Systems and methods for noise reduction in imaging

By estimating and correcting for noise outside the imaging band using a reference receiver circuit and adaptive filtering, the method addresses the challenge of low signal-to-noise ratios in medical imaging, improving image quality in ultrasound and MRI technologies.

JP2024156743AActive Publication Date: 2024-11-06SUNNYBROOK RES INST
View PDF 24 Cites 0 Cited by

Patent Information

Application Number
JP2024119350
Authority / Receiving Office
JP · JP
Patent Type
Applications
Current Assignee / Owner
Priority Date
2017-02-24
Filing Date
2024-07-25
Publication Date
2024-11-06
Estimated Expiration
2038-02-23

AI Technical Summary

Technical Problem

Medical imaging technologies such as ultrasound and MRI face challenges in maintaining image quality due to low signal-to-noise ratios, particularly in minimally invasive procedures where ultrasound transducers detect weak electrical signals that are prone to noise from various sources, including electromagnetic interference and system components, making it difficult to effectively filter and remove noise without compromising signal strength.

Method used

The method involves estimating broadband noise outside the imaging band to compensate for in-band noise by using a reference receiver circuit or out-of-band noise detection, applying adaptive filtering, and employing pattern recognition to dynamically suppress noise, thereby enhancing the signal-to-noise ratio and improving image quality.

Benefits of technology

This approach effectively reduces noise in imaging signals, enhancing the signal-to-noise ratio and improving the quality of medical images by accurately estimating and correcting for noise within the imaging band, even in the presence of broadband noise.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 2024156743000001_ABST
    Figure 2024156743000001_ABST
Patent Text Reader

Abstract

To provide systems and methods for the denoising of images in the presence of broadband noise based on the detection and / or estimation of in-band noise.SOLUTION: According to various example embodiments, an estimate of broadband noise that lies within the imaging band is made by detecting or characterizing the out-of-band noise that lies outside of the imaging band. This estimated in-band noise may be employed for denoising the detected imaging waveform. According to other example embodiments, a reference receive circuit that is sensitive to noise within the imaging band, but is isolated from the imaging energy, may be employed to detect and / or characterize the noise within the imaging band. The estimated reference noise may be employed to denoise the detected in-band imaging waveform.SELECTED DRAWING: Figure 1A
Need to check novelty before this filing date? Find Prior Art

Description

[Technical field]

[0001] <CROSS REFERENCE TO RELATED APPLICATIONS> This application claims priority to U.S. Provisional Application No. 62 / 463,431, entitled “SYSTEMS AND METHODS FOR NOISE REDUCTION IN IMAGING,” filed February 24, 2017, the entire contents of which are incorporated herein by reference.

[0002] The present disclosure relates to image processing techniques and to processing image data to remove noise. [Background technology]

[0003] Medical imaging, both ultrasound and MRI imaging, relies on detecting low amplitude signals in the radio frequency spectrum, which typically ranges from 2 MHz to 200 MHz. Image quality is heavily affected by the signal-to-noise ratio.

[0004] In intravascular ultrasound (IVUS), intracardiac echocardiography (ICE), and other forms of minimal invasion, an ultrasound transducer detects ultrasound signals from surrounding structures and converts the acoustic energy into an electrical signal. This signal propagates along one or more conductive channels (e.g., coaxial conductors, twisted pair conductors, flex circuits, etc.). For many reasons (including cost, manufacturability, safety, biocompatibility, thermal issues, and power supply requirements), a minimally invasive imaging probe that can be inserted into the body may not have amplifiers to increase the signal strength. The electrical signals detected by minimally invasive ultrasound transducers can be very small (<10 mV, typically <1 mV), and most of the information about tissue structures that can be imaged by ultrasound is often on the low side of the dynamic range of the detected electrical signal. The signal amplitude of the received ultrasound signal is limited by any or all of the following: the mechanical efficiency of the transducer, the small amplitude of the detected acoustic signal, the small size of the transducer, or attenuation along the conductors that carry the electrical signal from the transducer to the body. From this perspective, the signals of minimally invasive ultrasound imaging systems tend to be very weak.

[0005] Noise can be introduced into the system from many sources, including radio transmitters, power electronics, transmission lines, switching transistors, and others known in the art. Noise can be introduced by induction, direct conduction, or suboptimal isolation between components sensitive to electromagnetic interference. Some noise can be generated by components within the imaging system itself, such as scan actuators, motor controller pulse width modulators, switching mode power components, clock circuits, transistors, and the like in any electronic components of the imaging system. Additionally, other systems connected to the subject or in the treatment environment can introduce noise, such as impedance monitors, tracking systems (e.g., Carto® 3, Carto® XP, NavX™ systems), temperature sensors, infusion pumps, ablation systems, ECGs, hemodynamic monitors. RFID inventory control systems used in the medical field can also introduce noise.

