Result filter and method for selecting the result data of an application for automatic pattern recognition

• US 8,116,574 B2
• Filed: 09/18/2007
• Issued: 02/14/2012
• Est. Priority Date: 09/19/2006
• Status: Expired due to Fees

• Current Assignee: Siemens Healthcare GMBH (Siemens AG)
• Original Assignee: Siemens AG
• Inventors: Engels, Heinz-Peter, Feuerlein, Ute
• Primary Examiner(s): Liew, Alex
• Application Number:
  US11/898,980
• Publication Number:
  US 20080069477A1
• Time in Patent Office: 1,610 Days
• Field of Search: 382128-134, 382141-152
• US Class Current: 382/224
• CPC Class Codes: G06F 18/41 Interactive pattern learning with a human teacher
• CPC Class Codes: G06V 2201/03 Recognition of patterns in medical or anatomical images

Abstract

A method is disclosed for automatically classifying and graphically visualizing image objects that are segmented and, given undershooting of a prescribed distance and compliance with a similarity criterion, are combined to form clusters. In at least one embodiment, for object classification, the method includes preselecting result data, using a prescribable selection criterion, from a result set of an application, executed in the background on an image data record of an imaging system, for feature extraction and automatic pattern recognition of segmented and clustered image objects, and/or rendered image data of an image rendering application, executed in the background, for two-dimensional and/or three-dimensional graphic visualization of these image objects; and/or marking the data in a graphically visible fashion as preselected on a screen of a screen terminal.

Claims

1. A method for at least one of automatically classifying and graphically visualizing image objects:

combining a plurality of segmented image objects, given undershooting of a distance and compliance with a similarity criterion, to form clusters;

automatically preselecting at least one of,

result data, for object classification, using a prescribable attributive selection criterion, from a result set of an application for feature extraction and automatic pattern recognition of the segmented and clustered image objects, and

rendered image data of an image rendering application for at least one of two-dimensional and three-dimensional graphic visualization of the image objects; and

marking the preselected data, in a graphically visible fashion, as preselected on a screen of a screen terminal, wherein the automatically preselected at least one of result data and rendered image data is executed simultaneously in a background on an image data record of an imaging system, with the marking the preselected data, and the result set is a list of findings to be evaluated that is sorted according to a specific sorting criterion, the result set is automatically updated upon recognition, identification and deletion of a wrongly classified image object, the result data following as a respective next list element is automatically added to the result set upon the recognition, identification and deletion of the wrongly classified image object.

2. The method as claimed in claim 1, wherein the automatic preselection of at least one of the result data and the rendered image data is performed directly after termination of the application for automatic pattern recognition of segmented and clustered image objects.

3. The method as claimed in claim 1, wherein the automatic preselection of at least one of the result data and the rendered image data is performed subsequent to an image rendering application, executed in the background at least one of during and after execution of the application for automatic pattern recognition, for providing rendered image data for at least one of a two-dimensional and three-dimensional graphic visualization of at the at least one of segmented and clustered image objects.

4. The method as claimed in claim 2, wherein object classes determined by the pattern recognition application are selected by restricting an object class space, forming the basis of the pattern recognition application, via user-specified selection criteria of a result filter, and image data of identified image objects whose object classes are not included in the restricted object class space are suppressed.

5. The method as claimed in claim 3, wherein only classified image objects that are preselected and marked in a graphically visible fashion as preselected within an image section selected by a region marking are displayed on the screen of the screen terminal by suppressing the transmission to the screen terminal of rendered image data of all remaining image sections and classified image objects.

6. The method as claimed in claim 3, wherein the rendered image data of at least one of the non-selected image sections and of the classified image objects that are not at least one of preselected and marked in a graphically visible fashion as preselected, are transmitted upon request to the screen terminal and graphically visualized on the screen.

7. The method as claimed in claim 1, wherein the prescribable attributive selection criterion includes at least one of a designation, a type, a status, a size, a number, a prescribable priority, and a relevance of classified image objects that is to be taken into consideration for a later evaluation.

