Denoising sensed signals from artifacts from cardiac signals

A system for artifact suppression in a sensed signal, the system comprising:

• memory;
• and
• processing circuitry coupled to the memory and configured to:
• receive the sensed signal sensed in a brain of a patient, wherein the sensed signal includes a neural signal and artifacts from a cardiac signal;
• decompose the sensed signal into a plurality of components of the sensed signal, each component having a plurality of frequencies;
• determine a first group of components, from the plurality of components of the sensed signal, that are correlated with one another, the first group of components including two or more components of the sensed signal, wherein to determine the first group of components, the processing circuitry is configured to:
• determine sets of adjacent components in the plurality of components, wherein each set of adjacent components includes sequential frequency bands, wherein the sets of adjacent components comprises at least a first set of two components having sequential frequency bands and a second set of two components having sequential frequency bands;
• determine a set of adjacent components from the sets of adjacent components having a maximum correlation, wherein the set of adjacent components includes a first component and a second component, and wherein a frequency band of the second component immediately follows a frequency band of the first component;
• determine components in the plurality of components that correlate to at least one of the first component or the second component by a threshold value determined relative to the maximum correlation;
• and
• include the determined components in the first group of components;
• determine an estimate of the cardiac signal based on the first group of components, wherein the estimate of the cardiac signal includes the cardiac signal and components of the neural signal;
• generate a denoised neural signal based on the estimate of the cardiac signal and a second group of components of the plurality of components of the sensed signal, wherein the cardiac signal is suppressed in the denoised neural signal, and wherein the second group of components excludes the first group of components;
• and
• cause delivery of deep brain stimulation (DBS) therapy based on the denoised neural signal.

A non-transitory computer-readable storage medium comprising instructions that when executed cause one or more processors to:

• receive a sensed signal sensed in a brain of a patient, wherein the sensed signal includes a neural signal and artifacts from a cardiac signal;
• decompose the sensed signal into a plurality of components of the sensed signal, each component having a plurality of frequencies;
• determine a first group of components, from the plurality of components of the sensed signal, that are correlated with one another, the first group of components including two or more components of the sensed signal, wherein the instructions that cause the one or more processors to determine the first group of components comprise instructions that cause the one or more processors to:
• determine sets of adjacent components in the plurality of components, wherein each set of adjacent components includes sequential frequency bands, wherein the sets of adjacent components comprises at least a first set of two components having sequential frequency bands and a second set of two components having sequential frequency bands;
• determine a set of adjacent components from the sets of adjacent components having a maximum correlation, wherein the set of adjacent components includes a first component and a second component, and wherein a frequency band of the second component immediately follows a frequency band of the first component;
• determine components in the plurality of components that correlate to at least one of the first component or the second component by a threshold value determined relative to the maximum correlation;
• and
• include the determined components in the first group of components;
• determine an estimate of the cardiac signal based on the first group of components, wherein the estimate of the cardiac signal includes the cardiac signal and components of the neural signal;
• generate a denoised neural signal based on the estimate of the cardiac signal and a second group of components of the plurality of components of the sensed signal, wherein the cardiac signal is suppressed in the denoised neural signal, and wherein the second group of components excludes the first group of components;
• cause delivery of deep brain stimulation (DBS) therapy based on the denoised neural signal.