IMPLANTABLE MEDICAL DEVICE WITH ADAPTIVE SIGNAL PROCESSING AND ARTIFACT CANCELLATION

• US 20100179395A1
• Filed: 01/15/2010
• Published: 07/15/2010
• Est. Priority Date: 01/15/2009
• Status: Active Grant

• Current Assignee: Medtronic Incorporated (Medtronic Plc)
• Original Assignee: Medtronic Incorporated (Medtronic Plc)
• Inventors: Bloem, David R.
• Granted Patent: US 8,587,426 B2
• Time in Patent Office: Days
• US Class Current: 600/301
• CPC Class Codes: A61B 5/0205 Simultaneously evaluating both cardiovascular conditions and different types of body conditions, e.g. heart and respiratory condition
• CPC Class Codes: A61B 5/1455 using optical sensors, e.g. spectral photometrical oximeters
• CPC Class Codes: A61B 5/7207 of noise induced by motion artifacts
• CPC Class Codes: A61N 1/3718 Monitoring of or protection against external electromagnetic fields or currents
• CPC Class Codes: A61N 1/3925 Monitoring; Protecting
• CPC Class Codes: G06F 18/2135 based on approximation criteria, e.g. principal component analysis
• CPC Class Codes: G06F 2218/02 Preprocessing
• CPC Class Codes: G06V 40/15 Biometric patterns based on physiological signals, e.g. heartbeat, blood flow

Abstract

A medical device includes multiple sensors used to acquire sensor signals grouped into multiple sets to obtain multiple multi-dimensional signals. Principal component analysis of the multi-dimensional signals is performed to compute principal components of variation of the multi-dimensional signals. Features extracted from the principal components are used in detecting physiological events.

Claims

1. A medical device, comprising:

a sensor module for sensing a plurality of physiological signals

a signal grouping module receiving the plurality of physiological signals and grouping the plurality of physiological signals into a plurality of multi-dimensional signals;

a plurality of processing modules each receiving one of the plurality of multi-dimensional signals, each of the plurality of processing modules configured to extract a first feature set from the sensor signal set, perform principal component analysis of the plurality of multi-dimensional signals to compute principal components of variation for each of the plurality of multi-dimensional signals, and extract a second feature set from the principal components; and

a controller receiving the first feature set and the second feature set from at least one of the plurality of processing modules, the controller responsive to the first feature set and the second feature set to detect a physiological event.

2. The device of claim 1 wherein the plurality of processing modules are further configured to compute a compensated signal from the principal components computed for a respective one of the plurality of multi-dimensional signals, the compensated signal having a reduced dimensionality as compared to the respective multi-dimensional signal, and extract the second feature set from the compensated signal.

3. The device of claim 1 wherein the signal grouping module is configured to group the plurality of physiological signals in response to one of a first feature set and a second feature set.

4. The device of claim 1 wherein the signal grouping module is configured to group the plurality of physiological signals in response to a signal from the controller corresponding to a detected event.

5. The device of claim 1 wherein the sensor module comprises a plurality of physiological sensors and is configured to selectively enable a physiological sensor in response to the signal grouping module.

6. The device of claim 2 wherein the controller is further configured to detect a known patient condition in response to one of a physiological sensor signal and patient input, and enable a training flag in response to the known patient condition, wherein one of the plurality of processing modules is configured to receive the training flag and compute a template of one of the multi-dimensional signals corresponding to the known patient condition, the controller being configured to use the template in computing the compensated signal.

7. The device of claim 1 wherein the controller detects the event using a weighted combination of the first set and the second set of extracted features.

8. The device of claim 1 wherein the controller detects the event using a Boolean logic rule applied to the first set and the second set of extracted features.

9. The device of claim 1 wherein the plurality of processing modules are further configured to identify an extraneous signal in one of the plurality of multi-dimensional signals in response to the computed principal components, and wherein the sensing module is configured to disable sensing of a physiological signal in response to one of the plurality of processing modules identifying the extraneous signal.

10. A method, comprising:

sensing a plurality of physiological signals

grouping the plurality of physiological signals into a plurality of multi-dimensional signals;

extracting a first feature set from the sensor signal set;

performing principal component analysis of each of the plurality of multi-dimensional signals to compute principal components of variation for each of the plurality of multi-dimensional signals;

extracting a second feature set from the computed principal components; and

detecting a physiological event in response to the first feature set and the second feature set.

11. The method of claim 10 further comprising:

computing a compensated signal from the principal components compute for one of the plurality of multi-dimensional signals, the compensated signal having a reduced dimensionality as compared to the respective multi-dimensional signal, and

extracting the second feature set from the compensated signal.

12. The method of claim 10 further comprising grouping the plurality of physiological signals in response to one of a first feature set and a second feature set.

13. The method of claim 10 further comprising grouping the plurality of physiological signals in response to a detected event.

14. The method of claim 10 further comprising selectively enable a physiological sensor in response to the signal grouping.

15. The method of claim 11 further comprising

detecting a known patient condition in response to one of a physiological sensor signal and patient input;

and

enabling a training flag in response to the known patient condition;

computing a template of one of the multi-dimensional signals corresponding to the known patient condition using principal component analysis, and

using the template to compute the compensated signal.

16. The method of claim 10 further comprising detecting the event using a weighted combination of the first set and the second set of extracted features.

17. The method of claim 10 further comprising detecting the event using a Boolean logic rule applied to the first set and the second set of extracted features.

18. The method of claim 10 further comprising:

identifying an extraneous signal in one of the plurality of multidimensional signals in response to the computed principal components, and

disabling sensing of one of the plurality of physiological signals in response to identifying the extraneous signal.

19. A computer readable medium storing a set of instructions which when implemented in a medical device system cause the system to:

sense a plurality of physiological signals

group the plurality of physiological signals into a plurality of multi-dimensional signals;

extract a first feature set from the sensor signal set;

perform principal component analysis of each of the plurality of multi-dimensional signals to compute principal components of variation for each of the plurality of multi-dimensional signals;

extract a second feature set from the computed principal components; and

detect a physiological event in response to the first feature set and the second feature set.