Algorithms for detecting atrial arrhythmias from discriminatory signatures of ventricular cycle lengths

• US 7,031,765 B2
• Filed: 11/11/2002
• Issued: 04/18/2006
• Est. Priority Date: 11/11/2002
• Status: Active Grant

• Current Assignee: Medtronic Incorporated (Medtronic Plc)
• Original Assignee: Medtronic Incorporated (Medtronic Plc)
• Inventors: Sarkar, Shantanu, Ritscher, David E.
• Primary Examiner(s): Pezzuto, Robert E.
• Assistant Examiner(s): DROESCH, KRISTEN L
• Application Number:
  US10/292,285
• Publication Number:
  US 20040092836A1
• Time in Patent Office: 1,254 Days
• Field of Search: 600/508, 600/510, 600/515, 600/516, 600/518
• US Class Current: 600/518
• CPC Class Codes: A61B 5/363 Detecting tachycardia or bradycardia
• CPC Class Codes: A61N 1/3622 comprising two or more electrodes co-operating with different heart regions
• CPC Class Codes: G06F 2218/18 by plotting the signal segments against each other, e.g. analysing scattergrams

Abstract

Detection of arrhythmias is facilitated using irregularity of ventricular beats measured by delta-RR (ΔRR) intervals that exhibit discriminatory signatures when plotted in a Lorenz scatter-plot. An “AF signature metric” is established characteristic of episodes of AF that exhibit highly scattered (sparse) distributions or formations of 2-D data points. An “AFL signature metric” is established characteristic of episodes of AFL that exhibit a highly concentrated (clustered) distribution or formation of 2-D data points. A set of heart beat interval data is quantified to generate highly scattered (sparse) formations as a first discrimination metric and highly concentrated (clustered) distributions or formations as a second discrimination metric. The first discrimination metric is compared to the AF signature metric, and/or the second discrimination metric is compared to the AFL signature metric. AF or HFL is declared if the first discrimination metric satisfies either one of the AF signature metric.

Claims

1. A method of detecting atrial arrhythmias, comprising one or more of atrial fibrillation (AF) and atrial flutter (AFL) within a segment of ventricular heartbeats signified by intervals between successive ventricular heartbeats that exhibit discriminatory signatures when plotted as data points in a scatter-plot comprising:

defining an AF signature metric for AF representative of the sparse distribution of data points during AF episodes and an AFL signature metric representative of the clustering distribution of data points during AFL episodes;

determining each interval between successive ventricular heartbeats;

plotting each succeeding interval as a data point in a scatter-plot;

developing a discriminatory metric signifying the degree of sparseness or clustering of the data points of the scatter-plot;

declaring AF if the discriminatory metric satisfies the AF signature metric representative of the distribution of data points during AF; and

declaring AFL if the discriminatory metric satisfies the AFL signature metric representative of the distribution of data points during AFL.

2. The method of claim 1, wherein the scatter plot is replaced by a 3-D plot having a first interval on the X axis, a succeeding interval on the Y axis, and a net succeeding interval on the Z axis.

3. The method of claim 2, wherein the scatter plot is replaced by a multidimensional plot having multi-dimensional axis in successive orders thereof.

4. The method of claim 1, wherein the scatter-plot is a Lorenz plot having an origin, and the plotting step further comprises assigning an abscissa value and an ordinate value referenced to the origin of the Lorenz plot to each data point.

5. The method of claim 1, wherein the step of determining the interval comprises determining the change in interval between each succeeding ventricular heart beat from a preceding ventricular heartbeat.

6. The method of claim 4 or 5, wherein rather than using RR or Δ

RR, the absolute value RR or the absolute value of Δ

RR is used.

7. The method of claim 4 or 5, wherein rather than using RR or Δ

RR, a non-linear function RR or a no-linear function of Δ

RR is used, such as the log of the absolute value multiplied by the sign of the value.

8. The method of claim 5, wherein the step of determining the interval includes one interval axis and a change between intervals axis.

