Detection of outlier lesions based on extracted features from skin images

• US 10,242,442 B2
• Filed: 10/27/2016
• Issued: 03/26/2019
• Est. Priority Date: 10/27/2016
• Status: Active Grant

• Current Assignee: International Business Machines Corporation, MoleMap NZ, Ltd.
• Original Assignee: International Business Machines Corporation
• Inventors: Abedini, Mani, Bowling, Adrian, Chakravorty, Rajib, Demyanov, Sergey, Garnavi, Rahil
• Primary Examiner(s): Coleman, Stephen P
• Application Number:
  US15/336,293
• Publication Number:
  US 20180122065A1
• Time in Patent Office: 880 Days
• Field of Search: None
• CPC Class Codes: G06T 2207/10016 Video; Image sequence
• CPC Class Codes: G06T 2207/10024 Color image
• CPC Class Codes: G06T 2207/20081 Training; Learning
• CPC Class Codes: G06T 2207/30088 Skin; Dermal
• CPC Class Codes: G06T 2207/30096 Tumor; Lesion
• CPC Class Codes: G06T 7/0012 Biomedical image inspection
• CPC Class Codes: G06T 7/0016 involving temporal comparison
• CPC Class Codes: G06T 7/73 using feature-based methods
• CPC Class Codes: G06V 10/50 by performing operations within image blocks; by using histograms, e.g. histogram of oriented gradients [HoG]; by summing image-intensity values; Projection analysis
• CPC Class Codes: G06V 10/758 Involving statistics of pixels or of feature values, e.g. histogram matching
• CPC Class Codes: G06V 2201/03 Recognition of patterns in medical or anatomical images

Abstract

A method for image analysis comprises receiving one or more images of a plurality of lesions captured from a body of a person, extracting one or more features of the plurality of lesions from the one or more images, analyzing the extracted one or more features, wherein the analyzing comprises determining a distance between at least two lesions with respect to the extracted one or more features, and determining whether any of the plurality of lesions is an outlier based on the analyzing.

Claims

1. A method for image analysis, comprising:

receiving one or more images of a plurality of lesions captured from a body of a person;

extracting one or more features of the plurality of lesions from the one or more images;

analyzing the extracted one or more features;

wherein the analyzing comprises determining a distance between at least two lesions with respect to the extracted one or more features; and

determining whether any of the plurality of lesions is an outlier based on the analyzing;

wherein the extracting comprises;

converting an image of a lesion of the plurality of lesions into a matrix comprising the image;

applying each of a plurality clinical pattern classifiers to the matrix, wherein each clinical pattern classifier corresponds to a clinical feature of a plurality of clinical features;

quantifying an amount of correspondence with each clinical feature of the plurality of clinical features at one or more locations in a plurality of output matrices respectively corresponding to each clinical pattern classifier; and

converting the image of the lesion into a clinical based image representation representing one or more structural properties of the lesion, wherein the converting comprises merging the quantified amounts from each of the plurality of output matrices to digitally transform the image of the lesion into the clinical based image representation;

wherein the method is performed by at least one computer system comprising at least one memory and at least one processor coupled to the memory.

2. The method according to claim 1, wherein the one or more features comprise at least one visual feature, and the method further comprises generating a histogram based on the at least one visual feature.

3. The method according to claim 1, wherein the one or more features comprise at least one clinical feature selected from the group consisting of lesion size, lesion shape, lesion symmetry, lesion asymmetry, lesion boundaries and lesion color.

4. The method according to claim 1, wherein the extracting further comprises:

converting the clinical based image representation into a feature vector.

5. The method according to claim 1, wherein determining the distance between the at least two lesions comprises calculating at least one of a Euclidean distance and a divergence measure.

6. The method according to claim 1, further comprising applying a weight to the distance between the at least two lesions, wherein the weight is based on a proximity of the at least two lesions with respect to each other.

