Discriminating between atrial fibrillation and sinus rhythm in physiological signals obtained from video

• US 9,320,440 B2
• Filed: 04/01/2014
• Issued: 04/26/2016
• Est. Priority Date: 04/01/2014
• Status: Active Grant

• Current Assignee: Xerox Corporation (Xerox Holdings Corp.)
• Original Assignee: Xerox Corporation (Xerox Holdings Corp.)
• Inventors: Kyal, Survi, Mestha, Lalit Keshav, Couderc, Jean-Philippe
• Primary Examiner(s): REMALY, MARK DONALD
• Application Number:
  US14/242,322
• Publication Number:
  US 20150272456A1
• Time in Patent Office: 756 Days
• US Class Current: 1/1
• CPC Class Codes: A61B 5/0077 Devices for viewing the surface of the body, e.g. camera, magnifying lens
• CPC Class Codes: A61B 5/0082 adapted for particular medical purposes
• CPC Class Codes: A61B 5/0295 using plethysmography, i.e. measuring the variations in the volume of a body part as modified by the circulation of blood therethrough, e.g. impedance plethysmography
• CPC Class Codes: A61B 5/1073 Measuring volume, e.g. of limbs

Abstract

What is disclosed is a system and method for determining whether a subject is in atrial fibrillation. A video is received of a region of exposed skin of a subject. The video is acquired of a region where a videoplethysmographic (VPG) signal can be registered by at least one imaging channel of a video imaging device. For each batch of image frames, pixels associated with the region of exposed skin are isolated and processed to obtain a time-series signal. A VPG signal is extracted from the time-series signal. The power spectral density (PSD) is computed across all frequencies within the VPG signal. A pulse harmonic strength (PHS) is calculated for this VPG signal. The pulse harmonic strength is compared to a discrimination threshold, defined herein. A determination is made whether the subject in the video is in atrial fibrillation or in normal sinus rhythm.

Claims

1. A method for determining whether a subject is having an atrial fibrillation event, the method comprising:

receiving a video of a region of exposed skin of a subject where a videoplethysmographic (VPG) signal can be registered by at least one imaging channel of a video imaging device used to capture that video;

defining a size N of a batch of image frames such that N_min≦

N≦

N_max, where N_min is a minimum size of a batch of image frames and N_max is a maximum size of a batch of image frames; and

for each batch of image frames of size N;

processing this batch of image frames to isolate pixels associated with said region of exposed skin;

processing said isolated pixels to obtain a time-series signal for this batch;

extracting a VPG signal from said time-series signal;

computing a power spectral density across all frequencies within said VPG signal;

calculating a pulse harmonic strength for this VPG signal;

comparing the pulse harmonic strength (PHS) to a discrimination threshold obtained from a Receiver Operating Characteristic (ROC) curve constructed from health vitals obtained for a group of cardiac patients, said threshold discriminating between low PHS values and high PHS values; and

classifying, as a result of said comparison, that said subject is one of;

in atrial fibrillation, and in sinus rhythm, wherein when PHS is below said threshold said subject is classified as being in atrial fibrillation, and in normal sinus rhythm otherwise.

2. The method of claim 1, wherein said video image frames comprise any combination of:

Near Infrared (NIR) images, color images, color with NIR images, multispectral images, and hyperspectral video images.

3. The method of claim 1, wherein said pulse harmonic strength for said VPG signal is a normalized pulse harmonic strength.

4. The method of claim 1, wherein, in advance of obtaining said time-series signal, pre-processing said video to compensate for any of:

an imaging blur, and slow illuminant variation.

5. The method in claim 1, wherein, said time-series signal is detrended to remove non-stationary signal components.

6. The method of claim 1, further comprising filtering said time-series signal with a bandpass filter with a low and high cutoff frequency defined as a function of said subject'"'"'s cardiac pulse frequency.

7. The method of claim 6, further comprising filtering said VPG signal to remove unwanted artifacts, said filtering using at least one of:

a Fast Fourier Transform (FFT)-based phase preservation filter, a zero-phase digital filter, a linear time invariant filter, a linear time varying filter, a finite impulse response filter, an infinite impulse response filter, a non-linear filter and a moving average filter.