9. The method of claim 5, wherein the scatter-plot is a Lorenz plot having an origin, and the plotting step further comprises assigning an abscissa value and an ordinate value referenced to the origin of the Lorenz plot to each data point, whereby the data point is a two-dimensional (2-D) data point.

10. The method of claim 9, further comprising the steps of defining 2-D bins of the Lorenz plot with reference to the origin of the Lorenz plot and correlating each 2-D data point of the Lorenz plot with a bin of the Lorenz plot that the 2-D data point falls within and wherein:

the step of developing a discriminatory metric further comprises;

determining the nearest neighbor distance (NND) between each bin having a 2-D data point and the nearest bin having a 2-D data point;

processing the NND of each bin to derive an NND metric quantifying the degree of dispersion of 2-D data points among the bins;

the defining step comprises establishing an NND signature metric for AF; and

the declaring step comprises declaring AF if the derived NND metric satisfies the NND signature metric for AF.

11. The method of claim 5, 9 or 10, wherein the step of determining the interval includes arranging input variables comprising one of and combinations thereof, of interval, change of interval, filtered interval including a running media filter, sensor input including systolic pressure for that beta, blood oxygen pulse, or volume signals, timing information, including A-V intervals, beat morphology parameters, including wavelet match, QRS width, or frequency content.

12. The method of claim 10, wherein correlating with the bin is performed in a non-uniform manner, permitting some bins to occupy more space than other bins.

13. The method of claim 10, wherein points falling outside a central bin area are assigned to the nearest.

14. The method of claim 10, wherein points falling outside a central bin area are discarded (not counted).

15. The method of claim 10, wherein points falling outside a central bin area are assigned to a separate, extra bin, which is counted separately from the other bins.

16. The method of claim 9, further comprising the steps of defining 2-D bins of the Lorenz plot with reference to the origin of the Lorenz plot and correlating each 2-D data point of the Lorenz plot with a bin of the Lorenz plot that the 2-D data point falls within and wherein:

the step of developing a discriminatory metric further comprises;

defining a local search window;

determining the nearest neighbor distance (NND) between each bin having a 2-D data point and the nearest bin having a 2-D data point and that is within the defined search window;

for each 2-D data point in a bin, determining the distance from origin (DFO);

weighting the NND for each 2-D data point as a function of distance from origin (DFO); and

summing the weighted NNDs of all 2-D data points to derive an NND metric quantifying the degree of dispersion of 2-D data points among the bins;

the step of defining an AF signature metric comprises establishing an NND signature metric for AF; and

the declaring step comprises declaring AF if the derived NND metric satisfies the NND signature metric for AF.

17. The method of claim 16, wherein the weighting step further comprises:

defining a minimum DFO and a maximum DFO;

according a high weight to the NND for each 2-D data point having a DFO greater than the minimum DFO and less than the maximum DFO; and

according a low weight to the NND for each 2-D data point having a DFO less than the minimum DFO and greater than the maximum DFO;

whereby the NND metric is high when the dispersion of 2-D data points among the bins is great signifying AF affecting the ventricular heartbeats.

18. The method of claim 9, further comprising the steps of defining 2-D bins of the Lorenz plot with reference to the origin of the Lorenz plot and correlating each 2-D data point of the Lorenz plot with a bin of the Lorenz plot that the 2-D data point falls within and wherein:

the step of developing a discriminatory metric further comprises;

determining the nearest neighbor distance (NND) between each bin having a 2-D data point and the nearest bin having a 2-D data point;

processing the NND of each bin to derive an NND metric quantifying the degree of clustering of 2-D data points within one or more adjacent bin;

the step of defining an AFL signature metric comprises establishing an NND signature metric for AFL; and

the declaring step comprises declaring AFL if the derived NND metric satisfies the NND signature metric for AFL.

