System and method for automation of morphological segmentation of bio-images

• US 7,668,351 B1
• Filed: 01/20/2004
• Issued: 02/23/2010
• Est. Priority Date: 01/17/2003
• Status: Expired due to Fees

• Current Assignee: Kestrel Corp.
• Original Assignee: Kestrel Corp.
• Inventors: Wilson, Mark, Raman, Balaji, Soliz, Peter
• Primary Examiner(s): Ahmed; Samir A.
• Assistant Examiner(s): TSAI, TSUNG YIN
• Application Number:
  US10/760,923
• Time in Patent Office: 2,226 Days
• Field of Search: 351/237, 356/401, 382/117, 382/132, 382/224, 702/19
• US Class Current: 382/128
• CPC Class Codes: G06T 2207/10048 Infrared image
• CPC Class Codes: G06T 2207/10064 Fluorescence image
• CPC Class Codes: G06T 2207/20012 Locally adaptive
• CPC Class Codes: G06T 2207/30041 Eye; Retina; Ophthalmic
• CPC Class Codes: G06T 7/155 involving morphological operators

Abstract

Medical images are automatically segmented by customizing the morphological segmentation of features identified in the image based upon statistical analysis of the features within each region to be analyzed. The statistical description of the features, as reported through a feature vector, informs the system as to which input variables to select for further segmentation analysis for features residing within the region of the image analyzed. By customizing the automatic segmentation analysis to produce an enhanced image, features within the image are characterized more efficiently and precisely. False positive identification of lesions are minimized without sacrifice of true positive identifications.

Claims

1. A method for characterizing a digital image comprising:

identifying a feature of interest within the image;

applying a feature algorithm to the image based on the feature of interest to determine a feature vector for the image;

selecting among predefined gray level morphology algorithm input variables unique for the image, based upon the determined feature vector for the image, to provide a customized gray level morphology algorithm to apply to the image;

applying the customized gray level morphology algorithm to the image to provide enhanced image data; and

calculating feature statistics based on the enhanced image data.

2. The method of claim 1, wherein the feature algorithm comprises a symmetry algorithm.

3. The method of claim 1, wherein the feature algorithm comprises a skewness algorithm.

4. A method for characterizing a retinal image comprising:

identifying a feature of interest within the retinal image;

applying a feature algorithm to the retinal image based on the feature of interest to determine a feature vector for the retinal image;

selecting among predefined gray level morphology algorithm input variables for the retinal image, based upon the feature vector for the retinal image, to provide a customized gray level morphology algorithm for the retinal image;

applying the customized gray level morphology algorithm to the retinal image to provide enhanced retinal image data; and

calculating feature statistics based on the enhanced retinal image data.

5. The method of claim 4, wherein the feature algorithm comprises a symmetry algorithm.

6. The method of claim 4, wherein the feature algorithm comprises a skewness algorithm.

7. A method for characterizing a digital image comprising:

defining the digital image into plural defined regions;

identifying features of interest within the defined regions;

applying feature algorithms to each of the defined regions based on the features of interest to determine a feature vector for each defined region;

selecting among predefined morphology algorithm input variables for each defined region, based upon the feature vector for that respective defined region, to provide a customized morphology algorithm for each defined region;

applying the customized morphology algorithms to each respective defined region to provide enhanced image data; and

calculating feature statistics based on the enhanced image data.

8. The method of claim 7, wherein the feature algorithm comprises a symmetry algorithm.

9. The method of claim 7, wherein the feature algorithm comprises a skewness algorithm.

10. A method for characterizing a digital image of a retina, the method comprising:

defining the digital image into one or more defined regions;

identifying a feature of interest within the defined regions;

applying a symmetry algorithm and a skewness algorithm to each of the defined regions based on the feature of interest to determine a feature vector for each defined region;

selecting among predefined morphology algorithm input variables for each defined region, based upon the feature vector for that respective defined region, to provide customized morphology algorithms for each defined region;

applying the customized morphology algorithms to each respective defined region to provide regional results having enhanced image data;

combining the regional results to form an enhanced image; and

calculating feature statistics for the enhanced image.

11. A computer program product for enabling a computer to characterize a digital image, the computer program product comprising

software instructions for enabling the computer to perform predetermined operations, and

a computer readable medium embodying the software instructions;

wherein the predetermined operations comprise;

identifying a feature of interest within the image;

applying a feature algorithm to the image based on the feature of interest to determine a feature vector for the image;

selecting among predefined morphology algorithm input variables for the image, based upon the feature vector for the image, to provide a customized morphology algorithm for the image;

applying the customized morphology algorithm to the image to provide enhanced image data; and

calculating feature statistics based on the enhanced image data.

12. A computer program product for enabling a computer to characterize a digital image of a retina, the computer program product comprising

software instructions for enabling the computer to perform predetermined operations, and

a computer readable medium embodying the software instructions;

wherein the predetermined operations comprise;

defining the digital image into one or more defined regions;

identifying a feature of interest within the defined regions;

applying a symmetry algorithm and a skewness algorithm to each of the defined regions based on the feature of interest to determine a feature vector for each defined region;

selecting among predefined morphology algorithm input variables for each defined region, based upon the feature vector for that respective defined region, to provide customized morphology algorithms for each defined region;

applying the customized morphology algorithms to each respective defined region to provide regional results having enhanced image data;

combining the regional results to form an enhanced image; and

calculating feature statistics for the enhanced image.

