DETECTION OF WAVEFORM ARTIFACT

• US 20110105927A1
• Filed: 10/30/2009
• Published: 05/05/2011
• Est. Priority Date: 10/30/2009
• Status: Active Grant

• Current Assignee: Medtronic Incorporated (Medtronic Plc)
• Original Assignee: Medtronic Incorporated (Medtronic Plc)
• Inventors: Davie, Scott W., Greenhut, Saul E., Kleckner, Karen J., Bennett, Tommy D., Karamanoglu, Mustafa
• Granted Patent: US 8,380,295 B2
• Time in Patent Office: Days
• US Class Current: 600/513
• CPC Class Codes: A61N 1/3702 Physiological parameters A61N1/365 takes precedence; evoked response A61N1/371

Abstract

An implantable medical device system including a physiological sensor detects signal artifact in a signal waveform acquired by the sensor. Features of individual waveforms in the sensor signal are extracted. Sample waveforms are classified by expert observation into at least two classes including an artifact class. A distribution range for each of the extracted features from the sample waveforms is determined for each of the classes. Waveform classification criteria are established in response to the determined distribution ranges.

Claims

1. A method for detecting signal artifact in a signal waveform acquired by an implantable physiological sensor, the method comprising:

sensing a signal in a patient using the implantable physiological sensor and separating the signal into a plurality of waveforms;

extracting a plurality of features from individual waveforms of the plurality of waveforms;

classifying the individual waveforms according to expert observation, the individual waveforms classified into at least two classes comprising an artifact class;

determining a distribution range for the plurality of features for each of the classes;

establishing waveform classification criteria in response to the determined distribution ranges;

acquiring unknown waveforms from an implantable physiological sensor;

extracting the plurality of features from the unknown waveforms;

classifying the unknown waveforms according to the established waveform classification criteria; and

detecting signal artifact in response to a waveform being classified in the artifact class.

2. The method of claim 1 further comprising computing a metric for monitoring a physiological condition of the patient in response to an unknown waveform not being classified in the artifact class, the metric computed as a function of at least one of the features extracted from the unknown waveforms.

3. The method of claim 1 wherein determining the distribution range comprises computing a morphology score as a function of the extracted features and determining the distribution range of the morphology score for each of the classes.

4. The method of claim 1 wherein extracting the plurality of features comprises extracting a non-physiological feature of the waveform, the non-physiological feature used for classifying the unknown waveforms;

the method further comprising;

extracting a physiological feature from the unknown waveforms not classified in the artifact class; and

computing a monitoring metric as a function of the physiological feature for monitoring a physiological condition of a patient.

5. The method of claim 4 further comprising:

extracting a physiological feature from the unknown waveforms classified in the artifact class, computing the monitoring metric using the physiological feature extracted from the unknown waveforms classified in the artifact class and from the unknown waveforms not classified in the artifact class;

determining a frequency of the unknown waveforms classified in the artifact class; and

storing the monitoring metric with the frequency.

6. The method of claim 1 further comprising:

sensing a second physiological signal in the patient, the second physiological signal correlated to a potential source of artifact in the plurality of unknown waveforms;

determining a potential artifact source level from the second physiological signal;

determining a frequency of artifact waveform classifications occurring in the unknown waveforms; and

storing the artifact source level with the frequency of artifact waveform classifications.

7. A medical device system for monitoring a physiological signal in a patient and detecting artifact in the physiological signal, the system comprising:

an implantable physiological sensor for sensing the physiological signal;

a display for displaying the physiological signal to enable an expert to classify individual waveforms of the physiological signal based on visual observation, the individual waveforms classified into at least two classes comprising an artifact class;

a user interface for receiving user input to establish the classification of each of the individual waveforms;

a memory for storing the user input classifications of the individual waveforms;

a processor configured to;

receive the implantable physiological sensor signal;

separate the signal into a plurality of individual waveforms;

extract a plurality of features from each individual waveform;

determine a distribution range for the plurality of features for each of the classes, and

establish waveform classification criteria in response to the determined distribution ranges; and

an implantable processor adapted to be coupled to an implantable physiological sensor, the implantable processor configured to;

acquire unknown waveforms from the implantable physiological sensor,

extract the plurality of features from the unknown waveforms,

classify the unknown waveforms according to the established waveform classification criteria, and

detect signal artifact in response to a waveform being classified in the artifact class.

