METHOD AND APPARATUS FOR RETRIEVAL OF SIMILAR HEART SOUNDS FROM A DATABASE

• US 20100049072A1
• Filed: 08/22/2008
• Published: 02/25/2010
• Est. Priority Date: 08/22/2008
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: SYEDA-MAHMOOD, TANVEER FATHIMA, Wang, Fei
• Granted Patent: US 8,137,283 B2
• Time in Patent Office: Days
• US Class Current: 600/528
• CPC Class Codes: A61B 7/00 Instruments for auscultation

Abstract

The present invention exploits a visual rendering of heart sounds and models the morphological variations of audio envelopes through a constrained non-rigid translation transform. Similar heart sounds are then retrieved by recovering the corresponding alignment transform using a variant of shape-based dynamic time warping.

Claims

1. A computer-implemented method for detecting audio similarity of heart sounds comprising:

recording a first heart sound;

pre-processing the first heart sound to create a second heart sound;

constructing a line segment approximation of the second heart sound;

defining an audio envelope around the second heart sound using the line segment approximation;

isolating a plurality of fiducial points on the audio envelope; and

matching the second heart sound to a plurality of similar heart sounds.

2. The method of claim 1, wherein the step of pre-processing comprises identifying a single periodic segment.

3. The method of claim 1, wherein the step of pre-processing the first heart sound comprises:

forming a curve f(t) by shifting g(t) by a candidate period τ

; and

defining a function R(τ

) that satisfies the periodicity condition based on the candidate period τ

as

$R<br><br>\left(\mathrm{\tau <br><br>}\right)=\frac{<br><br>\left\{g<br><br>\left(t_i\right)\right\}<br><br>}{\max <br><br>\left\{N-\mathrm{\tau <br><br>},\mathrm{\tau <br><br>}\right\}}<br><br><br><br>\mathrm{such}<br><br><br><br>\mathrm{that}<br><br><br><br><br><br>g<br><br>\left(t_i\right)-f<br><br>\left(t_j\right)<br><br>\le <br><br>\mathrm{\varepsilon <br><br>}$

wherein N is the total number of points on curve g(t), t_i and t_j are time values of matching fiducial points, and ε

is an error tolerance.

4. The method of claim 1, wherein the matching comprises:

identifying a non-rigid alignment transform based on a shape-based dynamic time warping to determine a correspondence between the fiducial points of the audio envelope and fiducial points from a database of heart sounds;

defining a measure of shape similarity as a function of a number matched fiducial points and a total number of fiducial points; and

outputting the measure of shape similarity.

5. The method of claim 4, wherein the function comprises a ratio of the number of matched fiducial points to the total number of fiducial points.

6. The method of claim 5, wherein the matching further comprises ranking the matching based on the ratio.

7. The method of claim 4, wherein the database of heart sounds comprises a database of heart sounds associated with diseases.

8. The method of claim 7, wherein the step of outputting the measure of shape similarity further comprises:

assembling disease statistics from the database of heart sounds; and

presenting the disease statistics as a decision support system showing disease labels that are associated with a plurality of heart sounds from the database of heart sounds.

9. The method of claim 4, wherein the matching further comprises time scaling and amplitude scaling.

10. The method of claim 1, wherein the fiducial points comprise local extremes of the audio envelope only only one side of a baseline of the second heart sound.

11. A computer program product comprising:

a computer usable medium having computer usable program code for detecting audio similarity of heart sounds, said computer program product including;

computer usable program code for recording a first heart sound;

computer usable program code for pre-processing the first heart sound, wherein the computer usable program code for pre-processing comprises;

program code for selecting a time duration of a portion of the first heart sound to use as a second heart sound;

computer usable program code for constructing a line segment approximation of the second heart sound;

computer usable program code for defining an audio envelope around the second heart sound using the line segment approximation;

computer usable program code for isolating a plurality of fiducial points on the audio envelope; and

computer usable program code for matching the second heart sound to a plurality of similar heart sounds, wherein the computer usable program code matching comprises;

program code for identifying a non-rigid alignment transform based on a shape-based dynamic time warping to determine a correspondence between the fiducial points of the audio envelope and fiducial points from a database of heart sounds;

program code for defining a measure of shape similarity by determining a ratio of matched fiducial points to a total number of fiducial points;

program code for ranking the matching based on the ratio.

12. A system for detecting audio similarity of heart sounds comprising:

an audio transducer for recording a first heart sound;

a processor;

wherein the processor pre-processes the first heart sound to create a second heart sound, constructs a line segment approximation of the second heart sound, defines an audio envelope around the second heart sound using the line segment approximation, and isolates a plurality of fiducial points on the audio envelope;

an input/output interface in communication with a database of heart sounds;

wherein the processor matches the second heart sound to a plurality of similar heart sounds retrieved from the database through the input-output interface.

13. The system of claim 12, wherein the processor pre-processes the first heart sound by:

forming a curve f(t) by shifting g(t) by a candidate period τ

; and

defining a function R(τ

) that satisfies the periodicity condition based on the candidate period τ

as

$R<br><br>\left(\mathrm{\tau <br><br>}\right)=\frac{<br><br>\left\{g<br><br>\left(t_i\right)\right\}<br><br>}{\max <br><br>\left\{N-\mathrm{\tau <br><br>},\mathrm{\tau <br><br>}\right\}}<br><br><br><br>\mathrm{such}<br><br><br><br>\mathrm{that}<br><br><br><br><br><br>g<br><br>\left(t_i\right)-f<br><br>\left(t_j\right)<br><br>\le <br><br>\mathrm{\varepsilon <br><br>}$

wherein N is the total number of points on curve g(t), t_i and t_j are time values of matching fiducial points, and ε

is an error tolerance.

14. The system of claim 12, wherein the processor matches the second heart sound to a plurality of heart sounds by:

identifying a non-rigid alignment transform based on a shape-based dynamic time warping to determine a correspondence between the fiducial points of the audio envelope and fiducial points from a database of heart sounds;

defining a measure of shape similarity as a function of a number matched fiducial points and a total number of fiducial points; and

outputting the measure of shape similarity.

15. The system of claim 14, wherein the function comprises a ratio of the number of matched fiducial points to the total number of fiducial points.

16. The system of claim 15, wherein the processor ranks the matching based on the ratio.

17. The system of claim 14, wherein the database of heart sounds comprises a database of heart sounds associated with diseases.

18. The system of claim 17, wherein outputting the measure of shape similarity further comprises:

assembling disease statistics from the database of heart sounds; and

presenting the disease statistics as a decision support system showing disease labels that are associated with a plurality of heart sounds from the database of heart sounds.

19. The system of claim 14, wherein the processor adjusts the second heart sound by time scaling and amplitude scaling.

20. The system of claim 12, wherein the fiducial points comprise local extremes of the audio envelope only on one side of a baseline of the second heart sound.