19. The method of claim 9, further comprising the steps of defining 2-D bins of the Lorenz plot with reference to the origin of the Lorenz plot and correlating each 2-D data point of the Lorenz plot with a bin of the Lorenz plot that the 2-D data point falls within and wherein:

the step of developing a discriminatory metric further comprises;

determining the nearest neighbor distance (NND) between each bin having a 2-D data point and the nearest bin having a 2-D data point and that is within a defined search window;

subtracting the determined NND from a maximum NND of the scatter-plot to provide an inverted NND (iNND);

for each 2-D data point in a bin, determining the distance from origin (DFO);

weighting the iNND for each 2-D data point as a function of distance from origin (DFO); and

summing the weighted iNNDs of all 2-D data points to derive an iNND metric quantifying the degree of dispersion of 2-D data points among the bins;

the step of defining an AFL signature metric comprises establishing an AFL signature metric quantifying the degree of clustering of 2-D data points within one or more adjacent bin; and

the declaring step comprises declaring AFL if the derived iNND discriminatory metric satisfies the AFL signature metric for AFL.

20. The method of claim 19, wherein the weighting step further comprises:

defining a minimum DFO and a maximum DFO;

according a low weight to the NND for each 2-D data point having a DFO greater than the minimum DFO and less than the maximum DFO; and

according a high weight to the NND for each 2-D data point having a DFO less than the minimum DFO and greater than the maximum DFO;

whereby the NND metric is high when the 2-D data points are highly concentrated at a short DFO signifying AFL affecting the ventricular heartbeats.

21. The method of claim 9, further comprising the steps of defining 2-D bins of the Lorenz plot with reference to the origin of the Lorenz plot and correlating each 2-D data point of the Lorenz plot with a bin of the Lorenz plot that the 2-D data point falls within and wherein:

the step of developing a discriminatory metric further comprises;

for each bin containing at least one 2-D data point, counting the number of 2-D data points within adjacent bins that are within a defined local window surrounding the bin;

determining a Null Count related to the number of bins that are within the local window that do not contain a 2-D data point; and

processing the Null Counts to derive an weighted null count (WNC) metric quantifying the degree of dispersion of 2-D data points among the bins;

the step of defining an AFL signature metric comprises establishing a WNC signature metric for AF; and

the declaring step comprises declaring AF if the derived WNC metric satisfies the WNC signature metric for AF.

22. The method of claim 9, further comprising the steps of defining 2-D bins of the Lorenz plot with reference to the origin of the Lorenz plot and correlating each 2-D data point of the Lorenz plot with a bin of the Lorenz plot that the 2-D data point falls within and wherein:

the step of developing a discriminatory metric further comprises;

for each bin containing at least one 2-D data point, counting the number of 2-D data points within adjacent bins that are within a defined local window surrounding the bin;

determining a Null Count related to the number of bins that are within the local window that do not contain a 2-D data point; and

for each 2-D data point in a bin having a Null Count associated with it, determining the distance from origin (DFO);

weighting the Null Count for each 2-D data point as a function of distance from origin (DFO); and

summing the Null Counts to derive an weighted null count (WNC) metric quantifying the degree of dispersion of 2-D data points among the bins;

the step of defining an AF signature metric comprises establishing a WNC signature metric for AF; and

the declaring step comprises declaring AF if the derived WNC metric satisfies the WNC signature metric for AF.

23. The method of claim 22, wherein the weighting step further comprises:

defining a minimum DFO and a maximum DFO;

according a high weight to the Null Count for each 2-D data point having a DFO greater than the minimum DFO and less than the maximum DFO; and

according a low weight to the Null Count for each 2-D data point having a DFO less than the minimum DFO and greater than the maximum DFO;

whereby the WNC metric is high when the dispersion of 2-D data points among the bins is great signifying AF affecting the ventricular heartbeats.

