Risk assessment based on patient similarity determined using image analysis

• US 10,283,221 B2
• Filed: 10/27/2016
• Issued: 05/07/2019
• Est. Priority Date: 10/27/2016
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Abedini, Mani, Bozorgtabar, Seyedbehzad, Chakravorty, Rajib, Garnavi, Rahil
• Primary Examiner(s): Coleman, Stephen P
• Application Number:
  US15/336,252
• Publication Number:
  US 20180121626A1
• Time in Patent Office: 922 Days
• Field of Search: None
• 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 7/0014 using an image reference approach
• CPC Class Codes: G16H 30/40 for processing medical images, e.g. editing
• CPC Class Codes: G16H 50/30 for calculating health indices; for individual health risk assessment
• CPC Class Codes: G16H 50/70 for mining of medical data, e.g. analysing previous cases of other patients

Abstract

A method for risk assessment comprises receiving one or more images of a plurality of lesions captured from a body of a target person, generating one or more digital signatures based on the one or more images from the body of the target person, comparing the generated one or more digital signatures to digital signatures of respective reference persons, wherein the comparing comprises measuring similarities between the generated one or more digital signatures and the digital signatures of the respective reference persons, and determining a risk factor for the target person of developing a disease based on the measured similarities and predetermined risk factors of developing the disease for the reference persons.

Claims

1. A method for risk assessment, comprising:

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

generating one or more digital signatures based on the one or more images from the body of the target person;

comparing the generated one or more digital signatures to digital signatures of respective reference persons, wherein the comparing comprises measuring similarities between the generated one or more digital signatures and the digital signatures of the respective reference persons;

determining a risk factor for the target person of developing a disease based on the measured similarities and predetermined risk factors of developing the disease for the reference persons;

separating the reference persons into a plurality of groups based on one or more characteristics of the reference persons; and

training an auto encoder for each group of the plurality of groups;

wherein generating the one or more digital signatures comprises processing the one or more images from the body of the target person with each trained auto encoder to generate respective codes corresponding to each trained auto encoder;

wherein the processing comprises;

encoding the one or more images from the body of the target person into one or more vectors each having a predetermined size; and

digitally reconstructing the encoded one or more images into one or more reconstructed images; and

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 training is performed using lesion images of the reference persons for each group of the plurality of groups.

3. The method according to claim 1, wherein the one or more characteristics are selected from the group comprising age, gender, race, geographic location, behavior, and family history.

4. The method according to claim 1, wherein a digital signature of the one or more digital signatures comprises a matrix of the respective codes.

5. The method according to claim 1, wherein the processing with each trained auto encoder is performed separately for images from different regions on the body of the target person.

6. The method according to claim 1, wherein a digital signature of the one or more digital signatures corresponds to a region on the body of the target person.

7. The method according to claim 1, wherein determining the risk factor for the target person of developing the disease comprises calculating a summation of the predetermined risk factors for each reference person, wherein each of the predetermined risk factors is adjusted based on a value for similarity between the generated one or more digital signatures of the target person and one or more digital signatures of each reference person.

8. The method according to claim 1, wherein measuring the similarities between the generated one or more digital signatures and the digital signatures of the respective reference persons is performed using sparse coding.

9. The method according to claim 8, further comprising defining the generated one or more digital signatures of the target person as a linear combination of the digital signatures of the respective reference persons.

10. A system for risk assessment, 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 target person;

generate one or more digital signatures based on the one or more images from the body of a target person;

compare the generated one or more digital signatures to digital signatures of respective reference persons, wherein the processor is further configured to measure similarities between the generated one or more digital signatures and the digital signatures of the respective reference persons;

determine a risk factor for the target person of developing a disease based on the measured similarities and predetermined risk factors of developing the disease for the reference persons;

separate the reference persons into a plurality of groups based on one or more characteristics of the reference persons; and

train an auto encoder for each group of the plurality of groups;

wherein in generating the one or more digital signatures, the processor is further configured to process the one or more images from the body of the target person with each trained auto encoder to generate respective codes corresponding to each trained auto encoder;

wherein in processing with each trained auto encoder, the processor is further configured to;

encode the one or more images from the body of the target person into one or more vectors each having a predetermined size; and

digitally reconstruct the encoded one or more images into one or more reconstructed images.

11. The system according to claim 10, wherein a digital signature of the one or more digital signatures comprises a matrix of the respective codes.

12. The system according to claim 10, wherein the processing with each trained auto encoder is performed separately for images from different regions on the body of the target person.

13. The system according to claim 10, wherein a digital signature of the one or more digital signatures corresponds to a region on the body of the target person.

14. The system according to claim 10, wherein:

measuring the similarities between the generated one or more digital signatures and the digital signatures of the respective reference persons is performed using sparse coding; and

the processor is further configured to define the generated one or more digital signatures of the target person as a linear combination of the digital signatures of the respective reference persons.

15. The system according to claim 10, wherein in determining the risk factor for the target person of developing the disease, the processor is further configured to calculate a summation of the predetermined risk factors for each reference person, wherein each of the predetermined risk factors is adjusted based on a value for similarity between the generated one or more digital signatures of the target person and one or more digital signatures of each reference person.

16. A computer program product for risk assessment, the computer program product comprising a 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 target person;

generating one or more digital signatures based on the one or more images from the body of the target person;

comparing the generated one or more digital signatures to digital signatures of respective reference persons, wherein the comparing comprises measuring similarities between the generated one or more digital signatures and the digital signatures of the respective reference persons;

determining a risk factor for the target person of developing a disease based on the measured similarities and predetermined risk factors of developing the disease for the reference persons;

separating the reference persons into a plurality of groups based on one or more characteristics of the reference persons; and

training an auto encoder for each group of the plurality of groups;

wherein generating the one or more digital signatures comprises processing the one or more images from the body of the target person with each trained auto encoder to generate respective codes corresponding to each trained auto encoder;

wherein the processing comprises;

encoding the one or more images from the body of the target person into one or more vectors each having a predetermined size; and

digitally reconstructing the encoded one or more images into one or more reconstructed images.