7. The method according to claim 1, wherein the one or more images and the extracting of the one or more features from the one or more images are associated with a first time, and the method further comprises:

receiving one or more additional images of the plurality of lesions captured from the body of the person;

extracting the one or more features of the plurality of lesions from the one or more additional images, wherein the one or more additional images and the extracting of the one or more features from the one or more additional images are associated with a second time; and

determining a change in the extracted one or more features between the first time and the second time.

8. The method according to claim 1, wherein determining whether any of the plurality of lesions is an outlier is performed using a representative distance method.

9. The method according to claim 1, further comprising creating a lesion signature for the person based on the analyzing.

10. The method according to claim 1, further comprising applying a weight to the distance between the at least two lesions, wherein the weight is based on an importance of the extracted one or more features.

11. The method according to claim 10, wherein the importance of the extracted one or more features is determined by using at least one machine learning technique.

12. A system for image analysis, comprising:

a memory and at least one processor coupled to the memory, wherein the at least one processor is configured to;

receive one or more images of a plurality of lesions captured from a body of a person;

extract one or more features of the plurality of lesions from the one or more images;

analyze the extracted one or more features;

wherein the analyzing comprises determining a distance between at least two lesions with respect to the extracted one or more features; and

determine whether any of the plurality of lesions is an outlier based on the analyzing;

wherein in performing the extracting the processor is configured to;

convert an image of a lesion of the plurality of lesions into a matrix comprising the image;

apply each of a plurality clinical pattern classifiers to the matrix, wherein each clinical pattern classifier corresponds to a clinical feature of a plurality of clinical features;

quantify an amount of correspondence with each clinical feature of the plurality of clinical features at one or more locations in a plurality of output matrices respectively corresponding to each clinical pattern classifier; and

convert the image of the lesion into a clinical based image representation representing one or more structural properties of the lesion, wherein the converting comprises merging the quantified amounts from each of the plurality of output matrices to digitally transform the image of the lesion into the clinical based image representation.

13. The system according to claim 12, wherein the one or more features comprise at least one visual feature, and the processor is further configured to generate a histogram based on the at least one visual feature.

14. The system according to claim 12, wherein the one or more features comprise at least one clinical feature selected from the group consisting of lesion size, lesion shape, lesion symmetry, lesion asymmetry, lesion boundaries and lesion color.

15. The system according to claim 12, wherein in performing the extracting the processor is further configured to:

convert the clinical based image representation into a feature vector.

16. The system according to claim 12, wherein in determining the distance between the at least two lesions the processor is configured to calculate at least one of a Euclidean distance and a divergence measure.

17. The system according to claim 12, wherein the processor is further configured to apply a weight to the distance between the at least two lesions, wherein the weight is based on a proximity of the at least two lesions with respect to each other.

18. The system according to claim 12, wherein the processor is further configured to apply a weight to the distance between the at least two lesions, wherein the weight is based on an importance of the extracted one or more features.

19. A computer program product for image analysis, the computer program product comprising a non-transitory computer readable storage medium having program instructions embodied therewith, the program instructions executable by a processor to cause the processor to perform a method comprising:

receiving one or more images of a plurality of lesions captured from a body of a person;

extracting one or more features of the plurality of lesions from the one or more images;

analyzing the extracted one or more features;

wherein the analyzing comprises determining a distance between at least two lesions with respect to the extracted one or more features; and

determining whether any of the plurality of lesions is an outlier based on the analyzing;

wherein the extracting comprises;

converting an image of a lesion of the plurality of lesions into a matrix comprising the image;

applying each of a plurality clinical pattern classifiers to the matrix, wherein each clinical pattern classifier corresponds to a clinical feature of a plurality of clinical features;

quantifying an amount of correspondence with each clinical feature of the plurality of clinical features at one or more locations in a plurality of output matrices respectively corresponding to each clinical pattern classifier; and

converting the image of the lesion into a clinical based image representation representing one or more structural properties of the lesion, wherein the converting comprises merging the quantified amounts from each of the plurality of output matrices to digitally transform the image of the lesion into the clinical based image representation.