Artifact suppression in diagnostic images
View Patent ↗A method of reducing artifacts in a digital radiographic image identifies either a row or column direction for the artifacts in the image data as a predominant direction and obtains a measurement of the image data frequency content that is subject to artifacts from the image content according to the predominant direction. The measurement of image data frequency content subject to the artifacts is tested according to a predetermined threshold. Artifacts are reduced when the predetermined threshold is exceeded by generating one or more suppression factors according to the testing results, decomposing the image content into at least two frequency bands in each row and column direction, applying the one or more suppression factors to modify at least one of the frequency bands, and recomposing the image content by recombining the at least one modified frequency band with one or more other bands into which the image had been decomposed.
1. A method of reducing artifacts in a digital radiographic image comprising:
identifying either a row or column direction for the artifacts in the image data as a predominant direction;
obtaining a measurement of the image data frequency content that is subject to the artifacts from the image content according to the predominant direction;
testing the measurement of image data frequency content subject to the artifacts according to a predetermined threshold; and
reducing the artifacts when the predetermined threshold is exceeded by:
generating one or more suppression factors according to the testing results;
decomposing the image content into at least two frequency bands in each row and column direction;
applying the one or more suppression factors to modify at least one of the frequency bands; and
recomposing the image content by recombining the at least one modified frequency band with one or more other bands into which the image had been decomposed.
2. The method of claim 1 wherein identifying a predominant row or column direction comprises comparing statistical values obtained from row and column image data values.
3. The method of claim 1 wherein obtaining a measurement of the image data frequency content that is subject to the artifacts from the image content comprises conditioning the image data by obtaining the square root of each image data value.
4. The method of claim 1 wherein obtaining a measurement of the image data frequency content subject to the artifacts from the image content comprises performing a functional fit detrending process.
5. The method of claim 4 wherein the detrending process is linear.
6. The method of claim 1 wherein obtaining a measurement of the image data frequency content subject to the artifacts from the image content comprises applying a frequency transform.
7. The method of claim 6 further comprising obtaining logarithm values for the transformed data.
8. The method of claim 1 wherein the predominant row or column direction for the artifacts is predetermined.
9. The method of claim 1 wherein decomposing the image content further comprises:
obtaining a first filtered image having low frequency content in a direction orthogonal to the predominant direction;
filtering the first filtered image in the predominant direction and obtaining a second filtered image having low frequency content in the predominant direction;
filtering the second filtered image having low frequency content in the predominant direction and generating a third filtered image;
subtracting the second filtered image from the first filtered image to obtain a fine predominant direction edge image;
subtracting the third filtered image from the second filtered image to obtain a coarse predominant direction edge image; and
subtracting the first filtered image in the predominant direction from the original image to obtain an edge image in the direction orthogonal to the predominant direction.
10. The method of claim 9 wherein applying one or more suppression factors comprises:
applying a first scaling factor to the fine predominant direction edge image to obtain a scaled fine predominant direction edge image; and
applying a second scaling factor to the coarse predominant direction edge image to obtain a scaled coarse predominant direction edge image.
11. The method of claim 1 wherein decomposing the image content comprises defining the at least two frequency bands according to results from obtaining a measurement of the image data frequency content.
12. The method of claim 1 further comprising applying one or more constraints to the recomposed image content.
13. The method of claim 12 wherein applying one or more constraints further comprises determining the difference between the recomposed image content and the original input image content.
14. The method of claim 1 wherein reducing the artifacts comprises applying the one or more suppression factors to a frequency band in the predominant direction.
15. The method of claim 1 wherein decomposing the image content into at least two frequency bands comprises identifying one or more filter values according to the image data.