13. A system for mathematically characterizing a digital image, the system comprising:

a programmable machine;

software instruction for operating on the programmable machine and adapted to cause the programmable machine to execute;

identifying a feature of interest within the image;

applying a feature algorithm to the image based on the feature of interest to determine a feature vector for the image;

selecting among predefined morphology algorithm input variables for the image, based upon the feature vector for the image, to provide a customized morphology algorithm for the image;

applying the customized morphology algorithm to the image to provide enhanced image data; and

calculating feature statistics based on the enhanced image data.

14. A system for mathematically characterizing a digital image of a retina, the system comprising:

a programmable machine;

software instruction for operating on the programmable machine and adapted to cause the programmable machine to execute;

defining the digital image into one or more defined regions;

identifying a feature of interest within the defined regions;

applying a symmetry algorithm and a skewness algorithm to each of the defined regions based on the feature of interest to determine a feature vector for each defined region;

selecting among predefined morphology algorithm input variables for each defined region, based upon the feature vector for that respective defined region, to provide customized morphology algorithms for each defined region;

applying the customized morphology algorithms to each respective defined region to provide regional results having enhanced image data;

combining the regional results to form an enhanced image; and

calculating feature statistics for the enhanced image.

15. A method for characterizing features of interest in a digital image, the method comprising:

identifying a set of one or more potential features of interest (PFOIs) to characterize in the image;

characterizing each of the PFOIs quantitatively with respect to a selected aspect and obtaining a respective aspect value;

selecting a filter threshold value from a receiver operating characteristic curve specific for the selected aspect, wherein the filter threshold value is selected based upon a user input variable; and

filtering the PFOIs with a filter specific to the selected aspect and using the selected filter threshold value, wherein filtering removes from the set of identified PFOIs any of the PFOIs with a quantitative value for the aspect that is equal to or less than the filter threshold value thereby producing a filtered set of identified PFOIs;

wherein the selected aspect comprises a first ordered aspect;

wherein the steps of characterizing, selecting, and filtering are repeated for each of one or more subsequent ordered aspects;

wherein, after repeating the steps of characterizing, selecting, and filtering for each of one or more subsequent ordered aspects, any PFOIs remaining in the set of identified PFOIs are then characterized as being features of interest.

16. The method for characterizing features of interest of claim 15, wherein the first and subsequent ordered aspects have their order determined via a genetic algorithm.

17. The method for characterizing features of interest of claim 16, wherein the genetic algorithm selects for maximization of true positive characterization of features of interest.

18. The method for characterizing features of interest of claim 16, wherein the genetic algorithm selects for minimization of false positive characterization of features of interest.

19. The method for characterizing features of interest of claim 16, wherein the genetic algorithm selects for minimization of false positive characterization of features of interest without substantial sacrifice of true positive characterization of features of interest.

20. A method for assessing symptoms of disease by characterizing features of interest in a digital image of an organ, the method comprising:

identifying a set of one or more potential features of interest (PFOIs) to characterize in the image, the potential features of interest being lesions;

characterizing each of the PFOIs quantitatively with respect to a selected aspect and obtaining a respective aspect value;

selecting a filter threshold value from a receiver operating characteristic curve specific for the selected aspect, wherein the filter threshold value is selected based upon a user input variable; and

filtering the PFOIs with a filter specific to the selected aspect and using the selected filter threshold value, wherein filtering removes from the set of identified PFOIs any of the PFOIs with a quantitative value for the aspect that is equal to or less than the filter threshold value thereby producing a filtered set of identified PFOIs;

wherein the selected aspect comprises a first ordered aspect;

wherein the steps of characterizing, selecting, and filtering are repeated for each of a second through sixth ordered aspects;

wherein, after repeating the steps of characterizing, selecting, and filtering for each of the ordered aspects, any PFOIs remaining in the set of identified PFOIs are then characterized as being features of interest.

21. The method for assessing symptoms of disease of claim 20, wherein the first ordered aspect comprises aspect ratio, the second ordered aspect comprises edge sharpness, the third ordered aspect comprises complexity, the fourth ordered aspect comprises roundness, the fifth ordered aspect comprises hue, and the sixth ordered aspect comprises saturation.

22. A computer program product for enabling a computer to characterize features of interest in a digital image, the computer program product comprising

software instructions for enabling the computer to perform predetermined operations, and

a computer readable medium embodying the software instructions;

wherein the predetermined operations comprise;

identifying a set of one or more potential features of interest (PFOIs) to characterize in the image;

characterizing each of the PFOIs quantitatively with respect to a selected aspect and obtaining a respective aspect value;

selecting a filter threshold value from a receiver operating characteristic curve specific for the selected aspect, wherein the filter threshold value is selected based upon a user input variable; and

filtering the PFOIs with a filter specific to the selected aspect and using the selected filter threshold value, wherein filtering removes from the set of identified PFOIs any of the PFOIs with a quantitative value for the aspect that is equal to or less than the filter threshold value thereby producing a filtered set of identified PFOIs;

wherein the selected aspect comprises a first ordered aspect;

wherein the steps of characterizing, selecting, and filtering are repeated for each of one or more subsequent ordered aspects;

wherein, after repeating the steps of characterizing, selecting, and filtering for each of one or more subsequent ordered aspects, any PFOIs remaining in the set of identified PFOIs are then characterized as being features of interest.

