Method and apparatus for detecting a signal from a noisy environment and fetal heartbeat obtaining method

• US 5,209,237 A
• Filed: 05/24/1991
• Issued: 05/11/1993
• Est. Priority Date: 04/12/1990
• Status: Expired due to Fees

• Current Assignee: Felix Rosenthal
• Original Assignee: Felix Rosenthal
• Inventors: Rosenthal, Felix
• Primary Examiner(s): Kamm, William E.
• Application Number:
  US07/705,493
• Time in Patent Office: 718 Days
• Field of Search: 128/662.04, 128/696, 128/698, 128/702, 128/703, 128/704, 364/724.19, 364/724.2, 364/726, 367/38, 367/45
• US Class Current: 600/511
• CPC Class Codes: G06F 18/00 Pattern recognition
• CPC Class Codes: G06F 2218/04 Denoising

Abstract

To provide an accurate reproduction of a signal detected in a noisy environment, a primary signal sensor and a plurality of "signal-free" reference sensors are used. Information collected from the primary sensor and the reference sensors is compared to derive correlation data. A matrix, representing correlation data in the time domain or cross-spectral density information in the frequency domain, is constructed based on the computed correlation data. The matrix is transformed to a diagonal matrix using a singular value decomposition algorithm and a threshold value is selected with reference to one of the larger eigenvalues of the diagonalized matrix. Components of the diagonalized matrix are compared to the threshold and values greater than the threshold are used together with primary sensor correlation data to compute Wiener filter constants. The reference sensor outputs are modified by the filter constants and modified sensor outputs are subtracted from the primary sensor output. A noise-cancelled output signal is then synthesized. By selecting an appropriate threshold, the effect of redundant reference sensor contributions and the effect of small primary signal contributions to the reference sensors are eliminated.

Claims

1. A method of generating an electrocardiogram representative of a fetal heartbeat, comprising the steps of:

placing at least one primary sensor on the abdominal area of a pregnant female in proximity of the fetus to detect signals generated by the fetal heart;

placing a plurality of reference sensors in other regions of the female body to detect signals generated by sources other than the fetal heart;

detecting a primary sensor output signal from said at least one primary sensor and reference sensor output signals from each of said plurality of reference sensors;

comparing said reference sensor output signals and identifying mutually independent noise source components of said reference sensor output signals;

generating filter constants using said identified components;

modifying each of said reference sensor output signals by said filter constants to generate modified reference output signals;

generating a modified primary output signal by subtracting components corresponding to said modified reference output signals from said primary output signal; and

producing an electrocardiogram representing said modified primary output signal.

2. The method in accordance with claim 1 wherein the step of comparing comprises generating correlation data representative of differences between said reference output signals.

3. The method in accordance with claim 2 and further comprising the steps of constructing a matrix of correlation values derived from said plurality of reference output signals, converting said matrix to a diagonal matrix having elements having various values, and eliminating from said diagonal matrix elements having a values less than a predetermined threshold value.

4. The method in accordance with claim 3 wherein said threshold value is greater than primary signal contributions to said diagonal matrix elements.

5. The method in accordance with claim 3 wherein said threshold value is defined with reference to the value of at least one of said reduced matrix elements.

6. The method in accordance with claim 1 wherein said step of placing reference sensors comprising placing more reference sensors than the number of significant independent noise sources.

7. The method of generating a noise-cancelled output signal in a system comprising a primary sensor for detecting a primary signal from a primary signal source and a plurality of reference sensors for detecting noise source signals, comprising the steps of:

detecting a primary output signal from said primary sensor;

detecting a reference output signal from each of said reference sensors;

comparing said primary output signal with each of said reference output signals and generating primary comparison data;

comparing each of said reference output signals with other reference output signals and generating reference output signal comparison data;

processing said comparison data for identifying data representing mutually independent noise source signals in said reference sensor output signals;

generating filter constants based on information derived from said data representing mutually independent noise source signals, and said primary comparison data;

applying said filter constants to said reference sensor output signals to generate filtered reference sensor output signals; and

subtracting said filtered reference sensor output signals from said primary sensor output signal to generate a noise-cancelled primary signal.

8. The method in accordance with claim 7 wherein said step of comparing further comprises the steps of constructing a matrix of reference output comparison data and mathematically reducing said matrix to diagonal form to derive a reduced matrix, and wherein said step of identifying comprises the step of comparing values of components of said reduced matrix to a predetermined threshold, and wherein said step of generating filter constants comprises combining components of said reduced matrix having values greater than said predetermined threshold with said primary comparison data.

9. The method in accordance with claim 8 wherein said step of mathematically reducing said matrix of reference output comparison data comprises using a singular value decomposition algorithm.

10. The method in accordance with claim 8 wherein said matrix of reference output comparison data comprises time correlation values.

11. The method in accordance with claim 8 wherein said matrix of reference output comparison data comprises cross-spectral density values.

12. The method in accordance with claim 7 wherein said step of comparing comprises comparing a reference output at a specified time with other reference outputs at a specified different time.

13. The method in accordance with claim 7 in a system wherein said plurality of sensors comprises a number greater than the number of significant independent noise sources.

