- Patent Number:
10376,192
- Appl. No:
15/923225
- Application Filed:
March 16, 2018
- نبذة مختصرة :
A system and method for contactless blood pressure determination. The method includes: receiving a captured image sequence; determining, using a trained hemoglobin concentration (HC) changes machine learning model, bit values from a set of bitplanes in the captured image sequence that represent the HC changes of the subject; determining a blood flow data signal; extracting one or more domain knowledge signals associated with the determination of blood pressure; building a trained blood pressure machine learning model with a blood pressure training set, the blood pressure training set including the blood flow data signal of the one or more predetermined ROIs and the one or more domain knowledge signals; determining, using the blood pressure machine learning model trained with a blood pressure training set, an estimation of blood pressure; and outputting the determination of blood pressure.
- Inventors:
NURALOGIX CORPORATION (Toronto, CA)
- Assignees:
NURALOGIX CORPORATION (Toronto, Ontario, CA)
- Claim:
1. A method for contactless blood pressure determination of a human subject, the method executed on one or more processors, the method comprising: receiving a captured image sequence of light re-emitted from the skin of one or more humans; determining, using a trained hemoglobin concentration (HC) changes machine learning model trained with a HC changes training set, bit values from a set of bitplanes in the captured image sequence that represent the HC changes of the subject, the HC changes training set comprising the captured image sequence; determining a blood flow data signal of one or more predetermined regions of interest (ROIs) of the subject captured on the images based on the bit values from the set of bitplanes that represent the HC changes; applying a plurality of band-pass filters, each having a separate passband, to each of the blood flow data signals to produce a bandpass filter (BPF) signal set for each ROI; extracting one or more domain knowledge signals associated with the determination of blood pressure from the blood flow data signal of each of the ROIs; building a trained blood pressure machine learning model with a blood pressure training set, the blood pressure training set comprising the BPF signal set of the one or more predetermined ROIs and the one or more domain knowledge signals; determining, using the blood pressure machine learning model trained with the blood pressure training set, an estimation of blood pressure for the human subject; and outputting the determination of blood pressure.
- Claim:
2. The method of claim 1 , wherein determining the estimation of blood pressure comprises determining an estimation of systolic blood pressure (SBP) and diastolic blood pressure (DBP).
- Claim:
3. The method of claim 1 , wherein the set of bitplanes in the captured image sequence that represent the HC changes of the subject are the bitplanes that are determined to significantly increase a signal-to-noise ratio (SNR).
- Claim:
4. The method of claim 1 , further comprising preprocessing the blood flow data signal with a Butterworth filter or a Chebyshev filter.
- Claim:
5. The method of claim 1 , wherein extracting the one or more domain knowledge signals comprises determining a magnitude profile of the blood flow data signal of each of the ROIs.
- Claim:
6. The method of claim 5 , wherein determining the magnitude profile comprises using digital filters to create a plurality of frequency filtered signals of the blood flow data signal in the time-domain for each image in the captured image sequence.
- Claim:
7. The method of claim 1 , wherein extracting the one or more domain knowledge signals comprises determining a phase profile of the blood flow data signal of each of the ROIs.
- Claim:
8. The method of claim 7 , wherein determining the phase profile comprises: applying a multiplier junction to the phase profile to generate a multiplied phase profile; and applying a low pass filter to the multiplied phase profile to generate a filtered phase profile.
- Claim:
9. The method of claim 7 , wherein determining the phase profile comprises determining a beat profile, the beat profile comprising a plurality of beat signals based on a Doppler or an interference effect.
- Claim:
10. The method of claim 1 , wherein extracting the one or more domain knowledge signals comprises determining at least one of systolic uptake, peak systolic pressure, systolic decline, dicrotic notch, pulse pressure, and diastolic runoff of the blood flow data signal of each of the ROIs.
- Claim:
11. The method of claim 1 , wherein extracting the one or more domain knowledge signals comprises determining waveform morphology features of the blood flow data signal of each of the ROIs.
- Claim:
12. The method of claim 1 , wherein extracting the one or more domain knowledge signals comprises determining one or more biosignals, the biosignals comprising at least one of heart rate measured from the human subject, Mayer waves measured from the human subject, and breathing rates measured from the human subject.
- Claim:
13. The method of claim 1 , further comprising receiving ground truth blood pressure data, and wherein the blood pressure training set further comprises the ground truth blood pressure data.
- Claim:
14. The method of claim 13 , wherein the ground truth blood pressure data comprises at least one of an intra-arterial blood pressure measurement of the human subject, an auscultatory measurement of the human subject, or an oscillometric measurement of the human subject.
