Method for arbitrary two-dimensional scaling of phonocardiographic signals

• US 20060155203A1
• Filed: 10/09/2003
• Published: 07/13/2006
• Est. Priority Date: 10/09/2002
• Status: Active Grant

• Current Assignee: Bang & Olufsen Medicom A/S (KOCH Industries Incorporated)
• Original Assignee: Bang & Olufsen Medicom A/S (KOCH Industries Incorporated)
• Inventors: Munk, Flemming
• Granted Patent: US 7,438,689 B2
• Time in Patent Office: Days
• US Class Current: 600/528
• CPC Class Codes: A61B 7/00 Instruments for auscultation
• CPC Class Codes: A61B 7/04 Electric stethoscopes microphones, acoustic transducers therefor H04R

Abstract

A heart signal comprising both first and second heart sounds and murmurs occurring due to various heart conditions is modelled by a sum of sinusoids automatically selected to represent the heart sound without a noise component. In an analysis step the signal parameters are measured, in an interpolation and transformation part Ω(t) and A(t) are interpolated from one window position to the next window position, and a synthesis part reconstructs the transformed heart signal. The transformations are transformations in either time scale (with unchanged frequency content) or frequency scale (with unchanged time progression), or both. Furthermore a dynamic transformation of an arbitrary heart signal to a fixed number of beats per minute is made possible.

Claims

1. A method for arbitrary time and/or frequency scaling of phonocardiographic signals obtained from a transducer and subjected to digital spectral analysis, characterised in that the signals are converted on a running basis to a sinusoidal model representation, that a time and/or frequency axis scaling is defined and used to control the amplitudes and phases of the sinusoids, which are subsequently added to create a time and/or frequency scaled representation of the phonocardiographic signal.

2. A method according to claim 1, characterised in comprising the following steps:

obtaining the frequency content of the signal by applying a Short Time Fourier Transform with overlapping segments, a Discrete Fourier Transform being performed on each segment,

performing a frequency peak search on each segment by consecutive removal of the spectral components having the highest energy content, identifying each peak by its frequency value

repeating the peak search until a maximum number of peaks have been identified or until the energy content of the last peak is below a preset minimum,

establishing a segment-by-segment map of spectral peaks, said peaks forming a track over time

optionally subjecting the frequency values of the spectral peaks to a multiplication

defining a synthesis frame [frame m] of time, based on said segments

optionally subjecting each frame to a multiplication of the time scale

adjusting the phase of sine generators centered on the frequencies of the tracks

adjusting the amplitudes of said sine generators

summing the outputs of all sine generators active at any one instant for a given frame length T.

3. A method according to claim 1, characterised in including the following steps:

obtaining the frequency content of the signal by applying a Short Time Fourier Transform with overlapping segments, a Discrete Fourier Transform being performed on each segment,

performing a frequency peak search on each segment by repeated removal of the highest spectral “

hills”

identifying each peak by its frequency value

repeating the peak search until a maximum number of peaks have been identified or until the peak level of the last peak is below a preset minimum,

establishing a segment-by-segment map of spectral peaks, said peaks forming a track over time

adjusting the phase of sine generators centered on the frequencies represented by the tracks

summing the outputs of all sine generators active at any one instant for a given frame length T

creating a continuous output signal by joining consecutive frames.

4. A method according to claim 2 for scaling a phonocardiographic signal on the frequency axis by a desired factor, characterised in that the spectral peaks are multiplied by a factor q.

5. A method according to claim 2 for scaling a phonocardiographic signal on the time axis by a desired factor, characterised in that the frame length is multiplied by a factor p.

6. A method according to claim 2 for autoscaling a phonocardiographic signal on the time axis, characterised in that the scaling factor p is set such that the frame length [T] multiplying factor is equal to the heart rate divided by 60.

7. A method according to claim 2 or 3, characterised in that the number of sine generators is equal to or below 50.

8. An apparatus for performing the method of any of the above claims on a phonocardiographic time function,

characterised in that it comprises means for windowing the time function [x(t)], short time Fourier spectrum analysis means [1], means for searching and classifying spectral peaks [2, 3], means [4] for comparing phases of signals corresponding to said spectral peaks, means for controlling the phases of sine generators [6] providing signals corresponding to said spectral peaks, means [5, 7] for controlling the amplitudes of said sine generators [6], and means [8] for summing the signals of said controlled sine generators in order to obtain a synthesized and essentially noise free output signal [y(t)] representative of said time function.