System and method for predicting neurological disorders

Patent metadata

Field	Value
Publication number	US-9715622-B2
Application number	US-201514797365-A
Country	US
Kind code	B2
Filing date	Jul 13, 2015
Priority date	Dec 30, 2014
Publication date	Jul 25, 2017
Grant date	Jul 25, 2017

Abstract

A method and system for predicting neurological disorders is provided. The method comprises receiving videos of individuals and detecting Regions of Interest (ROI) in video frames. The method further comprises determining a Motion Vector (MV) for each ROI in a set of successive frames and comparing value of the determined MV with pre-stored values. Furthermore, the method comprises identifying a MV matching a pre-stored value thereby identifying a ROI and a frame corresponding to the identified MV, wherein the pre-stored value indicates onset of an expression. Also, the method comprises determining MVs for the identified ROI in subsequent sets of successive frames and comparing value of the determined MVs with a pre-stored value of MV corresponding to peak and offset of the indicated expression. The method further comprises identifying the frame corresponding to the peak and offset of the indicated expression and generating pictorial representation for predicting neurological disorders.

First claim

We claim:

1. A method for predicting one or more neurological disorders via a processor configured to execute program instructions stored in a memory, the method comprising:
receiving, via the processor, one or more videos of one or more individuals;
splitting, via the processor, the one or more videos into one or more frames;
detecting, via the processor, one or more regions of interest in each of the one or more frames;
analyzing, via the processor, each of the one or more regions of interest in each of the one or more frames by:
determining a motion vector for each of the one or more regions of interest in a set of successive frames; and
comparing value of the determined motion vector for each of the one or more regions of interest with corresponding pre-stored values, wherein the pre-stored values for the motion vector for each of the one or more regions are pre-stored in a database;

identifying, via the processor, at least one motion vector, from the determined motion vectors, that matches with a pre-stored value of a motion vector thereby identifying a region of interest and a frame corresponding to the identified at least one motion vector, wherein the pre-stored value indicates onset of an expression;

determining, via the processor, one or more motion vectors for the identified region of interest in subsequent sets of successive frames;
comparing, via the processor, value of the one or more determined motion vectors for the identified region of interest with at least one of a pre-stored value of motion vector corresponding to peak of an indicated expression and a pre-stored value of motion vector corresponding to offset of the indicated expression, wherein the peak of the indicated expression and the offset of the indicated expression are values in at least one of a graph, a histogram and a table depicting the indicated expression;
identifying, via the processor, the frame corresponding to the at least one of: the peak of the indicated expression and the offset of the indicated expression; and
generating, via the processor, a pictorial representation of the one or more videos depicting at least one of: the onset, peak and offset of the indicated expression of the one or more individuals captured in the one or more videos for predicting the one or more neurological disorders.

2. The method of claim 1 , wherein the one or more expressions include at least one of: happy, content, sad, disgust, surprise, clueless and angry.

3. The method of claim 1 , wherein the detected one or more regions of interest include at least one of: eyes, cheeks, nose, lips, ears, eyebrows, hands, arms, torso, legs and feet.

4. The method of claim 1 , wherein the regions of interest are detected using Viola-Jones algorithm.

5. The method of claim 1 , wherein the motion vector for each of the one or more regions of interest in a set of successive frames is determined using optical flow algorithm.

6. The method of claim 5 , wherein the optical flow algorithm uses Horn-Schunck method for determining the motion vector.

7. The method of claim 1 , wherein the one or more individuals are speaking while the one or more videos are being captured.

8. A system for predicting one or more neurological disorders, the system comprising:
a video acquisition module configured to:
receive one or more videos of one or more individuals; and
split the one or more videos into one or more frames;

a region of interest detection module configured to detect one or more regions of interest in each of the one or more frames;
a video processing module configured to analyze each of the one or more detected regions of interest in each of the one or more frames, wherein the video processing module comprises:
a feature extraction module configured to determine a motion vector for each of the one or more regions of interest in a set of successive frames; and
a comparator configured to compare value of the determined motion vector for each of the one or more regions of interest with corresponding pre-stored values, wherein the pre-stored values for the motion vector for each of the one or more regions are stored in a training module;

the feature extraction module further configured to:
identify at least one motion vector, from the determined motion vectors, that matches with a pre-stored value of a motion vector thereby identifying a region of interest and a frame corresponding to the identified at least one motion vector, wherein the pre-stored value indicates onset of an expression; and determine one or more motion vectors for the identified region of interest in subsequent sets of successive frames;

the comparator further configured to:
compare value of the one or more determined motion vectors for the identified region of interest with at least one of a pre-stored value of motion vector corresponding to peak of an indicated expression and a pre-stored value of motion vector corresponding to offset of the indicated expression, wherein the peak of the indicated expression and the offset of the indicated expression are values in at least one of a graph, a histogram and a table depicting the indicated expression; and
identify the frame corresponding to the at least one of: the peak of the indicated expression and the offset of the indicated expression; and

a testing module configured to generate a pictorial representation of the one or more videos depicting at least one of: the onset, peak and offset of the indicated expression of the one or more individuals captured in the one or more videos for predicting the one or more neurological disorders.

9. The system of claim 8 , wherein the one or more expressions include at least one of: happy, content, sad, disgust, surprise, clueless and angry.

10. The system of claim 8 , wherein the detected one or more regions of interest include at least one of: eyes, cheeks, nose, lips, ears, eyebrows, hands, arms, torso, legs and feet.

11. The system of claim 8 , wherein the regions of interest are detected using Viola-Jones algorithm.

12. The system of claim 8 , wherein the motion vector for each of the one or more regions of interest in a set of successive frames is determined using optical flow algorithm.

13. The system of claim 12 , wherein the optical flow algorithm uses Horn-Schunck method for determining the motion vector.

14. The system of claim 8 , wherein the one or more individuals are speaking while the one or more videos are being captured.

15. A computer program product for predicting one or more neurological disorders, the computer program product comprising:
a non-transitory computer-readable medium having computer-readable program code stored thereon, the computer-readable program code comprising instructions that when executed by a processor, cause the processor to:
receive one or more videos of one or more individuals;
split the one or more videos into one or more frames:
detect one or more regions of interest in each of the one or more frames;
analyze each of the one or more regions of interest in each of the one or more frames by:
determining a motion vector for each of the one or more regions of interest in a set of successive frames; and
comparing value of the determined motion vector for each of the one or more regions of interest with corresponding pre-stored values, wherein the pre-stored values for the motion vector for each of the one or more regions are pre-stored in a database;

identify at least one motion vector, from the determined motion vectors, that matches with a pre-stored value of a motion vector thereby identifying a region of interest and a frame cor

Assignees

• Cognizant Tech Solutions India Pvt Ltd
• Cognizant Tech Solutions India Pvt Ltd

Inventors

• Krishna Rao Geelapaturu Subrahmanya Venkata Radha
• Roy Kuhelee
• Margret Anouncia Savarimuthu

Classifications

• H04N19/17

  the unit being an image region, e.g. an object · CPC title

• G06T2207/10016

  Video; Image sequence · CPC title

• A61B5/7267

  involving training the classification device · CPC title

• A61B5/7257

  using Fourier transforms · CPC title

• A61B5/4082

  Diagnosing or monitoring movement diseases, e.g. Parkinson, Huntington or Tourette · CPC title