Tissue surface roughness quantification based on image data and determination of a presence of disease based thereon

1. An image data processor, comprising:

a surface roughness quantifier that is configured to generate a roughness metric that quantifies a roughness of a surface of a tissue of interest in 3D image data based on a surface model adapted to the tissue of interest in the 3D image data, wherein generating the roughness metric comprises:

calculating a position variation of a pre-defined contrast gradient with respect to a mean surface position along a direction normal to the model surface; and

determining an integral of a local position variance of the gradient, normalized to a total tissue of interest surface; and

a decision component that is configured to generate a value signal indicating a presence or an absence of disease in the tissue of interest based on the roughness metric, wherein generating the value signal comprises:

comparing the roughness metric to a predetermined threshold;

generating a first value of the value signal if the roughness metric exceeds the predetermined threshold; and

generating a second value of the value signal if the roughness metric does not exceed the predetermined threshold.

2. The image data processor of claim 1 , wherein at least one of the first value and the second value indicates a stage of the disease.

3. The image data processor of claim 1 , wherein the mean surface position is determined by an initial surface model adaptation.

4. The image data processor of claim 1 , wherein the position variation pattern indicates a stage of the disease.

5. The image data processor of claim 1 , wherein the surface roughness quantifier is further configured to determine one or more of a spatial position variation of at least one of an amplitude, a wavelength, a frequency or a distribution of the amplitude on the surface as a function of at least one of wavelength or frequency, and the decision component is further configured to compare the spatial position variation to a corresponding predetermined threshold to make the determination.

6. The image data processor of claim 1 , wherein surface roughness quantifier is further configured to compare the adapted surface model to predetermined patterns and identify a pattern with the strongest correlation to the adapted surface model, and the decision component is further configured to compare the identified pattern to a pattern-to-stage map and generate a value signal indicative of the mapping, wherein the mapping indicates whether the disease is present or absent.

7. The image data processor of claim 6 , wherein the identified pattern indicates a stage of the disease.

8. The image data processor of claim 1 , wherein the model includes a pre-defined triangulated surface, and the surface roughness quantifier is further configured to adapt the model to the surface of the tissue of interest based on an external energy term and an internal energy term.

9. The image data processor of claim 8 , wherein the external energy term is based on images features and the internal energy term includes a predefined shape of the tissue of interest.

10. The image data processor of claim 1 , wherein the pre-defined triangulated surface is trained on medical image data sets from different modalities.

11. A method, comprising:

generating a roughness metric that quantifies a roughness of a surface of a tissue of interest in 3D image data based on a surface model adapted to the tissue of interest in the 3D image data; and

generating a value signal indicating a presence or an absence of disease in the tissue of interest based on the roughness metric,

wherein the model includes a pre-defined triangulated surface which is adapted to the surface of the tissue of interest, and generating the roughness metric comprises:

calculating a position variation of a pre-defined contrast gradient with respect to a mean surface position along a direction normal to the model surface; and

determining an integral of a local position variance of the gradient, normalized to a total tissue of interest surface, thereby generating the roughness metric; and

wherein generating the value signal comprises:

comparing the roughness metric to a predetermined threshold;

generating a first value of the value signal if the roughness metric exceeds the predetermined threshold; and

generating a second value of the value signal if the roughness metric does not exceed the predetermined threshold.

12. The method of claim 11 , wherein generating the roughness metric and the value signal do not include using a statistical model in which the roughness is a random variable.

13. The method of claim 11 , further comprising:

increasing a number of surface triangles of the model, creating a refined model; and

comparing the refined model to the model to determine a more localized metric.

14. The method of claim 11 , further comprising:

determining a spatial position variation of at least one of a wavelength, a frequency of a distribution of the roughness; and

comparing the spatial position variation to a corresponding predetermined threshold to determine the presence or absence of the disease.

15. The method of claim 11 , further comprising:

comparing the adapted surface model to predetermined patterns;

identifying a pattern of the predetermined patterns with a strongest correlation to the adapted surface model;

comparing the identified pattern to a pattern-to-stage map; and

identifying a map corresponding to the identified pattern, wherein the map indicates the presence or absence of the disease.

16. The method of claim 15 , wherein the map indicates a stage of the disease.