Image processing used to estimate abnormalities

1. An image processing apparatus comprising:

a computer system comprising:

an image acquiring section that acquires an image of body tissue;

an extracting section that extracts an outline of the body tissue from the image, wherein the outline comprises an ordered plurality of outline points beginning with a reference point, and wherein the reference point and each outline point of the plurality of outline points is defined with a pair of coordinates;

a sequence generating section that generates an object sequence indicating a distance to the reference point for each of the remaining ordered plurality of outline points, wherein each distance is a linear measure from at least one coordinate of an outline point to at least one corresponding coordinate of the reference point;

a sequence converting section that converts the object sequence into a value sequence using a series expansion; and

an estimating section that estimates an abnormal shape of the body tissue by processing the value sequence using a neural network.

2. The image processing apparatus according to claim 1 , wherein the sequence generating section calculates the object sequence to be relative coordinate values that are relative to the reference point in a one-dimensional direction for each of the remaining ordered plurality of outline points.

3. The image processing apparatus according to claim 1 , wherein the sequence generating section supplies the sequence converting section with a scaled object sequence obtained by scaling a length of the object sequence to be a reference length.

4. The image processing apparatus according to claim 1 , wherein the series expansion ignores fluctuations caused by relatively large regions occurring in the object sequence and extracts fluctuations occurring in small regions.

5. The image processing apparatus according to claim 1 , wherein the series expansion comprises a wavelet decomposition.

6. The image processing apparatus according to claim 5 , wherein the wavelet decomposition is applied to the object sequence a plurality of times.

7. The image processing apparatus according to claim 5 , wherein the sequence converting section supplies a neural network with the value sequence based on a wavelet decomposition coefficient obtained by the wavelet decomposition.

8. The image processing apparatus according to claim 1 , further comprising:

a learning processing section that trains the neural network using learning data including a plurality of groups of images of body tissue and abnormal shape judgment results.

9. The image processing apparatus according to claim 1 , wherein:

the image acquiring section acquires a cross-sectional image of a lung as the image of the body tissue; and

the estimating section estimates whether a lung disease in contact with an outline of the lung is present.

10. The image processing apparatus according to claim 9 , wherein:

the extracting section extracts outlines of a right lung and a left lung from the cross-sectional image;

the sequence converting section converts an object sequence generated from the outline of the right lung into a value sequence for the right lung and converts an object sequence generated from the outline of the left lung into a value sequence for the left lung; and

the estimating section estimates an abnormal shape of the right lung by performing processing with a neural network trained for the right lung on the value sequence for the right lung, and estimates an abnormal shape of the left lung by performing processing with a neural network trained for the left lung on the value sequence for the left lung.

11. An image processing method executed by a computer, comprising:

acquiring an image of body tissue;

extracting an outline of the body tissue from the image, wherein the outline comprises an ordered plurality of outline points beginning with a reference point, and wherein the reference point and each outline point of the plurality of outline points is defined with a pair of coordinates;

generating an object sequence indicating a distance to the reference point for each of the remaining ordered plurality of outline points, wherein each distance is a linear measure from at least one coordinate of an outline point to at least one corresponding coordinate of the reference point;

converting the object sequence into a value sequence using a series expansion; and

estimating an abnormal shape of the body tissue by processing the value sequence using a neural network.

12. The method of claim 11 , wherein the distance to the reference point for each of the remaining ordered plurality of outline points is measured in a one-dimensional direction from the reference point to each outline point.

13. The method of claim 11 , wherein the series expansion ignores fluctuations caused by relatively large regions occurring in the object sequence and extracts fluctuations occurring in small regions.

14. The method of claim 11 , wherein the series expansion comprises a wavelet decomposition.

15. The method of claim 11 , further comprising:

training the neural network using learning data including a plurality of groups of images of body tissue and abnormal shape judgment results.

16. A computer program product comprising a computer readable storage medium having computer readable program code stored thereon, wherein the computer readable program code, when executed by a computer, causes the computer to:

acquire an image of body tissue;

extract an outline of the body tissue from the image, wherein the outline comprises an ordered plurality of outline points beginning with a reference point, and wherein the reference point and each outline point of the plurality of outline points is defined with a pair of coordinates;

generate an object sequence indicating a distance to the reference point for each of the remaining ordered plurality of outline points, wherein each distance is a linear measure from at least one coordinate of an outline point to at least one corresponding coordinate of the reference point;

converting the object sequence into a value sequence using a series expansion; and

estimate an abnormal shape of the body tissue by processing the value sequence using a neural network.

17. The computer program product of claim 16 , wherein the distance to the reference point for each of the remaining ordered plurality of outline points is measured in a one-dimensional direction from the reference point to each outline point.

18. The computer program product of claim 16 , wherein the series expansion ignores fluctuations caused by relatively large regions occurring in the object sequence and extracts fluctuations occurring in small regions.

19. The computer program product of claim 16 , wherein the series expansion comprises a wavelet decomposition.

20. The computer program product of claim 16 , further comprising:

training the neural network using learning data including a plurality of groups of images of body tissue and abnormal shape judgment results.