23. A computer program product for enabling a computer to assess symptoms of disease by characterizing features of interest in a digital image of an organ, the computer program product comprising

software instructions for enabling the computer to perform predetermined operations, and

a computer readable medium embodying the software instructions;

wherein the predetermined operations comprise;

identifying a set of one or more potential features of interest (PFOIs) to characterize in the image, the potential features of interest being lesions;

characterizing each of the PFOIs quantitatively with respect to a selected aspect and obtaining a respective aspect value;

selecting a filter threshold value from a receiver operating characteristic curve specific for the selected aspect, wherein the filter threshold value is selected based upon a user input variable; and

filtering the PFOIs with a filter specific to the selected aspect and using the selected filter threshold value, wherein filtering removes from the set of identified PFOIs any of the PFOIs with a quantitative value for the aspect that is equal to or less than the filter threshold value thereby producing a filtered set of identified PFOIs;

wherein the selected aspect comprises a first ordered aspect;

wherein the steps of characterizing, selecting, and filtering are repeated for each of a second through sixth ordered aspects;

wherein, after repeating the steps of characterizing, selecting, and filtering for each of the ordered aspects, any PFOIs remaining in the set of identified PFOIs are then characterized as being features of interest.

24. The computer program product of claim 23, wherein the first ordered aspect comprises aspect ratio, the second ordered aspect comprises edge sharpness, the third ordered aspect comprises complexity, the fourth ordered aspect comprises roundness, the fifth ordered aspect comprises hue, and the sixth ordered aspect comprises saturation.

25. A system for characterizing features of interest in a digital image, the system comprising:

a programmable machine;

software instruction for operating on the programmable machine and adapted to cause the programmable machine to execute;

identifying a set of one or more potential features of interest (PFOIs) to characterize in the image;

characterizing each of the PFOIs quantitatively with respect to a selected aspect and obtaining a respective aspect value;

selecting a filter threshold value from a receiver operating characteristic curve specific for the selected aspect, wherein the filter threshold value is selected based upon a user input variable; and

filtering the PFOIs with a filter specific to the selected aspect and using the selected filter threshold value, wherein filtering removes from the set of identified PFOIs any of the PFOIs with a quantitative value for the aspect that is equal to or less than the filter threshold value thereby producing a filtered set of identified PFOIs;

wherein the selected aspect comprises a first ordered aspect;

wherein the steps of characterizing, selecting, and filtering are repeated for each of one or more subsequent ordered aspects;

wherein, after repeating the steps of characterizing, selecting, and filtering for each of one or more subsequent ordered aspects, any PFOIs remaining in the set of identified PFOIs are then characterized as being features of interest.

26. A system for assessing symptoms of disease by characterizing features of interest in a digital image of an organ, the system comprising:

a programmable machine;

software instruction for operating on the programmable machine and adapted to cause the programmable machine to execute;

identifying a set of one or more potential features of interest (PFOIs) to characterize in the image, the potential features of interest being lesions;

characterizing each of the PFOIs quantitatively with respect to a selected aspect and obtaining a respective aspect value;

selecting a filter threshold value from a receiver operating characteristic curve specific for the selected aspect, wherein the filter threshold value is selected based upon a user input variable; and

filtering the PFOIs with a filter specific to the selected aspect and using the selected filter threshold value, wherein filtering removes from the set of identified PFOIs any of the PFOIs with a quantitative value for the aspect that is equal to or less than the filter threshold value thereby producing a filtered set of identified PFOIs;

wherein the selected aspect comprises a first ordered aspect;

wherein the steps of characterizing, selecting, and filtering are repeated for each of a second through sixth ordered aspects;

wherein, after repeating the steps of characterizing, selecting, and filtering for each of the ordered aspects, any PFOIs remaining in the set of identified PFOIs are then characterized as being features of interest.

27. The system for assessing symptoms of disease of claim 26, wherein the first ordered aspect comprises aspect ratio, the second ordered aspect comprises edge sharpness, the third ordered aspect comprises complexity, the fourth ordered aspect comprises roundness, the fifth ordered aspect comprises hue, and the sixth ordered aspect comprises saturation.

28. A method for characterizing images comprising:

obtaining a first digital image of a feature to create a universal image of the feature;

obtaining a subsequent digital image of the feature;

registering the subsequent image to the universal image to create a registered image;

defining the registered image into regions;

identifying features of interest within the defined regions;

applying one or more feature vector algorithms to determine feature vectors for each defined region;

classifying feature vectors from each defined region into classes;

applying an optimized morphology algorithm to each defined region; and

calculating feature statistics for features in a registered image.