24. The method of claim 9, further comprising the steps of defining 2-D bins of the Lorenz plot with reference to the origin of the Lorenz plot and correlating each 2-D data point of the Lorenz plot with a bin of the Lorenz plot that the 2-D data point falls within and wherein:

the step of developing a discriminatory metric further comprises;

for each bin containing at least one 2-D data point, counting the number of 2-D data points within adjacent bins that are within a defined local window surrounding the bin as a Point Count;

processing the Point Counts to derive an weighted point count (WPC) metric quantifying the degree of clustering of 2-D data points among the bins;

the step of defining an AFL signature metric comprises establishing a WPC signature metric for AFL; and

the declaring step comprises declaring AF if the derived WPC metric satisfies the WPC signature metric for AFL.

25. The method of claim 9, further comprising the steps of defining 2-D bins of the Lorenz plot with reference to the origin of the Lorenz plot and correlating each 2-D data point of the Lorenz plot with a bin of the Lorenz plot that the 2-D data point falls within and wherein:

the step of developing a discriminatory metric further comprises;

for each bin containing at least one 2-D data point, counting the number of 2-D data points within adjacent bins that are within a defined local window surrounding the bin to derive a Point Count;

for each 2-D data point in a bin having a Point Count associated with it, determining the distance from origin (DFO);

weighting the Point Count for each 2-D data point as a function of distance from origin (DFO); and

summing the Point Counts to derive an weighted point count (WPC) metric quantifying the degree of clustering of 2-D data points among the bins;

the step of defining an AFL signature metric comprises establishing a WPC signature metric for AFL; and

the declaring step comprises declaring AFL if the derived WPC metric satisfies the WPC signature metric for AFL.

26. The method of claim 25, wherein the weighting step further comprises:

defining a minimum DFO and a maximum DFO;

according a low weight to the Null Count for each 2-D data point having a DFO greater than the minimum DFO and less than the maximum DFO; and

according a high weight to the Null Count for each 2-D data point having a DFO less than the minimum DFO and greater than the maximum DFO;

whereby the WNC metric is high when the 2-D data points are clustered within one or more bin signifying AFL affecting the ventricular heartbeats.

27. The method of claim 9, further comprising the steps of defining 2-D bins of the Lorenz plot with reference to the origin of the Lorenz plot and correlating each 2-D data point of the Lorenz plot with a bin of the Lorenz plot that the 2-D data point falls within and wherein:

the step of developing a discriminatory metric further comprises;

counting the number of 2-D data points added to each bin over the segment;

determining the highest number of 2-D data points in a single bin as a MaxVal;

counting the number of bins containing at least one 2-D data point as a nZeroCntnZeroCnt;

counting the number of bins containing at least one 2-D data point and a predetermined number of 2-D data points as a loCnt; and

calculating the cluster signature metric (CSM) by;

CSM=loCnt+nZeroCnt−

MaxVal,

the step of defining an AF signature metric comprises establishing a CSM signature metric for AF; and

the declaring step comprises declaring AF if the derived CSM satisfies the CSM signature metric for AF.

28. The method of claim 9, further comprising the steps of defining 2-D bins of the Lorenz plot with reference to the origin of the Lorenz plot and correlating each 2-D data point of the Lorenz plot with a bin of the Lorenz plot that the 2-D data point falls within and wherein:

the step of developing a discriminatory metric further comprises;

counting the number of 2-D data points added to each bin over the segment;

determining the highest number of 2-D data points in a single bin as a MaxVal;

counting the number of bins containing at least one 2-D data point as a nZeroCnt;

counting the number of bins containing at least one 2-D data point and a predetermined number of 2-D data points as a loCnt; and

calculating the cluster signature metric (CSM) by;

CSM=loCnt+nZeroCnt−

MaxVal,

the step of defining an AFL signature metric comprises establishing a CSM signature metric for AFL; and

the declaring step comprises declaring AFL if the derived CSM satisfies the CSM signature metric for AFL.

