much stronger than in Fig. 1. The angular values in Fig. 4 can therefore be used to perform better segmentations of the white blood cells in Fig. 1. For example, pixels can be thresholded with respect to their angular values.

Fig. 5a shows an application of an arrangement according to the invention. The arrangement is part of a system with a microscope 51 and a computer system 52. The microscope 51 delivers microscopic digital color images to the computer system 52, which performs segmentation of interesting objects in the digital images. The computer system 52 and the microscope 51 may be integrated. Parts of segmentation procedure- anon- a | »a: lO l5 20 25 30 35 n ~ o u o.

~ Q ~ ou u un uvøoou s 1 s 457 .2 ïšßšíë 18 it can also be performed distributed over a network. Fig. 5b shows an arrangement according to an embodiment of the invention. The arrangement includes a number of functional modules 53-57.

In a first module (ID1) 53, a first coarse measure identifies a first set of pixels in a digital image, so that the pixels in the first set predominantly depict objects from a certain category, for example red blood cells. The first module 53 constitutes means for performing such identification.

In a second module (FKI) 54, a first, substantially linear function is identified, which gives an approximate description of the relationship between the two color components' respective intensities for pixels in the first set. The second module 54 constitutes means for performing such identification.

In a third module 55 (ID2), a second set of pixels in a digital image is identified with a second roughly measured, so that the pixels in the second set predominantly depict objects from a certain category, for example white blood cells. The third module 55 constitutes means for performing such identification.

In a fourth module (FK2) 56 a second, substantially linear function is identified, which gives an approximate description of the relationship between the two color components' respective intensities for picture elements in the second set. The fourth module 56 is a means of performing such identification.

In a fifth module (SEGM.) 57 it is determined whether or not pixels in an image depict an object from the first category, on the basis of their respective intensities for the two color components and the said first and second functions. The fifth module 57 constitutes means for effecting such a decision and outputs objects segmented from a digital image for further analysis.

All the modules shown above can be realized by means of computer programs in the computer system 52. Such a computer program can be stored separately on a digital storage medium.

However, it is also possible to design some modules as circuit solutions, for example in the form of ASIC or FPGA circuits (ASIC = Application Specific Integrated Circuit; FPGA = Field Programmable Gate Array).

Fig. 6 shows a flow chart of a method 60 according to an embodiment of the invention. The method is used to distinguish objects from a first category in an image series with at least one digital color image, which depicts biological material and comprises a plurality of image elements, which have at least two color components, whereby also objects from a second category are found in the image series.

In a first step 61, a first set of pixels in an image in a series of images is identified with a first rough measure, so that the pixels in the first set predominantly depict objects from the second category. By a measure is meant here and below a rule that is applied to a quantity of input data, ie in this case pixels in an image. The rule gives for a pixel either the result that the pixel depicts an object from the other category or that the pixel does not.

In a second step 62, a first, substantially linear function is identified, which gives an approximate description of the relationship between the two color components' respective intensities for picture elements in the first set.

In a third step 63, with a second rough measurement, a second set of pixels in an image in the image series is identified, so that the pixels in the second set predominantly depict objects from the first category.

In a fourth step 64, a second, substantially linear function is identified, which gives an approximate description of the relationship between the two color components' respective intensities for pixels in the second set. 10 15. . ; - °,; '- ,; .:, '. .3 .z. ... .. 518 457 ä- * íš .§. = FI = É 20 In a fifth step 65 it is determined whether or not a given pixel in an image in the image series depicts an object from the first category, on the basis of that given the respective elements of the picture element for the two color components as well as the mentioned first and second functions.

The image series referred to above preferably contains several images, but can also contain only one image.

It is not necessary to perform all parts of the procedure in one image. The first and second steps can be performed on a first image, the third and fourth on a second image in the image series, and the fifth step on a third.

The invention is not limited to the embodiments shown above, but can be varied within the scope of the appended claims.

Claims (14)

