System for detection of malignancy in pulmonary nodules

• US 6,738,499 B1
• Filed: 07/02/2001
• Issued: 05/18/2004
• Est. Priority Date: 11/13/1998
• Status: Expired due to Fees

• Current Assignee: ARCH Development Corporation (University of Chicago)
• Original Assignee: ARCH Development Corporation (University of Chicago)
• Inventors: Nakamura, Katsumi, Doi, Kunio
• Primary Examiner(s): Patel, Jayanti K.
• Application Number:
  US09/830,574
• Time in Patent Office: 1,051 Days
• Field of Search: 382/100, 382/128, 382/129-134, 382/108, 382/157, 382/156, 382/181, 382/286, 324/309, 378/37, 600/407, 606/130
• US Class Current: 382/128
• CPC Class Codes: G06T 2207/30061 Lung
• CPC Class Codes: G06T 7/0012 Biomedical image inspection

Abstract

A method, computer program product, and system (100) for computerized analysis of the likelihood of malignancy in a pulmonary nodule using artificial neural networks (ANNs) (S4). The method, on which the computer program product and the system is based on, includes obtaining a digital outline of a nodule; generating objective measures corresponding to physical features of the outline of the nodule; applying the generated objective measures to an ANN; and determining a likelihood of malignancy of the nodule based on an output of the ANN. Techniques include novel developments and implementations of artificial neural networks and feature extraction for digital images. Output from the inventive method yields an estimate of the likelihood of malignancy (S7) for a pulmonary nodule.

Claims

1. A method for analyzing a nodule, comprising:

obtaining an image of the nodule;

obtaining a digital outline of the nodule;

generating objective measures corresponding to physical features of the nodule based on the outline, said physical features selected from the group consisting essentially of effective diameter, degree of circularity, degree of ellipticity, root mean square variation, first moment of power spectrum of a function corresponding to the distance between a calculated ellipse and the outline, degree of irregularity, average gradient, radial gradient index, tangential gradient index, line enhancement index, average pixel value, and standard deviation of pixel values;

applying the generated objective measures to an artificial neural network; and

determining a likelihood of malignancy of the nodule based on an output of the artificial neural network.

2. The method of claim 1, wherein:

said step of obtaining an image of the nodule comprises obtaining a digital image of the nodule; and

said step of obtaining a digital outline of a nodule comprises extracting the outline of the nodule from the digital image.

3. The method of claim 1, wherein the applying step further comprises:

applying to the artificial neural network at least one clinical parameter corresponding to the nodule.

4. The method of claim 3, wherein the step of applying the at least one clinical parameter comprises:

selecting the at least one clinical parameter from the group consisting essentially of;

age and gender.

5. A method for training an artificial neural network to analyze a candidate nodule, comprising:

obtaining an image of a training nodule;

obtaining a digital outline of the training nodule;

generating objective measures corresponding to physical features of the training nodule based on the out line, said physical features selected from the group consisting essentially of effective diameter, degree of circularity, degree of ellipticity, root mean square variation, first moment of power spectrum of a function corresponding to the distance between a calculated ellipse and the outline, degree of irregularity, average gradient, radial gradient index, tangential gradient index, line enhancement index, average pixel value, and standard deviation of pixel values;

applying the generated objective measures to an artificial neural network; and

training the artificial neural network to determine a likelihood of malignancy in a candidate nodule, based on the objective measures.

6. The method of claim 5, wherein:

the step of obtaining the image of the training nodule comprises obtaining a digital image of the training nodule; and

the step of obtaining the outline of the training nodule comprises extracting the outline of the training nodule from the digital image.

7. The method of claim 5, wherein the training step further comprises:

training the artificial neural network based on at least one clinical parameter corresponding to the training nodule.

8. The method of claim 7, wherein the step of training the artificial neural network based on at least one clinical parameter comprises:

selecting the at least one clinical parameter from the group consisting essentially of;

age and gender.

9. A system for analyzing a nodule, comprising:

a mechanism configured to obtain an image of the nodule;

a mechanism configured to obtain a digital outline of the nodule;

a mechanism configured to generate objective measures corresponding to physical features of the nodule based on the outline, said physical measures selected from the group consisting essentially of;

effective diameter, degree of circularity, degree of ellipticity, root mean square variation, first moment of power spectrum of a function corresponding to the distance between a calculated ellipse and the outline, degree of irregularity, average gradient, radial gradient index, tangential gradient index, line enhancement index, average pixel value, and standard deviation of pixel values;

a mechanism configured to apply the generated objective measures to an artificial neural network; and

a mechanism configured to generate objective measures corresponding to physical features of the training nodule based on the outline;

a mechanism configured to apply the generated objective measures to an artificial neural network; and

a mechanism configured to train the artificial neural network to determine a likelihood of malignancy in a candidate nodule, based on the objective measures.