[0006] There are approaches to reducing the amount of noise that gets into the ultrasound receiver circuitry of an ultrasound imaging system, such as selecting components in the system that produce minimal RF noise, electrical isolation, shielding, proper grounding, and physically separating noise-producing components from components that are susceptible to electromagnetic noise. These approaches can be difficult to implement because noise sources may have desirable characteristics for other reasons (e.g., pulse-width modulated motor controllers are energy efficient and have good response times) or are difficult to physically isolate from one another (e.g., it is desirable to place power electronics in close proximity to the imaging probe or its associated circuitry).

[0007] Other approaches to reducing the effect of noise on ultrasound signal quality (ultrasound image quality) include filtering and image processing. Ultrasound signals have a known bandwidth, and the detected ultrasound signal is filtered using analog or digital filtering techniques (sometimes a combination of the two). Analog or digital filtering can be applied to limit a portion of the electrical signal output from the ultrasound receiver circuitry to frequency content (or its harmonic content) within the operating bandwidth of the ultrasound transducer. By selecting a filter with a narrow band and sharp cutoff, the amount of noise allowed in the signal used for imaging or ultrasound signal applications (e.g., Doppler measurement spectrum analysis of ultrasound signals, or evaluation of scattering flow in an ultrasound field). Notch or comb filters are useful for removing narrowband noise within the imaging range frequencies. Excessive filtering can have the unwanted effect of reducing the amount of signal power allowed in imaging or other applications of the ultrasound signal, and can also have a detrimental effect on other aspects of the ultrasound imaging system's performance, such as resolution. However, if the filter's passband is too large, the system will receive more noise.

[0008] Image processing can further reduce noise by filtering the generated image data, for example by averaging or outlier removal. Filtering can be applied in the image in the spatial domain, for example by applying a Gaussian filter to a pixel and its neighboring pixels to smooth out blurring and random noise in the image, but this tends to reduce the spatial resolution of the image. Similarly, spatial domain filtering can be applied to imaging structures that are not fast moving relative to the frame repetition frequency of the imaging device. For example, a pixel in an image frame is a pixel at a similar location in a previous and / or subsequent frame that is the result of averaging or Gaussian filtering.

[0009] A similar problem exists in MRI imaging systems, where a weak signal is detected in the presence of noise from unwanted sources of radio frequency energy.

[0010] What would be useful are methods, systems, and devices for identifying and dynamically removing noise from one or more imaging signals.

[0011] Many noises can get into the received ultrasound signal and be difficult to remove once in the system. This is especially true when the noise is broadband in nature, with some of the noise being within the passband of the ultrasound system. For example, in an imaging system with a transducer having a center frequency of 10 MHz and a passband of 7.5 to 12.5 MHz, the system is designed to filter out the noise portion below 7.5 MHz and above 12.5 MHz. However, the noise in the 7.5 to 12.5 MHz band is often significant relative to the amplitude of the ultrasound signal to be detected.

[0012] Many noise sources arise as a result of fast transitions, for example when field effect transistors or switches turn on and off. Fast transitioning electrical signals have a very wide frequency domain, which can easily cover all or part of the passband of the ultrasonic receiving signal chain. This is especially true in power circuits and pulse width modulation circuits, where the noise has a strong enough amplitude relative to the detection signal. Summary of the Invention [Means for solving the problem]

[0013] One approach to reducing broadband noise takes advantage of the fact that imaging energy lies primarily within a selectable imaging band, and broadband noise can be detected both within the imaging band and outside the imaging band. In principle, by detecting or characterizing noise outside the imaging band at any time, one can estimate the broadband noise that may be present within the imaging band and modify the detected signal to reduce the estimated in-band noise. By effectively estimating the in-band noise based on the out-of-band noise, one can generate a signal that estimates the desired imaging energy without the estimated in-band noise present.

[0014] Another approach to reducing noise in the imaging band is to use a reference receive circuit (comprising resistors, capacitors, inductors, transmission lines, amplifiers, transformers, inert transducers or components capable of emulating a transducer receive circuit) that can detect noise in the imaging band but is isolated from the imaging energy. By estimating the in-band imaging noise based on the in-band noise received by the reference receive circuit, a signal can be generated that estimates the desired imaging energy without the estimated in-band noise.

[0015] In one aspect, a method for removing noise from a detected imaging signal in the presence of broadband noise is provided, the method comprising: obtaining a noise characteristic discrimination waveform by detecting energy waves with an imaging transducer receive circuit when no imaging energy is being received, and filtering the noise characteristic discrimination waveform to generate an in-band noise characteristic discrimination waveform that is within an imaging band and an out-of-band noise characteristic discrimination waveform that is at least partially within a noise detection band outside the imaging band; segmenting the in-band noise characteristic identification waveform and the out-of-band noise characteristic identification waveform according to one or more time windows; processing the in-band noise characteristic identification waveform and the out-of-band noise characteristic identification waveform for at least one time window to determine a relationship between noise in the imaging band and noise in the noise detection band; obtaining one or more imaging waveforms by detecting an imaging signal by the imaging transducer receiving circuit; For at least one imaging waveform, a) filtering the imaging waveform to generate an in-band imaging waveform present within the imaging band and an out-of-band noise detection imaging waveform present within the noise detection band; b) segmenting the in-band imaging waveform and the out-of-band noise detection imaging waveform according to a time window of one shape; c) using the relationship and the out-of-band noise detected imaging waveform to estimate a measure of the amount of noise in the in-band imaging waveform within at least one time window; d) for at least one time window in step c), applying a noise suppression correction to a portion of the in-band imaging waveform within the time window; has.

