Method for tachycardia discrimination

• US 5,311,874 A
• Filed: 05/18/1992
• Issued: 05/17/1994
• Est. Priority Date: 05/18/1992
• Status: Expired due to Term

• Current Assignee: Cardiac Pacemakers Incorporated (Boston Scientific Corporation)
• Original Assignee: Cardiac Pacemakers Incorporated (Boston Scientific Corporation)
• Inventors: Swanson, David K., Baumann, Lawrence S., Lang, Douglas
• Primary Examiner(s): Kamm, William E.
• Assistant Examiner(s): SCHAETZLE, KENNEDY
• Application Number:
  US07/884,770
• Time in Patent Office: 729 Days
• Field of Search: 364/413.06, 128/702, 128/705, 128/703, 128/704, 128/696
• US Class Current: 600/518
• CPC Class Codes: A61B 5/363 Detecting tachycardia or bradycardia
• CPC Class Codes: A61B 5/7239 using differentiation including higher order derivatives

Abstract

A cardiac discrimination method in which feature values of a heart related signal (e.g., cardiac biopotentials) determined to be non-baseline are compared with feature values of a heart related signal determined to be normal baseline. The feature values are extracted on an event-basis (a periodically), for obtaining feature values of a complex. For each complex, the feature values form a sequence with the value having the largest absolute value for the sequence given special identity as a fiducial point. The normal baseline complexes'"'"' characteristic sequence and a non-baseline complex'"'"'s sequence are aligned according to the fiducial points, and unoccupied positions on the ends of the sequences resulting from the alignment are filled with zeros to create normal baseline and non-baseline vectors. The similarity value and dissimilarity value of the normalized non-baseline vector with respect to the normalized normal baseline vector are determined. The type of tachycardia (VT or non-VT) and consequently the selection of appropriate therapy are determined by the location in a discrimination plane of a discrimination point, the coordinates of which are equal to the similarity and dissimilarity values. In addition, the location of a similarly generated hemodynamic discrimination point can be used to discriminate hemodynamically stable and unstable tachycardias based on the features derived from events in hemodynamic related signals. Furthermore, physiological indicators may be processed in a similar manner to determine an ideal rate at which the heart should be paced by a pacemaker.

Claims

1. A method for discriminating between tachycardias comprising the steps of:

sensing a cardiac biopotential signal comprising consecutive biopotential complexes;

determining a characteristic cycle length associated with each cardiac biopotential complex;

designating a biopotential complex to be non-baseline if the characteristic cycle length is less than a predetermined threshold and otherwise designating the complex to be baseline;

obtaining a characteristic sequence of feature values for cardiac biopotential complexes determined to be baseline for a particular patient;

obtaining a sequence of feature values for each cardiac biopotential complex determined to be non-baseline for that patient;

creating a baseline vector and creating non-baseline vectors from said characteristic sequence of feature values for complexes determined to be baseline and from each sequence of feature values for complexes determined to be non-baseline, respectively;

comparing each of the non-baseline vectors with the baseline vector; and

determining the type of tachycardia of a non-baseline complex based on the comparison of each non-baseline vector with the baseline vector.

2. The method of claim 1, and further comprising the step of accumulating comparison data related to the comparison of each of the non-baseline vectors with the baseline vector, and the step of determining the type of tachycardia based on the accumulated comparison data.

3. The method of claim 2, wherein said step of accumulating comparison date related to the comparison of each of the non-baseline vectors with the baseline vector comprises the steps of:

initializing a VT discrimination counter to a VT initial value;

initializing a non-VT discrimination counter to a non-VT initial value;

limiting the maximum value of the VT discrimination counter to a maximum VT value;

limiting the maximum value of the non-VT discrimination counter to a maximum non-VT value;

limiting the minimum value of the VT discrimination counter to a minimum VT value;

limiting the minimum value of the non-VT discrimination counter to a minimum non-VT value;

adding a VT increment value to the VT discrimination counter value and subtracting a non-VT decrement value from the non-VT discrimination counter when a complex is determined to be a non-baseline VT complex;

adding a non-VT increment value to the non-VT discrimination counter and subtracting a VT decrement value from the VT discrimination counter when a complex is determined to be a non-baseline non-VT complex; and

subtracting the VT baseline complex decrement value from the VT discrimination counter and subtracting the non-VT baseline complex decrement value from the non-VT discrimination counter when a complex is determined to be a baseline complex.