29. The method of claim 9, further comprising the steps of defining 2-D bins of the Lorenz plot with reference to the origin of the Lorenz plot and correlating each 2-D data point of the Lorenz plot with a bin of the Lorenz plot that the 2-D data point falls within and wherein:

the step of developing a discriminatory metric further comprises;

counting the number of 2-D data points added to each bin over the segment;

determining the highest number of 2-D data points in a single bin as a MaxVal;

counting the number of bins containing at least one 2-D data point as a nZeroCnt;

counting the number of bins containing at least one 2-D data point and a predetermined number of 2-D data points as a loCnt; and

calculating the cluster signature metric (CSM) by;

CSM=loCnt+nZeroCnt−

MaxVal,

the step of defining an AF signature metric comprises establishing a CSM signature metric for AF;

the step of defining an AF signature metric comprises establishing a CSM signature metric for AF; and

the declaring step comprises declaring AF if the derived CSM satisfies the CSM signature metric for AF and declaring AFL if the derived CSM satisfies the CSM signature metric for AFL.

30. The method of claim 9, further comprising the steps of defining 2-D bins of the Lorenz plot with reference to the origin of the Lorenz plot and correlating each 2-D data point of the Lorenz plot with a bin of the Lorenz plot that the 2-D data point falls within and wherein:

the step of developing a discriminatory metric further comprises;

incrementing a count of the number of bins containing at least one 2-D data point as a nZeroCnt upon each occurrence of a 2-D data count;

determining the maximum number of 2-D data points accumulated in any bin over the segment as a MaxVal; and

calculating the cluster signature metric (CSM) by;

CSM=nZeroCnt−

MaxVal,

the step of defining an AF signature metric comprises establishing a CSM signature metric for AF; and

the declaring step comprises declaring AF if the derived CSM satisfies the CSM signature metric for AF.

31. The method of claim 9, further comprising the steps of defining 2-D bins of the Lorenz plot with reference to the origin of the Lorenz plot and correlating each 2-D data point of the Lorenz plot with a bin of the Lorenz plot that the 2-D data point falls within and wherein:

the step of developing a discriminatory metric further comprises;

incrementing a count of the number of bins containing at least one 2-D data point as a nZeroCnt upon each occurrence of a 2-D data count;

determining the maximum number of 2-D data points accumulated in any bin over the segment as a MaxVal; and

calculating the cluster signature metric (CSM) by;

CSM=nZeroCnt−

MaxVal,

the step of defining an AFL signature metric comprises establishing a CSM signature metric for AFL; and

the declaring step comprises declaring AFL if the derived CSM satisfies the CSM signature metric for AFL.

32. The method of claim 9, further comprising the steps of defining 2-D bins of the Lorenz plot with reference to the origin of the Lorenz plot and correlating each 2-D data point of the Lorenz plot with a bin of the Lorenz plot that the 2-D data point falls within and wherein:

the step of developing a discriminatory metric further comprises;

incrementing a count of the number of bins containing at least one 2-D data point as a nZeroCnt upon each occurrence of a 2-D data count;

determining the maximum number of 2-D data points accumulated in any bin over the segment as a MaxVal; and

calculating the cluster signature metric (CSM) by;

CSM=nZeroCnt−

MaxVal,

the step of defining an AF signature metric comprises establishing a CSM signature metric for AF;

the step of defining an AFL signature metric comprises establishing a CSM signature metric for AFL; and

the declaring step comprises declaring AF if the derived CSM satisfies the CSM signature metric for AF and declaring AFL if the derived CSM satisfies the CSM signature metric for AFL.

33. The method of claim 9, further comprising the steps of defining 2-D bins of the Lorenz plot with reference to the origin of the Lorenz plot and correlating each 2-D data point of the Lorenz plot with a bin of the Lorenz plot that the 2-D data point falls within and wherein:

the step of developing a discriminatory metric further comprises incrementing a count of the number of bins containing at least one 2-D data point as a nZeroCnt upon each occurrence of a 2-D data count;

the step of defining an AFL signature metric comprises establishing a CSM signature metric for AFL; and

the declaring step comprises declaring AFL if the derived nZeroCnt satisfies the CSM signature metric for AFL.