[0016] In one aspect, a method for removing noise from a detected imaging signal in the presence of noise is provided, the method comprising: When no imaging energy is being received, obtaining a noise characteristic discrimination waveform by detecting energy waves with an imaging transducer receiving circuit, and filtering the noise characteristic discrimination waveform to generate an in-band noise characteristic discrimination waveform that is within an imaging band; obtaining a reference noise characteristic identification waveform by detecting noise with a reference receive circuit configured to not transmit imaging energy while the imaging transducer receive circuit detects noise; processing the in-band noise characteristic identification waveform and the reference noise characteristic identification waveform to determine a relationship between noise in the imaging band and noise detected by the reference receiving circuit; obtaining one or more imaging waveforms by detecting an imaging signal by the imaging transducer receiving circuit; For at least one imaging waveform, a) filtering the imaging waveform to generate an in-band imaging waveform that resides within the imaging band; b) detecting a reference noise detection waveform by the reference receiving circuit; c) segmenting the in-band imaging waveform and the reference noise detection waveform according to one or more time windows; d) using the relationship and the reference noise detection waveform to estimate a measure of the amount of noise in the in-band imaging waveform within at least one time window; e) for at least one time window in step c), applying a noise suppression correction to a portion of the in-band imaging waveform within the time window; has.

[0017] In one aspect, a method for removing noise from a detected imaging signal in the presence of broadband noise is provided, the method comprising: obtaining an imaging waveform by detecting energy waves with an imaging transducer receive circuitry and filtering the imaging waveform to generate an in-band imaging waveform that resides within an imaging band and an out-of-band noise detection imaging waveform that resides at least in part within a noise detection band that is outside the imaging band; detecting an in-band imaging envelope of the in-band imaging waveform; detecting an out-of-band envelope of the out-of-band noise detected imaging waveform; applying a scaling factor to the out-of-band envelope to obtain a modified out-of-band envelope; combining the modified out-of-band envelope and the in-band imaging envelope to obtain a noise-corrected in-band envelope; having The scaling factor is selected to reduce the in-band noise contribution of the noise-corrected in-band envelope.

[0018] In one aspect, a method for removing noise from a detected imaging signal when noise is present is provided, the method comprising: obtaining an imaging waveform by detecting energy waves with an imaging transducer receive circuitry and filtering the imaging waveform to generate an in-band imaging waveform that resides within an imaging band and an out-of-band noise detection imaging waveform that resides at least in part within a noise detection band that is outside the imaging band; applying an amplitude scaling factor and a frequency shift to the out-of-band noise detected imaging waveform to obtain a modified waveform, such that the modified waveform includes frequency components that are within the imaging band; combining the modified waveform with the in-band captured waveform to obtain a noise-corrected in-band captured waveform; having The amplitude scaling factor is selected to reduce the in-band noise contribution of the noise-corrected in-band imaging waveform.

[0019] In one aspect, a method for removing noise from a detected imaging signal when noise is present is provided, the method comprising: obtaining an imaging waveform by detecting energy waves with imaging transducer receive circuitry and filtering the imaging waveform to generate an in-band imaging waveform that resides within an imaging band; obtaining a reference noise detection waveform by detecting noise with a reference receive circuit configured to not transmit imaging energy while detecting noise received by the imaging transducer receive circuit; detecting an in-band imaging envelope of the in-band imaging waveform; detecting a reference envelope of the reference noise detection waveform; applying a scaling factor to the reference envelope to obtain a modified reference envelope; combining the modified reference envelope and the in-band imaging envelope to obtain a noise-corrected in-band envelope; having The scaling factor is selected to reduce the in-band noise contribution of the noise-corrected in-band envelope.

[0020] In one aspect, a method for removing noise from a detected imaging signal when noise is present is provided, the method comprising: obtaining an imaging waveform by detecting energy waves with imaging transducer receive circuitry and filtering the imaging waveform to generate an in-band imaging waveform that resides within an imaging band; obtaining a reference noise detection waveform by detecting noise with a reference receive circuit configured to not transmit imaging energy while detecting noise received by the imaging transducer receive circuit; adaptively filtering the reference noise detection waveform according to one or more adaptive filter parameters; combining the filtered reference noise detection waveform with the in-band imaging waveform to obtain a noise-corrected in-band imaging waveform; having The adaptive filter parameters are dynamically determined by processing the noise-corrected in-band imaging waveform to minimize a power of the noise-corrected in-band imaging waveform.