8. The method as claimed in claim 1, wherein result data of additional findings that relate to further image objects at least one of identified and classified by a user are manually addable to the list of the findings to be evaluated.

9. The method as claimed in claim 8, wherein object markings of image objects from the result set of deleted findings are also displayed on the screen of the screen terminal after the deletion of these findings.

10. The method as claimed in claim 9, wherein, depending on at least one of designation, type, status, size, priority and relevance of at least one of the marked image objects and the image objects included in marked image sections, object markings and region markings for at least one of marking image objects and image sections are stored and graphically displayed as graphic symbols of different at least one of shape and color or as text labels of at least one of different font size, font type and font color.

15. A computer software program product for carrying out a method as claimed in claim 1 during operation on a screen terminal of a pattern recognition and image rendering system.

16. The method as claimed in claim 5, wherein the rendered image data of at least one of the non-selected image sections and of the classified image objects that are not at least one of preselected and marked in a graphically visible fashion as preselected, are transmitted upon request to the screen terminal and graphically visualized on the screen.

11. A pattern recognition and image rendering system for at least one of automatically classifying and graphically visualizing image objects, the system comprising:

a processor to combine a plurality of segmented image objects, given undershooting a distance and compliance with a similarity criterion, to form clusters; and

a result filter to automatically preselect result data for object classification from a result set of an application, simultaneously executed in a background on an image data record of an imaging system with a marking of the preselect result data, for feature extraction and automatic pattern recognition of the segmented and clustered image objects, and the result set is a list of findings to be evaluated that is sorted according to a specific sorting criterion, the result set is automatically updated upon recognition, identification and deletion of a wrongly classified image object, the result data following as a respective next list element is automatically added to the result set upon the recognition, identification and deletion of the wrongly classified image object.

12. The pattern recognition and image rendering system as claimed in claim 11, further comprising a second result filter to graphically visible mark preselected rendered image data of an image rendering application, executed in the background, for at least one of two-dimensional and three-dimensional graphic visualization of classified image objects on a screen of a screen terminal by using a prescribable attributive selection criterion.

13. The pattern recognition and image rendering system as claimed in claim 12, wherein the result filter is directly downstream of the application for automatic pattern recognition of the segmented and clustered image objects.

14. The pattern recognition and image rendering system as claimed in claim 13, wherein the second result filter is downstream of an image rendering application, executed at least one of during and after execution of the application for automatic pattern recognition, for graphically visualizing the at least one of segmented and clustered image objects.

17. The pattern recognition and image rendering system as claimed in claim 11, wherein the result filter is useable to at least one of automatically preselect and graphically visible marking of preselected rendered image data of an image rendering application, executed in the background, for at least one of two-dimensional and three-dimensional graphic visualization of classified image objects on a screen of a screen terminal by using a prescribable attributive selection criterion.

18. The pattern recognition and image rendering system as claimed in claim 11, wherein the result filter is directly downstream of the application for automatic pattern recognition of the segmented and clustered image objects.

19. A method for automatically classifying and graphically visualizing image objects, the method comprising:

combining a plurality of segmented image objects, given undershooting of a distance and compliance with a similarity criterion, to form clusters;

automatically preselecting at least one of

result data, for object classification, using a prescribable attributive selection criterion, from a result set of an application for feature extraction and automatic pattern recognition of segmented and clustered image objects, and

rendered image data of an image rendering application for at least one of two-dimensional and three-dimensional graphic visualization of the image objects; and

marking the preselected data, in a graphically visible fashion, as preselected on a screen of a screen terminal, wherein the automatically preselected at least one of result data and rendered image data is executed simultaneously in a background on an image data record of an imaging system, with the marking the preselected data, and the result set is a list of findings to be evaluated that is sorted according to a specific sorting criterion, the result set is automatically updated upon recognition, identification and deletion of a wrongly classified image object, the result data following as a respective next list element is automatically added to the result set upon the recognition, identification and deletion of the wrongly classified image object.