Systems and methods for filtering a signal using a continuous wavelet transform

• US 8,385,675 B2
• Filed: 10/10/2008
• Issued: 02/26/2013
• Est. Priority Date: 07/15/2008
• Status: Active Grant

• Current Assignee: Nellcor Puritan Bennett Ireland
• Original Assignee: Nellcor Puritan Bennett Ireland
• Inventors: Watson, James Nicholas, Addison, Paul Stanley
• Primary Examiner(s): ZARKA, DAVID PETER
• Application Number:
  US12/249,100
• Publication Number:
  US 20100014725A1
• Time in Patent Office: 1,600 Days
• Field of Search: 382/261
• US Class Current: 382/261
• CPC Class Codes: A61B 5/14551 for measuring blood gases
• CPC Class Codes: A61B 5/7203 for noise prevention, reduction or removal
• CPC Class Codes: A61B 5/7207 of noise induced by motion artifacts
• CPC Class Codes: A61B 5/726 using Wavelet transforms
• CPC Class Codes: G06F 2218/04 Denoising

Abstract

According to embodiments, systems and methods for reducing noise in a signal are provided. A signal may be transformed using a continuous wavelet transform and a corresponding scalogram may be generated. Regions of noise may be identified from the resulting scalogram. These regions may be masked by, for example, removing, altering, or appropriately tagging the regions. After masking the regions of noise, the scalogram may be converted to a filtered signal using an inverse wavelet transform. Alternatively or additionally, desirable regions of non-noise may instead be identified from the resulting scalogram. These desirable regions may be extracted from the scalogram and an inverse wavelet transform performed on the extracted regions in order to generate a filtered signal.

Claims

1. A method for filtering an input signal, the method comprising:

performing, using processor circuitry, a continuous wavelet transform of an input signal to produce a transformed signal;

generating a scalogram based, at least in part, on the transformed signal;

identifying at least one region of noise in the scalogram based, at least in part, on a sliding-wedge window, wherein the sliding-wedge window analyzes wedge-shaped areas of the scalogram, wherein the wedge-shaped areas span a plurality of scales in the scalogram, and wherein the width of the wedge-shaped areas at a first of the plurality of scales is different than the width of the wedge-shaped areas at a second of the plurality of scales; and

generating a filtered signal based, at least in part, on the at least one region of noise.

2. The method of claim 1, wherein generating a filtered signal comprises:

tagging the at least one region of noise for later processing.

3. The method of claim 1, wherein generating a filtered signal comprises:

removing the at least one region of noise from the scalogram; and

performing an inverse continuous wavelet transform on the scalogram.

4. The method of claim 3, wherein generating a filtered signal further comprises:

creating side skirts in the scalogram in an area surrounding the at least one region of noise.

5. The method of claim 1, further comprising:

identifying desired, non-noise components of the scalogram; and

wherein generating a filtered signal comprises extracting the desired, non-noise components.

6. The method of claim 5, wherein identifying desired, non-noise components of the scalogram further comprises:

identifying particular regions of the scalogram that correspond to physiological operations in the input signal; and

identifying the particular regions as desired components of the input signal.

7. A method for filtering an input signal, the method comprising:

performing, using processor circuitry, a continuous wavelet transform of an input signal to produce a transformed signal;

generating a scalogram based, at least in part, on the transformed signal;

identifying at least one region of noise in the scalogram based, at least in part, on a sliding-wedge window, wherein the sliding-wedge window analyzes wedge-shaped areas of the scalogram, and wherein identifying the at least one region of noise in the scalogram comprises;

comparing energy in a wedge-shaped area of the scalogram to a threshold value;

determining that the energy in the wedge-shaped area of the scalogram is greater than the threshold value; and

identifying the wedge-shaped area of the scalogram as a region of noise in response to the determining; and

generating a filtered signal based, at least in part, on the at least one region of noise.

8. A method of filtering an input signal, the method comprising:

performing, using processor circuitry, a continuous wavelet transform of an input signal to produce a transformed signal;

generating a scalogram based, at least in part, on the transformed signal;

identifying at least one region of noise in the scalogram;

identifying a repeating pattern in an area surrounding the at least one region of noise; and

replacing values of the scalogram in the at least one region of noise with values that are consistent with the repeating pattern.