[0021] In one aspect, a method for removing noise from a detected imaging signal in the presence of broadband noise is provided, the method comprising: obtaining an imaging waveform by detecting energy waves with an imaging transducer receive circuitry and filtering the imaging waveform to generate an in-band imaging waveform that resides within an imaging band and an out-of-band noise detection imaging waveform that resides at least in part within a noise detection band that is outside the imaging band; detecting an in-band imaging envelope of the in-band imaging waveform; detecting an out-of-band imaging envelope of the out-of-band noise detected imaging waveform; adaptively filtering the out-of-band imaging envelope according to one or more adaptive filter parameters; combining the filtered out-of-band imaging envelope and the in-band imaging envelope to obtain a noise-corrected in-band imaging envelope; having The adaptive filter parameters are dynamically determined by processing the noise-corrected in-band imaging envelope to minimize a power of the noise-corrected in-band imaging envelope.

[0022] In one aspect, a method for removing noise from a detected imaging signal in the presence of broadband noise is provided, the method comprising: obtaining an imaging waveform by detecting energy waves with an imaging transducer receive circuitry and filtering the imaging waveform to generate an in-band imaging waveform that resides within an imaging band and an out-of-band noise detection imaging waveform that resides at least in part within a noise detection band that is outside the imaging band; applying a frequency shift to the out-of-band noise detection imaging waveform to obtain a modified waveform, such that the modified waveform includes frequency components that are within the imaging band; adaptively filtering the out-of-band noise detected imaging waveform according to one or more adaptive filter parameters; combining the filtered modified waveform with the in-band captured waveform to obtain a noise corrected in-band captured waveform; having The adaptive filter parameters are dynamically determined by processing the noise-corrected in-band imaging waveform to minimize a power of the noise-corrected in-band imaging waveform.

[0023] In one aspect, a method for removing noise from a detected imaging signal in the presence of broadband noise is provided, the method comprising: obtaining an imaging waveform by detecting energy waves with an imaging transducer receive circuitry and filtering the imaging waveform to generate an in-band imaging waveform that resides within an imaging band and an out-of-band noise detection imaging waveform that resides at least in part within a noise detection band that is outside the imaging band; selecting appropriate filter parameters for a dynamic digital filter for filtering the in-band imaging waveform to remove in-band noise by processing the out-of-band noise detection imaging waveform; filtering the in-band imaging waveform with the dynamic digital filter according to the filter parameters; has.

[0024] In one aspect, a method is provided for performing noise suppression on a signal acquired by a detection system characterized by one or more noise sources, the method comprising: obtaining a noise characteristic discrimination waveform by detecting energy waves with an imaging transducer receive circuit when no imaging energy is being received, and filtering the noise characteristic discrimination waveform to generate an in-band noise characteristic discrimination waveform that resides within an imaging band and an out-of-band noise characteristic discrimination waveform that resides within a noise detection band that is at least partially outside of the imaging band; segmenting the in-band noise characteristic identification waveform and the out-of-band noise characteristic identification waveform according to one or more time windows; processing the in-band noise characteristic identification waveform and the out-of-band noise characteristic identification waveform according to a pattern recognition algorithm for at least one time window to identify noise patterns within the noise detection band that correlate with noise in the imaging band; detecting an imaging signal by the imaging transducer receiving circuit to obtain an imaging waveform, and filtering the imaging waveform to obtain an in-band imaging waveform present in the imaging band and an out-of-band noise detection imaging waveform present in the noise detection band; segmenting the in-band imaging waveform and the out-of-band imaging waveform according to one or more time windows; processing the out-of-band noise detection imaging waveform for at least one time window according to the pattern recognition algorithm to detect the noise pattern; if the noise pattern is detected, applying a noise suppression correction to the time window of the in-band imaging waveform specific to the noise pattern detected in the out-of-band noise detection imaging waveform; has.

[0025] In one aspect, a method is provided for performing noise suppression on a signal detected by a detection system characterized by one or more known noise sources, the method comprising: When no imaging energy is being received, obtaining a noise characteristic discrimination waveform by detecting energy waves with an imaging transducer, and filtering the noise characteristic discrimination waveform to generate an in-band noise characteristic discrimination waveform present within an imaging band; obtaining a reference noise characteristic identification waveform by detecting noise with a reference receive circuit configured to not transmit imaging energy while detecting noise received by the imaging transducer receive circuit; segmenting the in-band noise characteristic identification waveform and the reference noise characteristic identification waveform according to one or more time windows; processing the in-band noise characteristic identification waveform and the reference noise characteristic identification waveform for at least one time window to determine a relationship between noise in the imaging band and noise detected by the reference receiving circuit; processing the in-band noise characteristic discrimination waveform and the reference noise characteristic discrimination waveform according to a pattern recognition algorithm to identify the presence of noise patterns in the reference noise characteristic discrimination waveform that correlate with noise in the in-band noise characteristic discrimination waveform; detecting an imaging signal by the imaging transducer receiving circuit to obtain an imaging waveform, and detecting a reference noise detection waveform by the reference receiving circuit and filtering the imaging waveform to obtain an in-band imaging waveform present within the imaging band; segmenting the in-band imaging waveform and the reference noise detection waveform according to one or more time windows; processing the reference noise detection waveform according to the pattern recognition algorithm for at least one time window to detect the presence of the noise pattern; upon detecting the noise pattern, applying a noise suppression correction to the time window of the in-band imaging waveform specific to the noise pattern detected in the reference noise detection waveform; has.

