Noise suppression in diagnostic images
View Patent ↗A method for suppressing noise in a diagnostic image executes one or more iterations of segmenting the image to identify and label one or more regions in the image; and performing selective diffusion on at least one of the one or more labeled regions in the image. A homogeneity value is computed for the region. A diffusion conductance function is generated for the region according to an intensity gradient between adjacent digital image elements within the region. The diffusion process is applied to a plurality of digital image elements within the region.
1. A method for suppressing noise in a diagnostic image, the method comprising one or more iterations of:
segmenting the image to identify and label one or more regions in the image; and
performing selective diffusion on at least one of the one or more labeled regions in the image by:
computing a homogeneity value for the region; and
generating a diffusion conductance function for the region according to an intensity gradient between adjacent digital image elements within the region and applying the diffusion process to a plurality of digital image elements within the region.
2. The method of claim 1 wherein segmenting the image comprises comparing digital image element values within the image against a predetermined threshold and labeling one or more regions according to the comparison.
3. The method of claim 1 wherein segmenting the image comprises using 3-D model-based information.
4. The method of claim 1 wherein computing a homogeneity value for the region comprises computing an average value of magnitude of intensity differences or gradient for two or more digital image elements within the region.
5. The method of claim 4 further comprising computing a standard deviation for the magnitude of intensity differences or gradient for pairs of digital image elements within the region.
6. The method of claim 1 wherein the diagnostic image is a 2-D image.
7. The method of claim 1 wherein the diagnostic image is a 3-D image.
8. The method of claim 1 wherein the diagnostic image is a 4-D image or higher dimension image.
9. The method of claim 1 wherein the diagnostic image is reconstructed from a plurality of 2-D images.
10. The method of claim 1 wherein applying the diffusion conductance function further comprises reducing the magnitude of intensity differences or gradient for pairs of digital image elements within the region.
11. The method of claim 1 further comprising displaying the processed diagnostic image on a display screen.
12. The method of claim 1 wherein generating the diffusion conductance function comprises generating an anisotropic diffusion function.
13. A method for suppressing noise in a 3-D diagnostic image, the method comprising one or more iterations of:
forming the 3-D diagnostic image from two or more separate images;
segmenting the 3-D diagnostic image to identify and label one or more 3-D regions in the image; and
performing selective diffusion on at least one of the one or more labeled 3-D regions in the image by:
computing a homogeneity value for the 3-D region; and
generating an anisotropic diffusion conductance function for the 3-D region according to an intensity gradient between adjacent digital image elements within the region and applying the diffusion process to a plurality of digital image elements within the region.
14. The method of claim 13 wherein segmenting comprises grouping adjacent image voxels according to their relative gradient intensity values.
15. The method of claim 13 wherein the 3-D diagnostic image is formed from two or more computed tomography scans.
16. A method for suppressing noise in a diagnostic image, the method comprising:
segmenting the diagnostic image to identify and label one or more regions in the diagnostic image;
performing selective diffusion on at least one of the one or more labeled regions in the diagnostic image to generate a filtered image, wherein the selective diffusion comprises:
computing a homogeneity value for the at least one of the one or more labeled regions; and
generating a diffusion conductance function for the at least one of the one or more labeled regions according to an intensity gradient between adjacent digital image elements within the at least one of the one or more labeled regions and applying the diffusion process to a plurality of digital image elements within the at least one of the one or more labeled regions;
segmenting the filtered image to identify and label one or more regions in the filtered image;
performing the selective diffusion on at least one of the one or more labeled regions in the filtered image.