10 15 20 25 30 35 518 457 .n I,. . .a, nn 21 PATENT REQUIREMENTS A method of segmenting objects from a first category in an image series with at least one digital microscope color image, which depicts biological material and comprises a plurality of image elements, which have at least two color components, wherein objects from a second category are also found in the image series, k ä drawn by steps r - identifying (61) a first set of pixels in a color image in the image series with a first rough measure, so that the pixels in the first set predominantly depict objects from the second category; - identifying (62) a first function, which gives an approximate description of the relationship between the respective color components' intensities for pixels in the first set; - identifying (63) a second set of pixels in a color image in the image series with a second coarse measure, so that the pixels in the second set predominantly depict objects from the first category; (64) approximate description of the relationship between the two - to identify a second function, which gives one the respective intensities of the color components for pixels in the second set; and - determining (65) whether or not a given pixel in a color image in the image series depicts an object from the first category, based on the respective intensities of the given pixel for the two color components and said first and second functions . A method according to claim 1, wherein said first and second functions are substantially linear. The method of claim 2, wherein the step of determining whether or not said given pixel depicts an object from the first category comprises the sub-steps: SE0100142-9 10 15 20 25 30 35 s 1 s 457 f; ïï - §II2.LÉ: ~! .I! § “å '. n: I n 22 - to identify for the given pixel an element line which, in the intensity plane of the two color components, intersects both the intensities of the given pixel for the two colors and an intersection point between the first and the second function; - assigning the given pixel an angular value based on the angle of said element line in relation to lines corresponding to said first and second function; - to determine whether or not the given pixel represents an object from the first category based on the angular value. The angular value of the element assumes a first extreme value of the method of claim 3, wherein the given pixel line coincides with the line of the first function or has an angle relative to the line of the second function which is greater than the angle between the first and the second function lines; assumes a second extreme value if the element line coincides with the line of the second function or has an angle relative to the line of the first function which is greater than the angle between the lines of the first and the second function; and assumes a value between the first and the second extreme value if the element line has smaller angles in relation to the lines of the first and second functions than the mutual angle between the lines of the first and the second function. A method according to claim 4, wherein also the distance between the intensities of the given pixel and the intersection of the first and second functions in the plane of the two color components is used to determine whether the given pixel depicts an object from the first category or not. Said objects in the first category are white blood cells. A method according to any one of the preceding claims, wherein said objects in the second category are red blood cells. SE010Û142-9 lO 15 20 25 30 35,. v I 4 o I | -:: I s1s 457: z LI - ëë .ï - alšf 23 A method according to any one of the preceding claims, wherein said two color components are red and blue. A method according to any one of claims 6 or 7, wherein said first coarse measure comprises by threshold identifying and excluding picture elements depicting background. The method of claim 8, wherein said first coarse measure further comprises identifying and excluding areas of pixels, which areas can be assumed to contain white blood cells, the assumption being based on the intensity of the green color component. A method according to any one of claims 6-9, wherein said second coarse dimension comprises identifying at least one low intensity pixel cluster of green color component. 11. ll. The method of claim 10, wherein said second coarse measure further comprises the sub-steps; - selecting an area associated with said pixel cluster, which area has a size larger than the expected size of a white blood cell; selecting at least one pixel in the area, which pixel has an appearance different from the expected appearance of pixels depicting white blood cells, said at least one pixel forming an initial amount; - that object similarity values are determined for picture elements in the area; and - said second set of pixels being identified based on the initial set and the object similarity values. Arrangement for segmenting objects from a first category in a series of images with at least one digital color image, which depicts biological material and comprises a plurality of picture elements, which have at least two color components, wherein objects from a second category are also found in the image series, characterized by means of means (53) for identifying with a first coarse measure a first set of picture elements in a color image in the picture series, so that the picture elements in the first set predominantly depicts objects from the second category; - means (54) for identifying a first function, which gives an approximate description of the relationship between the respective color components' intensities for pixels in the first set; means (55) for identifying with a second coarse measure a second set of pixels in a color image in the image series, so that the pixels in the second set predominantly depict objects from the first category; - means (56) for identifying a second function, which gives an approximate description of the relationship between the respective color components' intensities for pixels in the second set; and - means (57) for determining whether or not a given pixel in a color image in the image series depicts an object of the first category, based on the respective intensities of the given "pixel for the two color components and said first and second functions. Computer program for segmenting objects from a first category into a series of images with at least one digital color image, which depicts biological material and comprises a plurality of picture elements, which have at least two color components, whereby also objects from a second category are present in the image series, characterized by instructions corresponding to the steps: - to identify with a first rough measure a first set of pixels in a color image in the image series, so that the pixels in the first set predominantly depict objects from the second category; - to identify a first function, which gives an approximate description of the relationship between the two 350100142-9 10 15 20 u 0 o u: O 1 una I 1 1 n no ~ n 518 457 n .una. u u »o o 25 the respective intensities of the color components for pixels in the first set; - to identify with a second coarse measure a second set of picture elements in a color image in the series of pictures, so that the picture elements in the second set predominantly depict objects from the first category; - to identify a second function, which gives an approximate description of the relationship between the respective intensities of the two color components for pixels in the second set; and - determining whether or not a given pixel in a color image in the image series depicts an object of the first category, based on the respective intensities of the given pixel for the two color components and said first and second functions. A digital storage medium, comprising computer programs according to claim 13.