Method and Device for the Segmentation of Regions and Related Computer Program Product

• US 20080193014A1
• Filed: 01/24/2008
• Published: 08/14/2008
• Est. Priority Date: 07/26/2005
• Status: Active Grant

• Current Assignee: Tissue Gnostics GmbH
• Original Assignee: Tissue Gnostics GmbH
• Inventors: Steiner, Georg E., Laurain, Vincent, Ecker, Rupert
• Granted Patent: US 8,023,737 B2
• Time in Patent Office: Days
• US Class Current: 382/173
• CPC Class Codes: G06T 2207/10056 Microscopic image
• CPC Class Codes: G06T 2207/20156 Automatic seed setting
• CPC Class Codes: G06T 2207/30024 Cell structures in vitro; Tissue sections in vitro
• CPC Class Codes: G06T 7/11 Region-based segmentation
• CPC Class Codes: G06T 7/187 involving region growing; involving region merging; involving connected component labelling
• CPC Class Codes: G06T 7/194 involving foreground-background segmentation
• CPC Class Codes: G06V 20/695 Preprocessing, e.g. image segmentation

Abstract

The invention relates to a method for segmentation of locally-defined regions. The input data obtained for the image region under investigation is separated into a foreground mask and a background mask by application of a threshold value, two different filter operations are applied to the input data to determine segmentation starting points, the resulting data sets are subtracted and the image regions viewed as segmentation starting points, a filter operation is applied to generate a less restrictive mask for the input data, the resulting data subjected to a threshold value reduction, the less restrictive mask subjected to a labeling with regard to determination of the local values for generation of a restrictive mask, the mean values determined, segmentation starting points not obtained in the less restrictive mask are removed, a growth into all regions of the restrictive mask is carried out and the regions obtained in the restrictive mask are viewed as the segmented brightness regions.

Claims

1. A method for the segmentation of locally defined, two-dimensional regions, the method which comprises:

separating input data acquired from an image region under investigation into a foreground mask containing foreground data corresponding to relatively larger or higher values, and a background mask containing background data corresponding to relatively lower values by applying a threshold value;

subjecting the input data to two different filtering operations to determine segmentation starting points, with a filter reducing sharp contours or contrasts, and a filter for determining local maxima in a region;

subtracting the data sets resulting from the two different filtering operations and defining image regions having a specific threshold value or threshold value range as segmentation starting points;

generating a less restrictive mask by subjecting the input data to a filtering operation accentuating high-frequency parts or a filtering operation detecting edges, and applying a specific threshold value to the resulting data, and correlating the data set obtained therewith the data of the foreground mask;

generating a restrictive mask by subjecting the less restrictive mask to a labeling with respect to the determination of local values of each individual locally defined region or cluster, and determining median values of the respective regions, and subjecting the values obtained in each cluster to a discrimination with a preset threshold value;

eliminating segmentation starting points not contained in the less restrictive mask;

effecting a growth in the respective regions of the restrictive mask proceeding from the individual segmentation starting points;

continuing a growth of these regions in the less restrictive mask; and

viewing the regions obtained in the restrictive mask as the segmented brightness regions.

2. The method according to claim 1, which comprises segmenting brightness distributions for showing nuclei agglomerates or unions, or particle size distribution within a sample under investigation.

3. The method according to claim 1, wherein the separating step comprises defining the foreground mask and the background mask according to brightness values.

4. The method according to claim 1, wherein the filtering operations include a low-pass filter and a local maximum filter and wherein the input data are correlated with the foreground mask.

5. The method according to claim 4, wherein the low-pass filter is a Gaussian filter and the input data are multiplied with the foreground mask.

6. The method according to claim 1, wherein the threshold value for the filtering operations is a value zero.

7. The method according to claim 1, wherein the less restrictive mask is generated after a median filtration and/or a Gaussian filtration and the filtering operation detecting edges is effected with a high-pass filter.

8. The method according to claim 1, wherein the high-pass filter is a Laplace filter.

9. The method according to claim 1, wherein the labeling is effected with respect to the determination of grey values, the median values are grey values, and the grey values in each cluster are subjected to a discrimination with a preset threshold value.

10. The method according to claim 1, which comprises continuing the growth in the less restrictive mask and subsequently in the foreground mask;

and viewing the regions obtained in the foreground mask as the segmented brightness regions.

11. The method according to claim 1, which comprises using a single or multi-step dilation on the ascertained segmentation starting points.

12. The method according to claim 9, which comprises discriminating with a preset grey value when the grey values or the grey value mask is discriminated during labeling.

13. The method according to claim 5, which comprises correlating or adjusting a filter size of the local maximum filter to a size of the Gaussian filter.

