Non-invasive image analysis techniques for diagnosing diseases

1. A device comprising:

an image acquiring module adapted to receive an image comprising at least a portion of animal or human tissue;

a delineation module adapted to indicate an analysis zone in said acquired image;

a feature extractor module adapted to extract quantitative information from said analysis zone;

a machine learning module adapted to receive said extracted information and apply at least one detection algorithm to assess a condition of said tissue,

wherein the feature extractor module comprises at least a rotation compensation module to compensate for the rotation of the analysis zone,

wherein the feature extractor module is further arranged to extract quantitative information corresponding to at least one of the following:

characteristics calculated from a co-occurrence matrix of said analysis zone;

first and second order statistical characteristics of the analysis zone;

characteristics obtained by gradients of the analysis zone in one of a whole of or in portions thereof;

characteristics obtained by cascading;

a characteristic orientation and a local phase of at least a portion of the analysis zone

wherein the feature extractor module is adapted to simultaneously extract quantitative information corresponding to a plurality of characteristics,

wherein the machine learning module is arranged to select from a plurality of algorithms depending on the characteristics used by the feature extractor module, and

wherein the machine learning module is further arranged to combine a plurality of algorithms to assess said condition of said tissue.

2. The device according to claim 1 , wherein, for a specific condition to be assessed, the image acquiring module is adapted to receive images corresponding to a particular anatomical plane.

3. The device according to claim 2 , wherein the image acquiring module is adapted to detect predefined landmarks in the received images to determine if the received image corresponds to the particular anatomical plane.

4. The device according to claim 1 , wherein the delineation module comprises a drawing module adapted to allow a user of the device to mark the boundary of the analysis zone.

5. The device according to claim 4 , wherein said drawing module comprises a free-hand module to allow the user to manually indicate said boundary.

6. The device according to claim 4 , wherein said drawing module comprises a polygon module to allow the user to indicate said boundary.

7. The device according to claim 1 , wherein said characteristics calculated from a co-occurrence matrix of said analysis zone are selected from a list including first order statistics, an angular second moment, a contrast, a correlation, an energy and an entropy of said analysis zone.

8. The device according to claim 1 , wherein said first and second statistical characteristics of the analysis zone are selected from a list including a mean value, a variance, a standard deviation, a skew and a kurtosis of the analysis zone.

9. The device according to claim 1 , wherein the machine learning module comprises a memory for storing quantitative information corresponding to characteristics of a plurality of images corresponding to said condition.

10. A method of assessing a condition of at least a portion of an animal or human tissue, comprising:

receiving an image of said at least one portion of animal or human tissue;

indicating an analysis zone in said received image;

extracting quantitative information from said analysis zone;

applying a machine learning algorithm to said extracted quantitative information to assess the condition of said tissue;

wherein said extracting quantitative information comprises at least compensating for a rotation of the analysis zone,

wherein said extracting quantitative information comprises extracting quantitative information corresponding to at least one of the following:

characteristics calculated from a co-occurrence matrix of said analysis zone;

first and second order statistical characteristics of the analysis zone;

characteristics obtained by gradients of the analysis zone either in the whole of or in portions thereof;

characteristics obtained by cascading;

a characteristic orientation and a local phase of at least a portion of the analysis zone,

wherein said extracting quantitative information comprises simultaneously extracting quantitative information corresponding to a plurality of characteristics, and

further comprising selecting from a plurality of machine learning algorithms depending on the extracted characteristics and combining a plurality of algorithms to assess said condition of said tissue.

11. The method according to claim 10 , wherein, for a specific condition to be assessed, the received image corresponds to a particular anatomical plane.

12. The method according to claim 10 , further comprising selecting said characteristics calculated from a co-occurrence matrix from a list including first order statistics, an angular second moment, a contrast, a correlation, an energy and an entropy of said analysis zone.

13. The method according to claim 10 , further comprising selecting said first and second order statistical characteristics of the analysis zone from a list including a mean value, a variance, a standard deviation, a skew and a kurtosis of the analysis zone.

14. A method of diagnosing a pathological condition of at least a portion of an animal or human tissue, comprising:

receiving an image of said at least one portion of animal or human tissue;

indicating an analysis zone in said received image;

extracting quantitative information from said analysis zone;

applying a machine learning algorithm to said extracted quantitative information to assess the condition of said tissue;

wherein said extracting quantitative information comprises compensating for a rotation of the analysis zone,

wherein, if the extracted quantitative information corresponds to stored quantitative information belonging to animal or human tissue of said pathological condition, then said portion of animal or human tissue is diagnosed of said pathological condition, and

wherein said extracting quantitative information comprises simultaneously extracting quantitative information corresponding to a plurality of characteristics.

15. The method according to claim 14 , further comprising selecting from a plurality of machine learning algorithms depending on the extracted characteristics.

16. The method according to claim 15 , further comprising combining a plurality of algorithms to assess said condition of said tissue.

17. The method according to claim 14 , further comprising storing quantitative information corresponding to characteristics of a plurality of images corresponding to said condition.