[0026] In one aspect, a method for removing noise from a detected imaging signal when noise is present is provided, the method comprising: obtaining a noise characteristic discrimination waveform by detecting energy waves with an imaging transducer receive circuit when no imaging energy is being received, and filtering the noise characteristic discrimination waveform to generate an in-band noise characteristic discrimination waveform that resides within an imaging band and an out-of-band noise characteristic discrimination waveform that resides within a noise detection band that is at least partially outside of the imaging band; detecting an imaging signal by the imaging transducer receiving circuit to obtain an imaging waveform, and filtering the imaging waveform to generate an in-band imaging waveform present in the imaging band and an out-of-band noise detection imaging waveform present in the noise detection band; performing a cross-correlation between the out-of-band imaging waveform and the out-of-band noise characteristic discrimination waveform to determine a time delay corresponding to a maximum cross-correlation; applying the time delay and amplitude adjustment to the in-band noise characteristic discrimination waveform to obtain a modified in-band noise characteristic discrimination waveform, and subtracting the modified in-band noise characteristic discrimination waveform from the in-band captured waveform; has.

[0027] In one aspect, a method for removing noise from a detected imaging signal when noise is present is provided, the method comprising: When no imaging energy is being received, obtaining a noise characteristic discrimination waveform by detecting energy waves with an imaging transducer receiving circuit, and filtering the noise characteristic discrimination waveform to generate an in-band noise characteristic discrimination waveform that is within an imaging band; obtaining a reference noise characteristic identification waveform by detecting noise with a reference receive circuit configured to not transmit imaging energy while detecting noise received by the imaging transducer receive circuit; detecting an imaging signal by the imaging transducer receiving circuit to obtain an imaging waveform, detecting a reference noise detection waveform by the reference receiving circuit, and filtering the imaging waveform to obtain an in-band imaging waveform present within the imaging band; performing a cross-correlation between the reference noise detection waveform and the reference noise characteristic identification waveform to determine a time delay corresponding to a maximum cross-correlation; applying the time delay and amplitude adjustment to the in-band noise characteristic discrimination waveform to obtain a modified in-band noise characteristic discrimination waveform, and subtracting the modified in-band noise characteristic discrimination waveform from the in-band captured waveform; has.

[0028] In one aspect, a method for removing noise from a detected imaging signal when noise is present is provided, the method comprising: obtaining a plurality of imaging waveforms by detecting imaging signals by imaging transducer receive circuitry along a plurality of adjacent scan lines; For at least two adjacent scan lines, filtering the imaging waveform to generate an in-band imaging waveform that resides within an imaging band and an out-of-band noise detection imaging waveform that resides at least in part within a noise detection band that is outside the imaging band; segmenting the in-band imaging waveform and the out-of-band noise detection imaging waveform according to a series of time windows; For at least one window, processing the out-of-band noise detected imaging waveform to determine whether a corresponding windowed portion of the in-band imaging waveform should be noise corrected; applying a noise suppression correction to the in-band imaging waveform within the time window if the in-band imaging waveform within the time window is deemed suitable for noise correction, the noise suppression correction for each sample within the window being based on one or more statistical measurements of samples in the in-band imaging waveform of two or more adjacent windows, each adjacent window being in an adjacent scan line; generating an image based on noise-suppressed intra-band imaging waveforms corresponding to a plurality of scan lines; has.

[0029] In one aspect, a method for removing noise from a detected imaging signal when noise is present is provided, the method comprising: obtaining an imaging waveform by detecting energy waves with an imaging transducer receive circuitry and filtering the imaging waveform to generate an in-band imaging waveform that resides within an imaging band and an out-of-band noise detection imaging waveform that resides at least in part within a noise detection band that is outside the imaging band; obtaining a plurality of imaging waveforms by detecting imaging signals by the imaging transducer receive circuitry along a plurality of adjacent scan lines; processing one or more out-of-band noise detection imaging waveforms to determine a periodicity of a noise source; adjusting a scan rate to cause the noise to be time-synchronously abolished in adjacent scan lines and corresponding in-band imaging waveforms; For at least two adjacent scan lines, segmenting the in-band imaging waveform according to a series of time windows; For at least one window, applying a noise suppression correction to the in-band imaging waveform within the time window, the noise suppression correction for each sample within the window based on one or more statistical measurements of samples in the in-band imaging waveform of two or more adjacent windows, each adjacent window being in a corresponding adjacent scan line; generating an image based on noise-suppressed intra-band imaging waveforms corresponding to a plurality of scan lines; has.

[0030] In one aspect, a method for removing noise from a detected imaging signal when noise is present is provided, the method comprising: For at least two adjacent scan lines, detecting an imaging signal by an imaging transducer receiving circuit to obtain an imaging waveform, detecting a reference noise detection waveform by a reference receiving circuit, and filtering the imaging waveform to obtain an in-band imaging waveform present within an imaging band; segmenting the in-band imaging waveform and the reference noise detection waveform according to a series of time windows; For at least one window, processing the reference noise detection waveform to determine whether a corresponding windowed portion of the in-band imaging waveform should be noise corrected; applying noise suppression correction to the in-band imaging waveform within the time window if the in-band imaging waveform within the time window is deemed suitable for noise correction, the noise suppression correction for each sample within the window being based on one or more statistical measurements of samples of the in-band imaging waveform of two or more adjacent windows, each adjacent window being in an adjacent scan line; generating an image based on the noise-suppressed in-band imaging waveform corresponding to a respective one of a plurality of scan lines; has.