34. The method of claim 9, further comprising the steps of defining 2-D bins of the Lorenz plot with reference to the origin of the Lorenz plot and correlating each 2-D data point of the Lorenz plot with a bin of the Lorenz plot that the 2-D data point falls within and wherein:

the step of developing a discriminatory metric further comprises determining the maximum number of 2-D data points accumulated in any bin over the segment as a MaxVal;

the step of defining an AFL signature metric comprises establishing a CSM signature metric for AFL; and

the declaring step comprises declaring AFL if the derived MaxVal satisfies the CSM signature metric for AFL.

35. The method of claim 9, further comprising the steps of defining 2-D bins of the Lorenz plot with reference to the origin of the Lorenz plot and correlating each 2-D data point of the Lorenz plot with a bin of the Lorenz plot that the 2-D data point falls within and wherein:

the step of developing a discriminatory metric further comprises;

incrementing a count of the number of bins containing at least one 2-D data point as a nZeroCnt upon each occurrence of a 2-D data count; and

determining the maximum number of 2-D data points accumulated in any bin over the segment as a MaxVal;

the step of defining an AF signature metric comprises establishing a CSM signature metric for AF;

the step of defining an AFL signature metric comprises establishing a CSM signature metric for AFL;

the declaring step comprises declaring AF if the derived nZeroCnt satisfies the CSM signature metric for AF and declaring AFL if the derived MaxVal satisfies the CSM signature metric for AFL.

36. Apparatus for detecting atrial arrhythmias, comprising one or more of atrial fibrillation (AF) and atrial flutter (AFL) within a segment of ventricular heartbeats signified by intervals between successive ventricular heartbeats that exhibit discriminatory signatures when plotted as data points in a scatter-plot comprising:

means for defining an AF signature metric for AF representative of the sparse distribution of data points during AF episodes and an AFL signature metric representative of the clustering distribution of data points during AFL episodes;

means for determining each interval between successive ventricular heartbeats;

means for plotting each succeeding interval as a data point in a scatter-plot;

means for developing a discriminatory metric signifying the degree of sparseness or clustering of the data points of the scatter-plot;

means for declaring AF if the discriminatory metric satisfies the AF signature metric representative of the distribution of data points during AF; and

means for declaring AFL if the discriminatory metric satisfies the AFL signature metric representative of the distribution of data points during AFL.

37. The apparatus of claim 36, wherein the scatter-plot is a Lorenz plot having an origin, and the plotting step further comprises assigning an abscissa value and an ordinate value referenced to the origin of the Lorenz plot to each data point.

38. The method of claim 36, wherein the step of determining the interval comprises determining the change in interval between each succeeding ventricular heart beat from a preceding ventricular heartbeat.

39. A processor readable medium containing instructions to cause the processor to utilize ventricular signals to access atrial patterns for monitoring, diagnosis and prediction in the treatment of atrial arrhythmias comprising:

determining each interval between successive ventricular heartbeats comprising;

plotting each succeeding interval as a data point in a scatter-plot having an origin;

developing a metric signifying the degree of sparseness or clustering of the data points within the scatter-plot and with respect to the origin;

defining one or more of a sparseness discriminatory signature of data points in the scatter-plot representative of the distribution of data points with respect to origin during AF and a clustering discriminatory signature of data points in the scatter-plot representative of the distribution of data points with respect to origin during AFL; and

declaring one or more of AF or AFL if the metric exhibits the sparseness discriminatory signature representative of the distribution of data points during AF and of AFL if the metric exhibits the clustering discriminatory signature representative of the distribution of data points during AFL.

40. The medium of claim 39, wherein the scatter-plot is a Lorenz plot and the plotting step further comprises assigning an abscissa value and an ordinate value referenced to the origin of the Lorenz plot to each data point.

41. The medium of claim 40, wherein the step of determining the interval comprises determining the change in interval between each succeeding ventricular heart beat from a preceding ventricular heartbeat.