4. The method of claim 3, wherein said step of determining the type of tachycardia based on the accumulated comparison data comprises the steps of:

determining the type of tachycardia to be VT when the VT discrimination counter value is greater than a VT absolute threshold value, and the VT discrimination counter value minus the non-VT discrimination counter value is greater then a VT relative threshold value; and

determining the type of tachycardia to be non-VT when the non-VT discrimination counter value is greater than a non-VT absolute threshold value, and the non-VT discrimination counter value minus the VT discrimination counter value is greater than a non-VT relative threshold value.

5. The method of claim 1, wherein said step of obtaining a characteristic sequence of feature values for complexes determined to be baseline comprises the steps of:

creating event vectors and cluster vectors from the sequence of feature values for each complex designated baseline and from the characteristic sequence of feature values associated with each cluster of like baseline complexes, respectively;

comparing each event vector with each cluster vector;

assigning the baseline complex to a cluster based on the comparison of each event vector with each cluster vector;

defining a new cluster-by a baseline complex which cannot be assigned to any existing cluster;

calculating an average of the corresponding components of said event vector and the cluster vector for the baseline complex and the cluster to which the baseline complex is assigned to obtain an updated characteristic sequence of feature values for said cluster;

sorting the clusters in descending order of the number of events assigned to each cluster; and

defining cluster number one, the cluster to which the most complexes are assigned, to be the normal baseline cluster.

6. The method of claim 5, wherein said step of creating each event vector and each cluster vector comprises the steps of:

identifying and designating a feature value within each of said sequence of feature values for said baseline complex and said characteristic sequence of feature values for said cluster as fiducial points for the respective sequences;

aligning said baseline complex sequence and said cluster characteristic sequence to each other according to the respective fiducial points; and

adding or removing feature values at the ends of said baseline complex sequence and said cluster characteristic sequence to create said event vector and said cluster vector such that the vectors have equal numbers of components and such that corresponding components are occupied by feature values.

7. The method of claim 6, wherein said step of adding feature values at the ends of said baseline complex sequence and said cluster characteristic sequence comprises the step of filling in zeros to sequence positions which are not occupied by feature values.

8. The method of claim 6, wherein said step of identifying and designating a fiducial point for a sequence comprises the step of identifying and designating the feature value with largest absolute value in the sequence.

9. The method of claim 5, wherein said step of comparing each event vector with each cluster vector comprises the steps of:

specifying a set of linearly independent basis vectors which span a subspace of an m-dimensional vector space of which an m-component event vector and an m-component cluster vector are elements;

projecting said event vector on said subspace, thereby determining a linear combination of said basis vectors;

calculating the values of the coefficients in said linear combination; and

designating the coefficient associated with each basis vector in said linear combination to be the basis vector'"'"'s coefficient.

10. The method of claim 9, wherein said step of assigning comprises the steps of:

designating said subspace spanned by said basis vectors to be said cluster'"'"'s subspace;

defining for each of said basis vectors of said cluster'"'"'s subspace a basis vector coordinate axis with the origin of said cluster'"'"'s subspace being located at the intersection of said basis vector coordinate axes;

designating certain regions in said cluster'"'"'s subspace as regions particular to baseline complexes assigned to said cluster;

locating for each baseline complex a cluster point in said cluster'"'"'s subspace, the coordinates of said cluster point defined by said basis vectors'"'"' coefficients so that the position of said cluster point along each basis vector coordinate axis is at a coordinate relative to the origin equal to said basis vector coefficient value; and

assigning each baseline complex to the first cluster for which the location of said cluster point of the baseline complex is within one of said regions particular to said cluster.

11. The method of claim 5, wherein the baseline vector is a normal baseline vector created on the basis of the characteristic sequence of feature values for said normal baseline cluster.

12. The method of claim 1, wherein said step of creating the baseline vector and each non-baseline vector comprises the steps of:

identifying and designating a feature value within each of said characteristic baseline sequence and said non-baseline sequence as a fiducial point for the respective sequences;

creating candidate aligned sequence pairs from said characteristic baseline sequence and non-baseline sequences;

creating from each candidate aligned sequence pair a candidate baseline vector and non-baseline vector pair; and

selecting a candidate baseline vector and non-baseline vector pair as the baseline vector and the non-baseline vector.

13. The method of claim 12, wherein said step of creating a candidate aligned sequence pair comprises the step of aligning said characteristic baseline sequence and non-baseline sequence according to their respective fiducial points to create said candidate aligned sequence pair.

