Ultrasound diagnosis apparatus, image processing apparatus, and image processing method

An image processing apparatus comprising:

• decomposing circuitry configured to decompose medical image data into low-frequency decomposed image data and high-frequency decomposed image data at each of a predetermined number of hierarchical levels, by performing a hierarchical multi-resolution analysis;
• diffusion filter circuitry configured to apply, at a lowest hierarchical level of the predetermined number of hierarchical levels, a non-linear anisotropic diffusion filter to the low-frequency decomposed image data and the high-frequency decomposed image data at the lowest hierarchical level, configured to apply, at each of the hierarchical levels higher than the lowest hierarchical level, a non-linear anisotropic diffusion filter to data output from a hierarchical level immediately underneath that has been reconstructed by performing a multi-resolution analysis and to the high-frequency decomposed image data at that hierarchical level, and configured to generate, for each of the hierarchical levels, edge information of a signal either from the low-frequency decomposed image data at the lowest hierarchical level or from the data output from the hierarchical level immediately underneath;
• adjusting circuitry configured to adjust a signal level of the high-frequency decomposed image data to which the non-linear anisotropic diffusion filter was applied, based on the edge information generated by the diffusion filter circuitry on the same hierarchical level as the hierarchical level of the high-frequency decomposed image data at each of the hierarchical levels;
• and
• reconstructing circuitry configured to:
• reconstruct data by performing the multi-resolution analysis, from the data that has been processed by the diffusion filter circuitry and was not used in the process performed by the adjusting circuitry and the data that has been processed by the adjusting circuitry at the same hierarchical level,
• output the reconstructed data as the output data to be processed by the diffusion filter circuitry at the hierarchical level immediately above, and
• obtain the reconstructed data as corrected data of the ultrasound image data at the highest hierarchical level,
• wherein at the lowest hierarchical level, the diffusion filter circuitry applies the non-linear anisotropic diffusion filter to the high-frequency decomposed image data at the lowest hierarchical level by using a diffusion filter coefficient calculated based on a structure tensor and the edge information detected from the low-frequency decomposed image data at the lowest hierarchical level, whereas at each of the hierarchical levels higher than the lowest hierarchical level, the diffusion filter circuitry applies the non-linear anisotropic diffusion filter to the high-frequency decomposed image data at that hierarchical level by using a diffusion filter coefficient calculated based on a structure tensor and the edge information detected from the output data that is output by the reconstructing circuitry at the hierarchical level immediately underneath.

An image processing method comprising:

• a process performed by decomposing circuitry to decompose medical image data into low-frequency decomposed image data and high-frequency decomposed image data at each of a predetermined number of hierarchical levels, by performing a hierarchical multi-resolution analysis;
• a process performed by diffusion filter circuitry to apply, at a lowest hierarchical level of the predetermined number of hierarchical levels, a non-linear anisotropic diffusion filter to the low-frequency decomposed image data and the high-frequency decomposed image data at the lowest hierarchical level, to apply, at each of the hierarchical levels higher than the lowest hierarchical level, a non-linear anisotropic diffusion filter to data output from a hierarchical level immediately underneath that has been reconstructed by performing a multi-resolution analysis and to the high-frequency decomposed image data at that hierarchical level, and to generate, for each of the hierarchical levels, edge information of a signal either from the low-frequency decomposed image data at the lowest hierarchical level or from the data output from the hierarchical level immediately underneath;
• a process performed by adjusting circuitry to adjust a signal level of the high-frequency decomposed image data to which the non-linear anisotropic diffusion filter was applied, based on the edge information generated by the diffusion filter circuitry on the same hierarchical level as the hierarchical level of the high-frequency decomposed image data at each of the hierarchical levels;
• and
• a process performed by reconstructing circuitry to:
• reconstruct data by performing the multi-resolution analysis, from the data that has been processed by the diffusion filter circuitry and was not used in the process performed by the adjusting circuitry and the data that has been processed by the adjusting circuitry at the same hierarchical level,
• output the reconstructed data as the output data to be processed by the diffusion filter circuitry at the hierarchical level immediately above, and
• obtain the reconstructed data as corrected data of the ultrasound image data at the highest hierarchical level,
• at the lowest hierarchical level, the process performed by the diffusion filter circuitry includes applying the non-linear anisotropic diffusion filter to the high-frequency decomposed image data at the lowest hierarchical level by using a diffusion filter coefficient calculated based on a structure tensor and the edge information detected from the low-frequency decomposed image data at the lowest hierarchical level, whereas at each of the hierarchical levels higher than the lowest hierarchical level, the process performed by the diffusion filter circuitry includes applying the non-linear anisotropic diffusion filter to the high-frequency decomposed image data at that hierarchical level by using a diffusion filter coefficient calculated based on a structure tensor and the edge information detected from the output data that is output by the reconstructing circuitry at the hierarchical level immediately underneath.