[0031] A further understanding of the functional and useful aspects of the present disclosure will be gained by reference to the following detailed description and drawings.

[0032] Embodiments will now be described, by way of example only, with reference to the drawings in which: [Brief description of the drawings]

[0033] [Figure 1A] 1 illustrates an example of an ultrasound imaging system configured for noise suppression. [Figure 1B] 1 shows an example of a conventional ultrasound receive signal chain that processes ultrasound signals before converting them into an ultrasound image. [Figure 1C] 1 illustrates an example of an ultrasound imaging system having an internal body imaging probe. [Figure 1D] 1 illustrates an example of an ultrasound imaging system having a reference transducer for detecting in-band noise. [Figure 1E] 1 illustrates an example ultrasound imaging system that includes a second imaging transducer having an imaging band outside the imaging band of the first imaging transducer, e.g., the second imaging transducer is part of a circuit suitable for detecting in-band noise affecting a signal received from the first imaging transducer. [Figure 1F] 1 illustrates an example ultrasound imaging system with a reference receive circuit for detecting in-band noise, the reference receive circuit extending to a location within the imaging probe. [Figure 1G] 1 illustrates an example ultrasound imaging system with a reference receiver circuit for detecting in-band noise, the reference receiver circuit being located at one or more locations external to the imaging probe of the system. [Figure 2A] This shows a system configuration that reduces the noise in the input waveform envelope by suppressing the estimated in-band noise. The in-band noise is estimated by performing envelope detection of the out-of-band waveform and performing delay, scale and shape adjustment before subtraction. [Figure 2B]An example system configuration is shown that reduces noise on an input waveform by suppressing estimated in-band noise. The in-band noise is estimated by frequency shifting the out-of-band waveform, filtering the frequency-shifted out-of-band noise, and applying delay, scale, and shape adjustments before subtraction. [Figure 2C] 1 illustrates an example system configuration for reducing noise, the system including a reference receiver circuit that detects some or all of the in-band noise detectable by the imaging transducer receive circuit and is at least partially isolated from the imaging signal detected by the imaging transducer receive circuit. The noise signal detected by the reference receiver circuit can be subtracted from the signal received by the imaging transducer receive circuit to reduce noise in the output signal. [Figure 3A] 1 shows an example of a system configuration for reducing noise by dynamic noise cancellation, in which the cancellation waveform of the dynamic noise cancellation is obtained from a reference receiving circuit. [Figure 3B] 1 shows an example of a system configuration for reducing noise on an input waveform by dynamic noise cancellation, where the cancellation waveform is obtained by envelope detection of the out-of-band waveform. [Figure 3C] 1 shows an example system configuration for reducing noise on an input waveform using dynamic noise cancellation, where the cancellation waveform is obtained by frequency shifting an out-of-band waveform and filtering the frequency-shifted out-of-band waveform. [Figure 4] An example system configuration is shown for reducing noise on an input waveform through dynamic noise cancellation, where the filtering is controlled based on feedback parameters obtained by a filter update algorithm that determines one or more parameters of the filter based on one or more characteristics of a noise detection waveform. [Figure 5A]An example system configuration is shown that reduces noise on an input waveform by applying one or more noise reduction algorithms that utilize noise parameters acquired in a first measurement stage, where no imaging signal is present, in a second measurement stage, where imaging signals are acquired. [Figure 5B] 1 illustrates a schematic of another example system for detecting in-band noise using a reference receive channel. [Figure 6A] 1 illustrates an example system configuration for reducing noise on an input waveform based on detecting noise in an out-of-band waveform, suppressing different time windows of an in-band waveform based on processing each window of the out-of-band waveform, and correcting the noise windows of the in-band waveform by subtracting a subtraction value that is dependent on the amount of power in the window of the out-of-band waveform. [Figure 6B] 1 shows an example scatter plot showing the signal power in a window of the in-band waveform and the signal power in each window of the out-of-band waveform at the noise characterization stage. [Figure 6C] 1 illustrates an example system configuration for reducing noise on an input waveform based on detecting noise in an out-of-band waveform, suppressing different time windows of an in-band waveform based on processing each window of the out-of-band waveform, and correcting the noise windows of the in-band waveform by subtracting a subtraction value that is dependent on the amount of power in the window of the out-of-band waveform. [Figure 6D] 1 illustrates an example system configuration for reducing noise on an input waveform based on detecting noise in an out-of-band waveform, suppressing different time windows of an in-band waveform based on processing each window of the out-of-band waveform, and correcting the noise windows of the in-band waveform by multiplying them by an attenuation factor that depends on the amount of power in the window of the out-of-band waveform. [Figure 6E] A chart is shown of a method in which different time windows of an in-band waveform are first identified as a main signal or noise, and then noise windows around the signal windows are identified and reclassified as possible errors, and signal windows around the noise windows are identified and reclassified as errors. [Figure 6F] A chart is shown of a method in which different time windows of an in-band waveform are first identified as a main signal or noise, and then noise windows around the signal windows are identified and reclassified as possible errors, and signal windows around the noise windows are identified and reclassified as errors. [Figure 6G] 1 illustrates an example system configuration for reducing noise on an input waveform based on detecting noise in a filtered reference waveform measured by a reference receive channel, suppressing different time windows of the in-band waveform based on processing each window of the reference waveform, and correcting noise windows in the in-band waveform by subtracting a subtraction value that is dependent on the amount