14. The method of claim 12, wherein said step of creating candidate aligned sequence pairs comprises the steps of:

aligning said characteristic baseline sequence and non-baseline sequence according to their respective fiducial points to create a first candidate aligned sequence pair;

aligning said characteristic baseline sequence and non-baseline sequence to each other such that said non-baseline sequence fiducial point is located one feature value to the right of said characteristic baseline sequence fiducial point to create a second candidate aligned sequence pair; and

aligning said characteristic baseline sequence and non-baseline sequence to each other such that said non-baseline sequence fiducial point is located one feature value to the left of said characteristic baseline sequence fiducial point to create a third candidate aligned sequence pair.

15. The method of claim 12, wherein said step of creating a candidate baseline vector and non-baseline vector pair from each candidate aligned sequence pair comprises the step of adding or removing feature values at the ends of said characteristic baseline sequence and non-baseline sequence comprising said candidate aligned sequence pair to create said candidate baseline vector and non-baseline vector pair such that said vectors have equal numbers of components and such that corresponding components are occupied by feature values.

16. The method of claim 15, wherein said step of adding feature values at the ends of said characteristic baseline sequence and non-baseline sequence comprises the step of filling in zeros to sequence positions which are not occupied by feature values.

17. The method of claim 12, wherein said step of selecting a candidate baseline vector and non-baseline vector pair as the baseline vector and non-baseline vector comprises the steps of:

calculating the magnitude of the vector difference between said baseline vector and non-baseline vector comprising each candidate pair; and

selecting the candidate pair with the smallest magnitude of vector difference as the baseline vector and non-baseline vector.

18. The method of claim 12, wherein said step of identifying and designating a fiducial point for a sequence comprises the step of identifying and designating a feature value with largest absolute value in the sequence.

19. The method of claim 1, wherein said step of comparing comprises the steps of:

normalizing said baseline vector and said non-baseline vector by dividing by the magnitude of the baseline vector thereby creating a normalized baseline vector and a normalized non-baseline vector;

calculating a similarity feature value of said normalized non-baseline vector with respect to said normalized baseline vector by determining the projection of said normalized non-baseline vector onto said normalized baseline vector; and

calculating a dissimilarity feature value of said normalized non-baseline vector with respect to said normalized baseline vector by determining the projection of said normalized non-baseline vector onto an axis in the plane defined by the normalized baseline vector and the normalized non-baseline vector which is perpendicular to said normalized baseline vector.

20. The method of claim 19, wherein said step of determining the type of tachycardia of a non-baseline complex comprises the steps of:

defining a discrimination plane having a similarity coordinate axis and a dissimilarity coordinate axis which is orthogonal to said similarity coordinate axis, the origin of said discrimination plane being located at an intersection of said similarity coordinate axis and said dissimilarity coordinate axis;

designating certain regions in said discrimination plane as regions particular to certain tachycardias;

locating for each non-baseline complex a discrimination point in said discrimination plane, the coordinates of said discrimination point defined by said similarity feature value and said dissimilarity feature value so that the position of said discrimination point is along said similarity axis at a rectangular coordinate relative to the origin equal to said similarity value and along said dissimilarity axis at a rectangular coordinate relative to the origin equal to said dissimilarity value; and

determining the type of tachycardia of the non-baseline complex based on the location of said discrimination point in said discrimination plane.

21. The method of claim 1, wherein said step of comparing comprises the steps of:

specifying a set of linearly independent basis vectors which span a subspace of an m-dimensional vector space of which an m-component non-baseline vector and an m-component baseline vector are elements;

projecting said non-baseline vector on said subspace thereby determining a linear combination of said basis vectors;

calculating the values of the coefficients in said linear combination; and

terming the coefficient associated with each basis vector in said linear combination to be the basis vector'"'"'s coefficient.

22. The method of claim 21, wherein said step of determining the type of tachycardia of a non-baseline complex comprises the steps of:

designating said subspace spanned by said basis vectors to be a discrimination subspace;

defining for each of said basis vectors of said discrimination subspace a basis vector coordinate axis with the origin of said discrimination subspace being located at the intersection of said basis vector coordinate axes;

designating certain regions in said discrimination subspace as regions particular to certain tachycardias;

locating for each non-baseline complex a discrimination point in said discrimination subspace, the coordinates of said discrimination point defined by said basis vectors'"'"' coefficients so that the position of said discrimination point along each basis vector coordinate axis is at a coordinate relative to the origin equal to said basis vector coefficient value; and

determining the type of tachycardia of the non-baseline complex based on the location of said discrimination point in said discrimination subspace.