- Claim:
15. A system for contactless blood pressure determination of a human subject, the system comprising one or more processors and a data storage device, the one or more processors configured to execute: a transdermal optical imaging (TOI) module to receive a captured image sequence of light re-emitted from the skin of one or more humans, the TOI module determines, using a trained hemoglobin concentration (HC) changes machine learning model trained with a HC changes training set, bit values from a set of bitplanes in the captured image sequence that represent the HC changes of the subject, the HC changes training set comprising the captured image sequence, the TOI module determines a blood flow data signal of one or more predetermined regions of interest (ROIs) of the subject captured on the images based on the bit values from the set of bitplanes that represent the HC changes; a profile module to extract one or more domain knowledge signals associated with the determination of blood pressure from the blood flow data signal of each of the ROIs; a filter module to apply a plurality of band-pass filters, each having a separate passband, to each of the blood flow data signals to produce a bandpass filter (BPF) signal set for each ROI; a machine learning module to build a trained blood pressure machine learning model with a blood pressure training set, the blood pressure training set comprising the BPF signal set of the one or more predetermined ROIs and the one or more domain knowledge signals, the machine learning module determines, using the blood pressure machine learning model trained with the blood pressure training set, an estimation of blood pressure of the human subject; and an output module to output the determination of blood pressure.
- Claim:
16. The system of claim 15 , wherein determination of the estimation of blood pressure by the machine learning module comprises determining an estimation of systolic blood pressure (SBP) and diastolic blood pressure (DBP).
- Claim:
17. The system of claim 15 , wherein the set of bitplanes in the captured image sequence that represent the HC changes of the subject are the bitplanes that are determined to significantly increase a signal-to-noise ratio (SNR).
- Claim:
18. The system of claim 15 , further comprising a filter module to preprocess the blood flow data signal with a Butterworth filter or a Chebyshev filter.
- Claim:
19. The system of claim 15 , wherein extracting the one or more domain knowledge signals by the profile module comprises determining a magnitude profile of the blood flow data signal of each of the ROIs.
- Claim:
20. The system of claim 19 , wherein determining the magnitude profile by the profile module comprises using digital filters to create a plurality of frequency filtered signals of the blood flow data signal in the time-domain for each image in the captured image sequence.
- Claim:
21. The system of claim 15 , wherein extracting the one or more domain knowledge signals by the profile module comprises determining a phase profile of the blood flow data signal of each of the ROIs.
- Claim:
22. The system of claim 21 , wherein determining the phase profile by the profile module comprises: applying a multiplier junction to the phase profile to generate a multiplied phase profile; and applying a low pass filter to the multiplied phase profile to generate a filtered phase profile.
- Claim:
23. The system of claim 22 , wherein determining the phase profile by the profile module comprises determining a beat profile, the beat profile comprising a plurality of beat signals based on a Doppler or an interference effect.
- Claim:
24. The system of claim 15 , wherein extracting the one or more domain knowledge signals by the profile module comprises determining at least one of systolic uptake, peak systolic pressure, systolic decline, dicrotic notch, pulse pressure, and diastolic runoff of the blood flow data signal of each of the ROIs.
- Claim:
25. The system of claim 15 , wherein extracting the one or more domain knowledge signals by the profile module comprises determining waveform morphology features of the blood flow data signal of each of the ROIs.
- Claim:
26. The system of claim 15 , wherein extracting the one or more domain knowledge signals by the profile module comprises determining one or more biosignals, the biosignals comprising at least one of heart rate measured from the human subject, Mayer waves measured from the human subject, and breathing rates measured from the human subject.
- Claim:
27. The system of claim 15 , wherein the profile module receives ground truth blood pressure data, and wherein the blood pressure training set further comprises the ground truth blood pressure data.
- Claim:
28. The system of claim 27 , wherein the ground truth blood pressure data comprises at least one of an intra-arterial blood pressure measurement of the human subject, an auscultatory measurement of the human subject, or an oscillometric measurement of the human subject.
- Patent References Cited:
7610166 October 2009 Solinsky
2013/0197377 August 2013 Kishi
2016/0098592 April 2016 Lee
2016037033 March 2016
WO 2016/037033 March 2016
- Other References:
Jain et al, “Face Video Based Touchless Blood Pressure and Heart Rate Estimation”, Sep. 2016, IEEE, pp. 1-5 (Year: 2016). cited by examiner
Xing et al, “Optical blood pressure estimation with photoplethysmography and FFT-based neural networks”, Jul. 2016, OSA, pp. 1-14 (Year: 2016). cited by examiner
International Search Report corresponding to PCT/CA2017/051533; Canadian Intellectual Property Office; dated Feb. 20, 2018. cited by applicant
Written Opinion of the International Searching Authority corresponding to PCT/CA2017/051533; Canadian Intellectual Property Office; dated Feb. 20, 2018. cited by applicant
Jain, M. et al., “Face video based touchless blood pressure and heart rate estimation.” IEEE 18th International Workshop on Multimedia Signal Processing (MMSP), Sep. 23, 2016 (Sep. 23, 2016). cited by applicant
Xing X. et al., “Optical blood pressure estimation with photoplethysmography and FFT-based neural networks”, Biomedical optics express. vol. 7(8), Aug. 1, 2016 (Aug. 1, 2016, pp. 3007-3020. cited by applicant
- Primary Examiner:
Park, Edward
- Attorney, Agent or Firm:
Lampert, Marc
Bhole, Anil
Bhole IP Law
- الرقم المعرف:
edspgr.10376192
No Comments.