of power in that window of the filtered reference waveform. [Figure 6H] 1 illustrates an example system configuration for reducing noise on an input waveform based on detecting noise in a filtered reference waveform measured by a reference receive channel, suppressing different time windows of the in-band waveform based on processing each window of the reference waveform, and correcting noise windows in the in-band waveform by subtracting a subtraction value that is dependent on the amount of power in that window of the filtered reference waveform. [Figure 6I] 1 illustrates an example system configuration for reducing noise on an input waveform based on detecting noise in a filtered reference waveform measured by a reference receive channel, correcting different time windows of an in-band imaging waveform based on processing each window of the filtered reference waveform, and correcting the noise windows of the in-band imaging waveform with an attenuation factor that is dependent on the amount of power in the window of the filtered reference waveform. [Figure 7A] 1 illustrates an example system configuration for reducing noise on an input waveform based on noise detected in one or more noise detection waveforms, at least one of which has an out-of-band signal of an imaging band, and wherein different time windows of the in-band waveform are denoised according to one or more patterns identified by processing the one or more noise detection waveforms. [Figure 7B]1 illustrates an example system configuration for reducing noise on an input waveform based on noise detected in one or more noise detection waveforms, at least one of which has an out-of-band signal of an imaging band, and wherein different time windows of the in-band waveform are denoised according to one or more patterns identified by processing the one or more noise detection waveforms. [Figure 7C] An example system configuration is shown for reducing noise on an input waveform based on noise detected in a reference waveform, where different time windows of an in-band imaging waveform are denoised according to one or more patterns identified by processing one or more reference waveforms. [Figure 7D] An example system configuration is shown for reducing noise on an input waveform based on noise detected in a reference waveform, where different time windows of an in-band imaging waveform are denoised according to one or more patterns identified by processing one or more reference waveforms. [Figure 8A] An example system configuration is shown for reducing noise on an input waveform based on noise detected in an out-of-band waveform. Different time windows of the in-band waveform are denoised according to the estimated in-band noise time-aligned before subtraction. [Figure 8B] An example system configuration is shown for reducing noise on an input waveform based on noise detected in an out-of-band waveform. Different time windows of the in-band waveform are denoised according to the estimated in-band noise time-aligned before subtraction. [Figure 8C] An example system configuration is shown for reducing noise on an input waveform based on noise detected in a filtered reference waveform. Different time windows of the in-band waveform are denoised according to the estimated in-band noise time-aligned before subtraction. [Figure 8D] An example system configuration is shown for reducing noise on an input waveform based on noise detected in a filtered reference waveform. Different time windows of the in-band waveform are denoised according to the estimated in-band noise time-aligned before subtraction. [Figure 8E]An example system configuration is shown that reduces noise on an input waveform based on noise detected in an out-of-band waveform, and employs measurements from adjacent or replicated scan lines when performing noise correction. [Figure 8F] An example system configuration is shown that reduces noise on an input waveform based on noise detected in a filtered reference waveform, and employs measurements from adjacent or replicated scan lines when performing noise correction. [Figure 9] 1 illustrates an example of a magnetic resonance imaging system configured for noise suppression. [Figure 10A] 10A and 10B show example images acquired using an intracardiac echo system, with FIG. 10A showing an image acquired without any noise sources. [Figure 10B] 10A and 10B show example images acquired using an intracardiac echo system, and FIG. 10B shows an image acquired in the presence of noise generated by an electroanatomical mapping system. [Figure 10C] 10A and 10B show example images acquired using an intracardiac echo system, and FIGURE 10C shows an image acquired in the presence of a noise source generated by an ablation generator. [Figure 11A] 11A and 11B show images acquired in the presence of noise from an electroanatomical mapping system: FIG. [Figure 11B] 11A shows an image acquired in the presence of noise from an electroanatomical mapping system, and FIG. 11B shows the denoised image. [Figure 11C] 11C shows an image acquired in the presence of noise from an electroanatomical mapping system after denoising by attenuation. The relaxation parameter in FIG. 11C is 0.5. [Figure 11D] 11D shows an image acquired in the presence of noise from an electroanatomical mapping system after denoising by attenuation. The relaxation parameter in FIG. 11D is 1. [Figure 11E] 11E shows an image acquired in the presence of noise from an electroanatomical mapping system after denoising by attenuation. The relaxation parameter in FIG. 11E is 1.5. [Figure 12A] The images shown are taken in the presence of noise from the ablation generator: Figure 12A shows an image without noise reduction; [Figure 12B] Figure 12A shows an image acquired in the presence of noise from the ablation generator, and Figure 12B shows the denoised image. [Figure 13A] Figure 13 shows images acquired in the presence of noise from the magnetic tracking system. Figure 13A shows the image without noise removal. [Figure 13B] Figure 13B shows an image acquired in the presence of noise from the magnetic tracking system. Figure 13B shows the image after noise has been removed. [Figure 14] 10 indicates the phrases used to refer to waveforms in the imaging band, in the noise detection band, and in the waveform from the reference receiver circuit. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0034] Various embodiments and aspects of the present disclosure are described in detail below. The following description and drawings are for the purpose of illustrating the present disclosure and should not be construed as limiting the present disclosure. Specific details are provided to allow the embodiments of the present disclosure to be understood. However, in some cases, known or existing details are not described in order to simplify the description of the embodiments of the present disclosure.