8. The system of claim 7 wherein determining the distribution range comprises computing a morphology score as a function of the extracted features and determining the distribution range of the morphology score for each of the classes.

9. The system of claim 7 wherein the implantable processor is further configured to compute a metric for monitoring a physiological condition of the patient, the metric computed as a function of at least one of the features extracted from the unknown waveforms in response to an unknown waveform not being classified in the artifact class.

10. The system of claim 7 wherein the plurality of features extracted comprises a non-physiological feature of the waveform, the non-physiological feature used for classifying the unknown waveforms;

and

the implantable processor further configured to extract a physiological feature from an unknown waveform in response to the unknown waveform not being classified in the artifact class, and

compute a metric as a function of the physiological feature for monitoring a physiological condition of a patient.

11. The system of claim 10 wherein the implantable processor is further configured to:

extract a physiological feature from the unknown waveforms classified in the artifact class;

compute the monitoring metric using the physiological feature extracted from unknown waveforms classified in the artifact class and the physiological features extracted from unknown waveforms not classified in the artifact class;

determine a frequency of the unknown waveforms classified in the artifact class; and

store the monitoring metric with the frequency.

12. The system of claim 10 wherein the non-physiological feature comprises one of a notch amplitude, a notch width, and a notch area.

13. The system of claim 10 wherein extracting the non-physiological feature comprises dividing the waveform into a plurality of segments and searching within one of the plurality of segments for the non-physiological feature.

14. The system of claim 7 wherein the established criteria comprise a plurality of rules defined for detecting a waveform class, the plurality of rules comprising a threshold comparison established in response to determining the distribution range of an extracted feature and at least one rule of the plurality of rules comprising a Boolean logic operation.

15. The system of claim 14 wherein the threshold comparison comprises comparing a first feature extracted from an unknown waveform to a variable threshold that is dependent on a second feature extracted from the unknown waveform and different than the first feature.

16. The system of claim 7 wherein extracting the plurality of features comprises computing a waveform derivative.

17. The system of claim 7 wherein the implantable processor is further configured to count a number of the unknown waveforms being classified in the artifact class and generate a notification in response to the count.

18. The system of claim 7 further comprising counting a number of the unknown waveforms being classified in the artifact class in a series of waveforms and reject the entire series of waveforms as artifact in response to the count.

19. The system of claim 7 wherein the at least two classes further comprise an atypical class and a typical class, the implantable processor further configured to:

determine a percentage of the unknown waveforms classified as each of the artifact class, atypical class, and typical class;

compute a monitoring metric using the unknown waveforms; and

report the monitoring metric with the percentage of one of the artifact, atypical and typical classes to establish a confidence level of the reported metric.

20. The system of claim 7 further comprising:

a second physiological signal in the patient for sensing signal correlated to a potential source of artifact in the plurality of unknown waveforms;

the implantable processor further configured to receive the second physiological signal, determine a potential artifact source level from the second physiological signal, determine a frequency of artifact waveform classifications occurring in the unknown waveforms, and store the artifact source level with the frequency of artifact waveform classifications.

21. An implantable medical device for monitoring a physiological signal in a patient and detect portions of the signal containing artifact, the device comprising:

a memory storing waveform classification criteria established based on expert observational classification of waveforms previously acquired by an implantable physiological sensor;

an implantable physiological sensor for sensing a signal in a patient; and

a processor configured to;

receive the sensor signal and separate the signal into a plurality of unknown waveforms,

extracting the plurality of features from the unknown waveforms,

classify the unknown waveforms according to the established waveform classification criteria, and

detect signal artifact in response to classifying an unknown waveform in the artifact class.