9. A system for filtering an input signal, the system comprising:

a sensor configured to detect an input signal; and

a processor configured to;

perform a continuous wavelet transform of the input signal to produce a transformed signal;

generate a scalogram based, at least in part, on the transformed signal;

identify at least one region of noise in the scalogram based, at least in part, on a sliding-wedge window, wherein the sliding-wedge window analyzes wedge-shaped areas of the scalogram, wherein the wedge-shaped areas span a plurality of scales in the scalogram, and wherein the width of the wedge-shaped areas at a first of the plurality of scales is different than the width of the wedge-shaped areas at a second of the plurality of scales; and

generate a filtered signal based, at least in part, on the at least one region of noise.

10. The system of claim 9, wherein the processor is further configured to:

tag the at least one region of noise for later processing.

11. The system of claim 9, wherein the processor is further configured to:

remove the at least one region of noise from the scalogram; and

perform an inverse continuous wavelet transform on the scalogram.

12. The system of claim 11, wherein the processor is further configured to:

create side skirts in the scalogram in an area surrounding the at least one region of noise.

13. The system of claim 9, wherein the processor is further configured to:

identify desired, non-noise components of the scalogram; and

extract the desired, non-noise components.

14. The system of claim 13, wherein the processor is further configured to:

identify particular regions of the scalogram that correspond to physiological operations in the input signal; and

identify the particular regions as desired components of the input signal.

15. A system for filtering an input signal, the system comprising:

a sensor configured to detect an input signal; and

a processor configured to;

perform a continuous wavelet transform of the input signal to produce a transformed signal;

generate a scalogram based, at least in part, on the transformed signal;

identify at least one region of noise in the scalogram based, at least in part, on a sliding-wedge window, wherein the sliding-wedge window analyzes wedge-shaped areas of the scalogram, and wherein the at least one region of noise is identified by;

comparing energy in a wedge-shaped area of the scalogram to a threshold value;

determining that the energy in the wedge-shaped area of the scalogram is greater than the threshold value; and

identifying the wedge-shaped area of the scalogram as a region of noise in response to determining that the energy in the wedge-shaped area of the scalogram is greater than the threshold value; and

generate a filtered signal based, at least in part, on the at least one region of noise.

16. A system for filtering an input signal, the system comprising:

a sensor configured to detect an input signal; and

a processor configured to;

perform a continuous wavelet transform of the input signal to produce a transformed signal;

generate a scalogram based, at least in part, on the transformed signal;

identify at least one region of noise in the scalogram;

identify a repeating pattern in the scalogram in the area surrounding the at least one region of noise; and

replace values of the scalogram in the at least one region of noise with values that are consistent with the repeating pattern.

17. A non-transitory computer-readable medium for use in filtering an input signal, the non-transitory computer-readable medium having computer program instructions recorded thereon for:

performing a continuous wavelet transform of an input signal to produce a transformed signal;

generating a scalogram based, at least in part, on the transformed signal;

identifying at least one region of noise in the scalogram based, at least in part, on a sliding-wedge window, wherein the sliding-wedge window analyzes wedge-shaped areas of the scalogram, wherein the wedge-shaped areas span a plurality of scales in the scalogram, and wherein the width of the wedge-shaped areas at a first of the plurality of scales is different than the width of the wedge-shaped areas at a second of the plurality of scales; and

generating a filtered signal based, at least in part, on the at least one region of noise.

18. A method for filtering an input signal, the method comprising:

performing, using processor circuitry, a continuous wavelet transform of an input signal to produce a transformed signal;

generating a scalogram based, at least in part, on the transformed signal;

identifying at least one region of interest in the scalogram based, at least in part, on a sliding-wedge window, wherein the sliding-wedge window analyzes wedge-shaped areas of the scalogram, wherein the wedge-shaped areas span a plurality of scales in the scalogram, and wherein the width of the wedge-shaped areas at a first of the plurality of scales is different than the width of the wedge-shaped areas at a second of the plurality of scales; and

performing an inverse continuous wavelet transform of the at least one region of interest.