Non-local mean filtering for electrophysiological signals

Patent metadata

Field	Value
Publication number	US-11324433-B2
Application number	US-201815965295-A
Country	US
Kind code	B2
Filing date	Apr 27, 2018
Priority date	Jan 17, 2013
Publication date	May 10, 2022
Grant date	May 10, 2022

Abstract

A method can include storing input electrical signal data representing at least a given electrophysiological signal acquired from a patient. A non-local mean filter can be applied to the given electrophysiological signal, the non-local mean filter including a spatial filter component and an intensity filter component. The method can also include controlling parameters to establish weighting of each of the spatial filter component and the intensity filter component in response to a control input. Filtered signal data can be stored based on the applying and the controlling.

First claim

What is claimed is:

1. A non-transitory computer-readable medium having instructions executable by a processor, the instructions programmed to perform a method comprising:
applying a non-local mean filter to a given electrophysiological signal measured via at least one electrode from a patient thereby providing a first recovered signal and a first residual signal;
applying the non-local mean filter to one of the first recovered signal and the first residual signal and thereby providing a feature signal and a second residual signal; and
combining the first recovered signal with the feature signal and thereby providing a filtered version of the given electrophysiological signal.

2. The medium of claim 1 , wherein the non-local mean filter comprises a spatial domain filter to spatially weight the given electrophysiological signal relative to electrophysiological signals provided from neighboring electrodes according to spatial parameter data during the application of the non-local mean filter.

3. The medium of claim 2 , wherein the non-local mean filter further comprises an intensity filter to weight based on a relative magnitude of a sample point of the given electrophysiological signal relative to corresponding sample points of the electrophysiological signals according to intensity parameter data during the application of the non-local mean filter.

4. The medium of claim 3 , further comprising configuring the spatial parameter data and the intensity parameter data to be the same for each application of the non-local mean filter to the given electrophysiological signal.

5. The medium of claim 3 , further comprising configuring the spatial parameter data and the intensity parameter data to be different for each application of the non-local mean filter to the given electrophysiological signal.

6. The medium of claim 3 , further comprising configuring at least one of the spatial parameter data and the intensity parameter data in response to a control input.

7. The medium of claim 3 , the method further comprising:
analyzing at least one of the first recovered signal and the first residual signal;
selectively configuring at least one of the spatial parameter data and the intensity parameter data based on the analyzing.

8. The medium of claim 1 , wherein the given electrophysiological signal comprises one of a plurality of electrophysiological signals representing electrical activity associated with a geometric surface of patient tissue,
wherein each of the applying and the combining are performed with respect to at least a substantial portion of each of the plurality of electrophysiological signals.

9. The medium of claim 8 , wherein each of the plurality of electrophysiological signals are derived at least one of invasively acquired electrical measurements for the patient or non-invasively acquired electrical data for the patient.

10. The medium of claim 1 , further comprising:
identifying at least one transient feature in the given electrophysiological signal;
determining a time window associated with each identified transient feature; and
selectively applying filtering to at least one of the given electrophysiological signal, the first recovered signal and the first residual signal according to the determined time window and thereby removing each identified transient feature,
wherein selectively applying filtering further comprises one of applying the non-local mean filter configured for removal of transient features or applying a smoothing function.

11. The medium of claim 10 ,
wherein the non-local mean filter applied to the given electrophysiological signal is configured for denoising the given electrophysiological signal such that the first recovered signal comprises a denoised signal that includes the at least one transient feature; and
wherein selectively applying filtering is performed on the denoised signal.

12. The medium of claim 1 , wherein the method further comprises generating a graphical output based on the filtered version of the given electrophysiological signal.

13. The medium of claim 1 , wherein for a given application the non-local mean filter is configured according to a first set of filter parameters and for another application the non-local mean filter is configured according to a second set of filter parameters, wherein the first set of filter parameters is different from the second set of filter parameters.

14. A system comprising:
a plurality of sensors configured to measure electrophysiological signals from locations distributed across tissue associated with a patient;
memory configured to store machine readable instructions and the measured electrophysiological signals;
at least one processor configured to access the memory and configured to execute the machine readable instructions, the machine readable instructions comprising:
a non-local mean filter comprising a spatial filter and an intensity filter configured to adaptively average neighboring samples of the electrophysiological signals in a neighborhood for each sample of a given electrophysiological signal; and
a filter control configured to:
set spatial parameter data to configure the spatial filter according to a distance in a sampling space for the given electrophysiological signal, and
set intensity parameter data to configure weighting of the intensity filter according to an intensity difference between samples in the given electrophysiological signal.

15. The system of claim 14 , wherein the filter control is further configured to apply the non-local mean filter to the given electrophysiological signal to provide a recovered signal and a residual signal, and wherein the system further comprises a signal analysis function configured to analyze at least one of the recovered signal and the residual signal, the filter control being further configured to set at least one of the spatial parameter data and the intensity parameter data based on the signal analysis.

16. The system of claim 15 , wherein the signal analysis further comprises a spike detector configured to identify at least one transient feature in the given electrophysiological signal, the filter control to selectively apply filtering to the signal in a region of each identified transient feature in the given electrophysiological signal.

17. The system of claim 16 , wherein the signal analysis further comprises a spatial region calculator configured to determine the region of each identified transient feature as a time window associated with each identified transient feature.

18. The system of claim 17 , wherein the filter control is configured to selectively apply the filtering as one of the non-local mean filter configured to remove transient features or another filter function.

19. The system of claim 15 , wherein the signal analysis further comprises a residual analysis function to analyze the residual signal, the filter control being further configured to control a number of iterations that the non-local mean filter is applied to the given electrophysiological signal based on the residual analysis function.

20. A system comprising:
memory to store machine readable instructions and store input electrical signal data representing at least a given electrophysiological signal acquired from a patient via a respective sensor; and
at least one processor to access the memory and execute the machine readable instructions configured to perform a method, the method comprising:
applying a non-local mean filter to the given electrophysiological signal, the non

Assignees

• Cardioinsight Technologies Inc

Inventors

• Zeng Qingguo
• Vasudevan Venkatesh
• Ramanathan Charulatha
• Jia Ping

Classifications

• A61B5/316 Primary

  Modalities, i.e. specific diagnostic methods · CPC title

• A61B5/307 Primary

  specially adapted for particular uses · CPC title

• A61B5/7203 Primary

  for noise prevention, reduction or removal · CPC title

• A61B5/30

  Input circuits therefor · CPC title