10. The system of claim 9, wherein:

the mechanism configured to obtain an image of the nodule comprises a mechanism configured to obtain a digital image of the nodule; and

the mechanism configured to obtain the digital outline comprises a mechanism configured to extract the outline of the nodule from the digital image.

11. The system of claim 9, wherein the mechanism configured to apply the generated objective measures to the artificial neural network comprises:

a mechanism configured to apply to the artificial neural network at least one clinical parameter corresponding to the nodule.

12. The system of claim 11, wherein the mechanism configured to apply the at least one clinical parameter comprises:

a mechanism configured to select the at least one clinical parameter from the group consisting essentially of;

age and gender.

13. A system for training an artificial neural network to analyze a candidate nodule, comprising:

a mechanism configured to obtain an image of a training nodule;

a mechanism configured to obtain a digital outline of a the training nodule;

a mechanism configured to generate objective measures corresponding to physical features of the training nodule based on the outline, said physical features selected from the group consisting essentially of;

effective diameter, degree of circularity, degree of ellipticity, root mean square variation, first moment of power spectrum of a function corresponding to the distance between a calculated ellipse and the outline, degree of irregularity, average gradient, radial gradient index, tangential gradient index, line enhancement index, average pixel value, and standard deviation of pixel values;

a mechanism configured to apply the generated objective measures to an artificial neural network; and

a mechanism configured to train the artificial neural network to determine a likelihood of malignancy in a candidate nodule, based on the objective measures.

14. The system of claim 13, wherein:

the mechanism configured to obtain an image of the training nodule comprises a mechanism configured to obtain a digital image of the training nodule; and

the mechanism configured to obtain the digital outline comprises a mechanism configured to extract the outline of the training nodule from the digital image.

15. The system of claim 13, wherein the mechanism configured to train the artificial neural network comprises:

a mechanism configured to train the artificial neural network based on at least one clinical parameter corresponding to the training nodule.

16. The system of claim 15, wherein the mechanism configured to train the artificial neural network based on at least one clinical parameter comprises:

a mechanism configured to select the at least one clinical parameter from the group consisting essentially of;

age and gender.

17. A computer readable medium storing computer instructions for analyzing a nodule, by performing the steps of:

obtaining an image of the nodule;

obtaining a digital outline of the nodule;

generating objective measures corresponding to physical features of the nodule based on the outline, said physical features selected from the group consisting essentially of effective diameter, degree of circularity, degree of ellipticity, root mean square variation, first moment of power spectrum of a function corresponding to the distance between a calculated ellipse and the outline, degree of irregularity, average gradient, radial gradient index, tangential gradient index, line enhancement index, average pixel value, and standard deviation of pixel values;

applying the generated objective measures to an artificial neural network; and

determining a likelihood of malignancy of the nodule based on an output of the artificial neural network.

18. The computer readable medium of claim 17, wherein:

said step of obtaining an image of the nodule comprises obtaining a digital image of the nodule; and

said step of obtaining an outline of the nodule comprises extracting the outline of the nodule from the digital image.

19. The computer readable medium of claim 17, wherein the applying step further comprises:

applying to the artificial neural network at least one clinical parameter corresponding to the nodule.

20. The computer readable medium of claim 19, wherein the step of applying the at least one clinical parameter comprises:

selecting the at least one clinical parameter from the group consisting essentially of;

age and gender.

21. A computer readable medium storing computer instructions for training an artificial neural network to analyze a candidate nodule, by performing the steps of:

obtaining an image of a training nodule;

obtaining a digital outline of the training nodule;

generating objective measures corresponding to physical features of the training nodule based on the outline, said physical features selected from the group consisting essentially of effective diameter, degree of circularity, degree of ellipticity, root mean square variation, first moment of power spectrum of a function corresponding to the distance between a calculated ellipse and the outline, degree of irregularity, average gradient, radial gradient index, tangential gradient index, line enhancement index, average pixel value, and standard deviation of pixel values;

applying the generated objective measures to an artificial neural network; and

training the artificial neural network to determine a likelihood of malignancy in a candidate nodule, based on the objective measures.

22. The computer readable medium of claim 21, wherein:

the step of obtaining an image of a training nodule comprises obtaining a digital image of the training nodule; and

the step of obtaining an outline of the training nodule comprises extracting the outline of the training nodule from the digital image.

23. The computer readable medium of claim 21, wherein the training step further comprises:

training the artificial neural network based on at least one clinical parameter corresponding to the training nodule.

24. The computer readable medium of claim 23, wherein the step of training the artificial neural network based on at least one clinical parameter comprises:

selecting the at least one clinical parameter from the group consisting essentially of;

age and gender.