8. The method in claim 1, wherein, said power spectral density is computed by performing any of:

a non-parametric and a parametric spectral density estimation on the time-series signal.

9. The method of claim 1, wherein said discrimination threshold is based on a obtained VPG signal from the subject during one of:

atrial fibrillation, sinus rhythm.

10. The method of claim 1, further comprising determining whether a movement occurred during video acquisition of a batch of image frames and, in response to said movement being above a predetermined threshold level for movement, not processing the current batch of image frames.

11. The method of claim 1, further comprising upsampling said time-series signal to a standard sampling frequency.

12. The method of claim 1, wherein, in response to said subject having an atrial fibrillation event, further comprising any of:

initiating an alert and signaling a medical professional.

13. The method of claim 1, further comprising dynamically adjusting a size of said batch of image frames.

14. The method of claim 1, wherein said video is a streaming video and said determination of an atrial fibrillation event occurs in real-time.

15. A system for determining whether a subject is having an atrial fibrillation event, the system comprising:

a memory and a storage device; and

a processor in communication with said memory and storage device, said processor executing machine readable instructions for performing;

receiving a video of a region of exposed skin of a subject where a videoplethysmographic (VPG) signal can be registered by at least one imaging channel of a video imaging device used to capture that video;

defining a size N of a batch of image frames such that N_min≦

N≦

N_max, where N_min is a minimum size of a batch of image frames and N_max is a maximum size of a batch of image frames; and

for each batch of image frames of size N;

processing this batch of image frames to isolate pixels associated with said region of exposed skin;

processing said isolated pixels to obtain a time-series signal for this batch;

extracting a VPG signal from said time-series signal;

computing a power spectral density across all frequencies within said VPG signal;

calculating a pulse harmonic strength for this VPG signal;

comparing the pulse harmonic strength (PHS) to a discrimination threshold obtained from a Receiver Operating Characteristic (ROC) curve constructed from health vitals obtained for a group of cardiac patients, said threshold discriminating between low PHS values and high PHS values; and

classifying, as a result of said comparison, that said subject is one of;

in atrial fibrillation, and in sinus rhythm, wherein when PHS is below said threshold said subject is classified as being in atrial fibrillation, and in normal sinus rhythm otherwise.

16. The system of claim 15, wherein said video image frames comprise any combination of:

Near Infrared (NIR) images, color images, color with NIR images, multispectral images, and hyperspectral video images.

17. The system of claim 15, wherein said pulse harmonic strength for said VPG signal is a normalized pulse harmonic strength.

18. The system of claim 15, wherein, in advance of obtaining said time-series signal, pre-processing said video to compensate for any of:

an imaging blur, and slow illuminant variation.

19. The system of claim 15, wherein, said time-series signal is detrended to remove non-stationary signal components.

20. The system of claim 15, further comprising filtering said time-series signal with a bandpass filter with a low and high cutoff frequency defined as a function of said subject'"'"'s cardiac pulse frequency.

21. The system of claim 20, further comprising filtering said VPG signal to remove unwanted artifacts, said filtering using at least one of:

a Fast Fourier Transform (FFT)-based phase preservation filter, a zero-phase digital filter, a linear time invariant filter, a linear time varying filter, a finite impulse response filter, an infinite impulse response filter, a non-linear filter and a moving average filter.

22. The system of claim 15, wherein, said power spectral density is computed by performing any of:

a non-parametric and a parametric spectral density estimation on the time-series signal.

23. The system of claim 15, wherein said discrimination threshold is based on a obtained VPG signal from the subject during one of:

atrial fibrillation, sinus rhythm.

24. The system of claim 15, further comprising determining whether a movement occurred during video acquisition of a batch of image frames and, in response to said movement being above a predetermined threshold level for movement, not processing the current batch of image frames.

25. The system of claim 15, further comprising upsampling said time-series signal to a standard sampling frequency.

26. The system of claim 15, wherein, in response to said subject having an atrial fibrillation event, further comprising any of:

initiating an alert and signaling a medical professional.

27. The system of claim 15, further comprising dynamically adjusting a size of said batch of image frames.

28. The system of claim 15, wherein said video is a streaming video and said determination of an atrial fibrillation event occurs in real-time.