14. A method for generating a noise-cancelled primary signal in a noisy environment, comprising the steps of:

placing at least one primary sensor in a position to primarily detect said primary signal;

placing a plurality of reference sensors, at least equal in number to the number of known noise sources, in a position to detect primarily noise source signals;

detecting over a predetermined period of time a primary output signal from said at least one primary sensor and an output signal from each of said reference sensors;

defining a plurality of time delays;

comparing said primary output signal and each of said reference output signals at a specified time with all of said reference output signals at a time delayed from said specified time by said defined delay times, and generating correlation data;

constructing a matrix having as the matrix component data values representing correlation among said reference output signals;

mathematically reducing said matrix to a diagonal matrix having diagonal matrix components;

comparing said diagonal matrix components to a predetermined threshold value;

inverting certain of said diagonal matrix components having values greater than said threshold value;

computing filter constants using said inverted components and data defining correlation among said primary output and said reference output signals;

generating filtered output signals by applying said filter constants to each of said reference output signals;

subtracting said filtered output signals from said primary output signal, thereby generating data representing said noise-cancelled primary signal; and

synthesizing said noise-cancelled primary signal.

15. A method of generating a modified output signal in a system comprising at least one primary sensor for detecting signals generated by a primary signal source and a plurality of reference sensors for detecting signals from sources other than said primary source, comprising the steps of:

detecting a primary sensor output signal from said at least one primary sensor and reference sensor output signals from each of said plurality of reference sensors;

comparing said reference sensor output signals and identifying mutually independent noise source components of said reference sensor output signals;

generating filter constants using said identified components;

modifying each of said reference sensor output signals by said filter constants to generate modified reference output signals; and

generating a modified output signal by subtracting components corresponding to said modified reference output signals from said primary sensor output.

16. The method in accordance with claim 15 wherein the step of comparing comprises generating correlation data representative of differences between said reference output signals.

17. The method in accordance with claim 16 and further comprising the steps of constructing a matrix of correlation values derived from said plurality of reference output signals. converting said matrix to a diagonal matrix having elements having various values, and eliminating from said diagonal matrix elements having a values less than a predetermined threshold value.

18. The method in accordance with claim 17 wherein said threshold value is greater than primary sensor output contributions to said diagonal matrix elements.

19. The method in accordance with claim 17 wherein said threshold value is defined with reference to the value of at least one of said diagonal matrix elements.

20. The method in accordance with claim 17 wherein said step of detecting a primary sensor output signal and reference sensor output signals comprises detecting a plurality of said primary sensor output signals at predetermined time intervals and a plurality of reference sensor output signals from each of said reference sensors at said predetermined time intervals, and said step of constructing a matrix comprises constructing a matrix of output signals of a reference at each of said predefined time intervals with respect to output signals of other references at each of said predetermined time intervals.

21. Apparatus for generating an output signal representative of a signal source, comprising:

a primary sensor for detecting signals from a primary signal source and generating a primary sensor output signal;

a plurality of reference sensors, each for detecting signals from other signal sources and for generating a reference sensor output signal;

means responsive to receipt of the reference sensor output signals to identify mutually independent noise source components of the received reference sensor output signals;

means for generating filtered reference signals from received reference sensor output signals using identified mutually independent noise source components; and

means responsive to a received primary sensor output signal for generating a system output signal by subtracting from the primary sensor output signal components corresponding to the filtered reference signals.

22. The apparatus in accordance with claim 21 and further comprising means responsive to receipt of the reference sensor output signals to construct a matrix of reference signal correlation values and to generate the filtered reference signals using only matrix values greater than a predetermined threshold value.

23. The apparatus in accordance with claim 22 and further comprising a graphical display unit responsive to the system output signal for graphically displaying a representation of the system output signal.

24. The method of generating a noise-cancelled output signal in a system comprising a primary sensor for detecting a signal from a primary signal source and a plurality of reference sensors for detecting noise source signals, comprising the steps of:

detecting primary output signals from the primary sensor;

detecting reference output signals from each of the reference sensors;

storing a data record of signal values representative of values of the detected primary output signals and reference output signals;

performing a transformation into the frequency domain of the stored signal values;

comparing frequency domain values representative of output signals of each of said reference sensors with frequency domain values representative of output signals of the primary sensor and other reference sensors and generating reference output comparison values;

processing said reference output comparison values for identifying values representing mutually independent noise source signals in the reference sensor output signals;

generating filter constants based on information derived from the values representing mutually independent noise source signals and storing the filter constants;

retrieving the stored data record and performing a transformation into the frequency domain of signal values representative of the primary sensor output signals and the reference sensor output signals;

applying the stored filter constant values to the data record values to generate values representative of filtered reference sensor output signals;

generating difference values by subtracting the values representative of filtered reference sensor output signals from values representative of primary sensor output signals; and

performing an inverse transformation from the frequency domain into the time domain of the difference values to generate a time-domain representation of a noise-cancelled primary signal.

25. The method in accordance with claim 24 and further comprising the steps of comparing frequency domain values, processing reference output comparison values, and generating filter constants, repetitively, for each of a plurality of frequency values prior to the step of retrieving the stored data record.

26. An arrangement for generating an output signal representative of a primary signal source in the presence of other signal sources, comprising:

a primary sensor for detecting signals from the primary signal source and generating an analog primary sensor output signal comprising a plurality of signal components;

a plurality of reference sensors for detecting signals from the other signal sources and generating a corresponding plurality of analog reference sensor output signals comprising signal components attributable to mutually independent signal sources;

analog to digital converter circuitry connected to the primary and reference sensors for converting the analog primary sensor output signal and the reference sensor output signals to a digital primary output signal and a plurality of digital reference output signals;

signal generating apparatus connected to the converter circuitry and responsive to the plurality of reference output signals to generate a corresponding plurality of modified reference output signals, the modified reference output signals comprising only signal components corresponding to signal components other than signal components attributable to mutually independent signal sources; and

signal generating apparatus responsive to the primary output signal to modify the primary output signal and to generate a modified primary output signal having only signal components other than signal components corresponding to components of the modified reference output signals.