Claims
- 1. A method of detecting a candidate abnormality in a sonographic medical image, comprising the steps of:
obtaining a sonographic medical image; determining a radial gradient index (RGI) at plural pixels to produce an RGI image; thresholding the RGI image; determining a candidate abnormality in the RGI image based on the thresholding step; and locating a center point of the candidate abnormality in the RGI image and a corresponding center point in the sonographic image.
- 2. The method of claim 1, further comprising:
segmenting the candidate abnormality in the sonographic image based on the center point in the sonographic image, including determining average radial gradients (ARDs) in the sonographic medical image.
- 3. The method of claim 2, further comprising:
extracting plural features from the segmented candidate abnormality; and determining a likelihood of the candidate abnormality being an actual abnormality based on the extracted plural features.
- 4. The method of claim 1, wherein the thresholding step comprises:
iteratively thresholding the RGI image.
- 5. The method of claim 4, wherein the extracting step comprises:
extracting at least four features from the group comprising candidate abnormality shape, margin sharpness, echogenic texture, RGI value, posterior acoustic enhancement, and shadowing.
- 6. The method of claim 3, wherein the determining a likelihood step comprises:
using a classifier; and determining a likelihood of the candidate abnormality being an actual abnormality based on the output of the classifier.
- 7. The method of claim 6, wherein the using step comprises:
using a Bayesian Neural Network (BNN).
- 8. The method of claim 7, wherein the using step comprises:
using a BNN having between 3 and 7 hidden layers.
- 9. The method of claim 7, wherein the using step comprises:
using a BNN having 7 layers, comprised of an input layer, and output layer, and 5 hidden layers.
- 10. The method of claim 6, wherein the using step comprises:
using an artificial neural network (ANN).
- 11. The method of claim 6, wherein the using step comprises:
using linear discriminants.
- 12. A system implementing the method of any one of claims 1 through 11.
- 13. A computer program product storing program instructions, which when executed by a computer system, cause the computer system to perform the method recited in any one of claims 1 through 11.
CROSS-REFERENCE TO RELATED APPLICATION
[0001] The present application claims priority to U.S. Provisional Patent Application 60/332,005 filed Nov. 23, 2001, entitled “Automated Method and System for the Detection of Abnormalities in Sonographic Images.”
Government Interests
[0002] The present invention was made in part with U.S. Government support under grant number CA89452 and T31 CA09649 from the USPHS, and U.S. Army Medical Research and Materiel Command 97-2445.
Provisional Applications (1)
|
Number |
Date |
Country |
|
60332005 |
Nov 2001 |
US |