14. The method according to claim 1, which comprises:

to separate the input data, generating a diagram showing the data distribution of the input data;

calculating a differential quotient of a smoothed diagram showing the data distribution;

calculating absolute values of the differential quotients and, optionally, adding a constant value;

correlating the data thus obtained with the input data and, optionally, resealing the data; and

selecting that value as threshold value at which the function obtained corresponds to a specific threshold value or has come close to the threshold value.

15. The method according to claim 14, which comprises separating the input data into bright foreground data and dark background data and showing the data distribution in a brightness histogram.

16. The method according to claim 1, which comprises reproducing the representations or images and also segmented nuclei or nuclei agglomerates which are brightly colored in comparison to other cell regions or in comparison to a cytoplasm.

17. The method according to claim 1, which comprises segmenting colored nuclei or nuclei agglomerates contained in tissue slices.

19. The device according to claim 18, configured for carrying out the method according to claim 1.

32. A computer program product with program code stored on a computer-readable data medium for carrying out the method according to claim 1 when the program product is executed on a computer.

18. A device for the segmentation of locally defined, two-dimensional regions, the device comprising:

a discriminator connected to receive input data related to an image region under investigation and configured to separate the input data into a foreground mask containing foreground data corresponding to greater values and a background mask containing background data corresponding to lower values;

a filter unit connected to said discriminator and configured to determine segmentation starting points by subjecting the input data to two different filtrations, said filter unit including a filter for reducing sharp contours or contrasts, and a local maximum filter;

a subtraction unit connected to said filter unit for subtracting the data sets resulting from the two different filtrations, and for determining, storing, and/or making available for further calculation, the image regions having a specific threshold value or threshold value range;

a filter for generating a less restrictive mask by detecting edges or accentuating the high-frequency parts of the input data;

a labeling unit for generating a restrictive mask, said labeling unit subjecting the less restrictive mask to a labeling with respect to a determination of the local values of each individual locally defined region or cluster, and wherein the median values of the respective regions are determined, whereby the values obtained in each cluster are subjected to a discrimination with a preset threshold value;

a comparator connected to said filter and configured to remove segmentation starting points not contained in the less restrictive mask;

a growing unit connected to said comparator, said growing unit undertaking a growth in the respective regions of the restrictive mask proceeding from individual segmentation starting points, and continuing the growth of these regions in the less restrictive mask, and optionally subsequently in the foreground mask; and

at least one further unit selected from the group consisting of a storage unit, a display unit, and a further evaluation unit for storing, displaying, and/or further processing the regions obtained in the restrictive mask, or in the foreground mask, as the desired segmented brightness regions.

20. The device according to claim 18, configured for segmentation of brightness distributions or regions, in particular for showing nuclei agglomerates or unions, particle size distribution, or data fields.

21. The device according to claim 18, wherein the foreground mask contains brightness values, and the background mask contains dark values.

22. The device according to claim 18, wherein said filter for reducing sharp contours or contrasts is a low-pass filter.

23. The device according to claim 22, wherein said low-pass filter is a Gaussian filter.

24. The device according to claim 18, wherein said filter for generating a less restrictive mask is a high-pass filter.

25. The device according to claim 24, wherein said filter for generating a less restrictive mask is a Laplace filter.

26. The device according to claim 25, wherein a median filter and/or Gaussian filter is connected upstream of said Laplace filter and a correlator connected downstream thereof, for subjecting the data to a threshold value application with a specific threshold value and the data set obtained therewith is correlated with the data of the foreground mask.

27. The device according to claim 18, which comprises a single-step dilation unit for the segmentation starting points.

28. The device according to claim 18, wherein:

said discriminator comprises a diagram component showing the data distribution for the input data;

a differential quotient component is provided for the smoothed diagram, said component calculating the absolute values of the differential quotients and optionally adding a constant value;

a correlator is provided for linking the data obtained from the differential quotient component with the input data; and

a threshold component is provided for selecting that value as threshold value at which the function assumes a specific threshold value or has come close to the specific threshold value.

29. The device according to claim 28, wherein said diagram component is a histogram component and the input data are separated into bright foreground data and dark background data, said correlator is a multiplier, and said specific threshold value is zero.

30. The device according to claim 28, configured in a computer with corresponding hardware and software required for carrying out the process.

31. A method for segmentation of nuclei or nuclei agglomerates which are colored dark in comparison to other cell regions or in comparison to the cytoplasm, and/or of colored nuclei or nuclei agglomerates contained in tissue slices, which comprises:

providing a device according to claim 18 and carrying out the functional steps required for the segmentation of the nuclei or nuclei agglomerates.