Detection of the heartbeat in cranial accelerometer data using independent component analysis

1. A system for determining a heartbeat signal describing the heartbeat of a patient, the system comprising:

a plurality of acceleration sensors disposable on a surface of an anatomical body part of the patient and configured to generate acceleration signals for measuring an acceleration of the surface;

at least one computer being operatively coupled to the acceleration sensors and configured to receive information corresponding to the acceleration signals from the acceleration sensors and to acquire acceleration measurement data based on that information,

wherein the at least one computer comprises at least one processor configured to execute computer-executable instructions for a method for determining a heartbeat signal describing the heartbeat of a patient in the time domain, the instructions, when executed, configure the at least one processor to:

acquire, at the at least one processor, acceleration measurement data describing an acceleration in the time domain of an anatomical body part measured on an external surface of the anatomical body part;

determine, by the at least one processor, component analysis data describing a result of an independent component analysis in the time domain of the acceleration measurement data;

acquire, at the at least one processor, heartbeat template data describing template shapes of heartbeat in the time domain;

determine, by the at least one processor and based on the component analysis data and the heartbeat template data, recurrent shape data describing a recurrence of certain signal shapes in the component analysis data; and

determine, by the at least one processor and based on the recurrent shape data, heartbeat signal data describing a time series of the heartbeat.

2. A computer-implemented method for determining a heartbeat signal describing the heartbeat of a patient in the time domain, the method comprising:

a) obtaining, from a plurality of acceleration sensors by at least one processor, acceleration measurement data describing an acceleration in the time domain of an anatomical body part measured on an external surface of the anatomical body part;

b) determining, by the at least one processor, component analysis data describing a result of an independent component analysis in the time domain of the acceleration measurement data;

c) acquiring, at the at least one processor, heartbeat template data describing template shapes of heartbeat in the time domain;

d) determining, by the at least one processor and based on the component analysis data and the heartbeat template data, recurrent shape data describing a recurrence of certain signal shapes in the component analysis data;

e) determining, by the at least one processor and based on the recurrent shape data, heartbeat signal data describing a time series of the heartbeat.

3. The method according to claim 2 , wherein the anatomical body part comprises the head.

4. The method according to claim 2 , comprising

preprocessing, by the at least one processor, the acceleration measurement data for at least one of noise reduction, low pass filtering and high pass filtering.

5. The method according to claim 2 , comprising

determining, by the at least one processor, whether a real-time independent component analysis on the acceleration measurement data is possible, and

if determined that the real-time independent component analysis is possible, determining, by the at least one processor, the component analysis data from a stream of the acceleration measurement data; and

if determined that the real-time independent component analysis is not possible, buffering, by the at least one processor, the acceleration measurement data and

determining, by the at least one processor, the component analysis data by conducting the independent component analysis on at least one batch of the buffered acceleration measurement data.

6. The method according to claim 2 , comprising

determining, by the at least one processor and based on the component analysis data, whether the independent component analysis of the acceleration measurement data describes recurring time domain shapes; and

if determined that the independent component analysis of the acceleration measurement data describes recurring time domain shapes,

determining, by the at least one processor, whether those shapes are described by the heartbeat template data; and

if determined that those shapes are not described by the heartbeat template data,

adding, by the at least one processor, those shapes to the heartbeat template data.

7. The method according to claim 2 , comprising

performing the independent component analysis by running, by the at least one processor, a fast independent component analysis algorithm on the acceleration measurement data.

8. The method according to claim 2 , comprising

determining, by the at least one processor and based on the heartbeat signal data, a frequency domain representation of the time series of the heartbeat; and

determining, by the at least one processor and based on the frequency domain representation, whether a peak frequency described by the frequency domain representation matches a predetermined heartrate.

9. The method according to claim 8 , wherein the frequency domain representation is determined by determining a Fourier transform of the time series of the heartbeat.

10. The method according to claim 8 , wherein the predetermined heartrate is acquired, at the at least one processor, as second measurement data received from a second heartbeat sensor, wherein the second heartbeat sensor includes at least one of a photoplethysmography device or an electrocardiography device.

11. The method according to claim 2 , wherein the heartbeat signal data is determined by executing at least one of the following:

determining, by the at least one processor and based on the recurrent shape data, a single independent component;

determining, by the at least one processor and based on the recurrent shape data, a combination of single independent components; or

determining, by the at least one processor, at least one of a convolution cross correlation of the component analysis data and the heartbeat template data.

12. The method according to claim 9 , wherein the heartbeat signal data is determined by determining, by the at least one processor, the harmonics of the frequency domain representation of the heartbeat.

13. A non-transitory computer-readable program storage medium on which a program is stored which, when executed by at least one processor of at least one computer, causes the at least one computer to perform a method for determining a heartbeat signal describing the heartbeat of a patient in the time domain, the program for the method configures the at least one processor to:

obtain, from a plurality of acceleration sensors , acceleration measurement data describing an acceleration in the time domain of an anatomical body part measured on an external surface of the anatomical body part;

determine component analysis data describing a result of an independent component analysis in the time domain of the acceleration measurement data;

acquire heartbeat template data describing template shapes of heartbeat in the time domain;

determine, based on the component analysis data and the heartbeat template data, recurrent shape data describing a recurrence of certain signal shapes in the component analysis data;

determine, based on the recurrent shape data, heartbeat signal data describing a time series of the heartbeat.

14. The non-transitory computer-readable program storage medium of claim 13 , wherein the program further configures the at least one processor to preprocess the acceleration measurement data prior to the independent component analysis.

15. The non-transitory computer-readable program storage medium of claim 13 , wherein, to determine recurrent shape data, the program configures the at least one processor to compare a heartbeat template shape, from the heartbeat template data, to a component channel from the component analysis data.

16. The non-transitory computer-readable program storage medium of claim 15 , wherein the program further configures the at least one processor to slide the heartbeat template shape over the component channel according to a predetermined step size and identify positions where a similarity value exceeds a threshold.

17. The non-transitory computer-readable program storage medium of claim 15 , wherein the program further configures the at least one processor to determine recurrence of the heartbeat template shape on the component channel within an anatomical relevant distance from the identified positions.

18. The non-transitory computer-readable program storage medium of claim 15 , wherein the program further configures the at least one processor to:

identify a recurring shape in at least one component channel following the independent component analysis of the acceleration measurement data;

determine whether the recurring shape is described in the heartbeat template data; and

add the recurring shape to the heartbeat template data when the recurring shape is not described.

19. The non-transitory computer-readable program storage medium of claim 15 , wherein the program further configures the at least one processor to determine a heartrate from the heartbeat signal data.

20. The non-transitory computer-readable program storage medium of claim 19 , wherein the program further configures the at least one processor to:

acquire second measurement data from a second heartbeat sensor; and

validate the heartrate determined from the heartbeat signal data based on the second measurement data.