[0035] In this specification, the terms "comprises" and "comprising" are to be construed as inclusive and open ended, not exclusive. Specifically, in this specification and claims, the terms "comprises" and "comprising" and their derivatives mean that certain features, steps, or components are included. These terms are not to be construed as excluding other features, steps, or components.

[0036] As used herein, the word "exemplary" means "serving as an example, example, or illustrative example" and should not be construed as preferred or advantageous over other features in the specification.

[0037] In this specification, the terms "about" and "approximately" are meant to cover variations that exist within the upper and lower limits of a range of values, such as variations in properties, parameters, and sizes. Unless otherwise specified, the terms "about" and "approximately" mean plus or minus 25%.

[0038] Unless otherwise stated, specific ranges and groups should be understood as a shorthand way of referring to each member of that range or group individually, as well as any sub-ranges or subgroups contained therein, and the disclosure includes each and every combination of sub-ranges and subgroups unless otherwise stated.

[0039] As used herein, the term "on the order of" when used in conjunction with an amount or parameter means a range from about one-tenth to about ten times that amount or parameter.

[0040] Ultrasound imaging relies on receiving echoes from a medium and optionally transmitting narrow acoustic pulses into the medium in a particular direction. As used herein, the term "scanline" refers to a line that represents a spatial direction in which imaging energy is received in the medium. By receiving echoes from multiple scanlines in the medium, a 2D image is acquired. The inventors have conceived, developed, and tested various methods and systems that effectively reduce broadband noise from ultrasound acquisition and / or processing systems.

[0041] Referring to FIG. 1A, an example ultrasound imaging system is shown. One or more ultrasound transducers 10 are controlled to perform ultrasound imaging across multiple scan lines 12. The transducers 10 are interfaced with control and processing hardware 100, which optionally controls a transmitter 15 that generates and emits imaging energy by the transducers 10. The control and processing hardware 100 is configured to receive ultrasound energy detected by the transducers 10. The ultrasound energy is routed via a Tx / Rx (transmit-receive) switch 25 to one or more amplifiers 20.

[0042] The ultrasound transducer 10 may optionally be configured to image a region of space corresponding to multiple scan lines 12, for example by mechanical scanning of the transducer 10, or by electronic scanning using an imaging array, for example, but not limited to, a phased array, a ring array, a linear array, a matrix array, a curvilinear array, etc. In the latter case, a transmit beamformer 26 and a receive beamformer 27 may be used to generate multiple transmit signals and beamform multiple receive signals.

[0043] As used herein, the term "receive circuitry" generally refers to transmission lines (e.g., coaxial, PCB tracing, etc.), connectors, MUX / DEMUX, RX / TX switch 25, amplifiers 20, slip rings, transformers, and other components known in the art.

[0044] As used herein, the term "transducer receive circuitry" includes receive circuitry coupled to one or more ultrasonic transducer elements 10 that are configured, in use, to receive ultrasonic signals.

[0045] As used herein, the term "ultrasound receive signal chain" includes receive circuitry, but may also include other components, such as an analog-to-digital converter (ADC) and digital processing components and / or processing logic, including, but not limited to, a noise reduction processing module 150 before the signal is converted into an image (e.g., via scan conversion) for subsequent image processing.

[0046] As used herein, the term "channel" refers to a conductive electrical circuit, a wireless channel, an optical channel, or other signal path. For example, FIG. 1A shows imaging receive channels 13 that indicate paths across which detected imaging signals cross. A system may use one or more receive channels per transducer (e.g., in the case of an array transducer, where there is a channel for each electrostatic transducer element or group thereof). ASICs or other devices along the signal receive chain may also be used to multiplex signals along the channels from one or more electrostatic transducer elements.

[0047] In the "imaging mode," the system can be configured to control the transducer 10 to optionally transmit energy into the medium and detect imaging energy within an imaging frequency band (referred to as the "imaging band"). An imaging band can consist of a single continuous frequency band or two or more frequency intervals for detecting imaging energy. Imaging energy or noise within the imaging band is considered to be "in-band."

[0048] FIG. 14 shows the groupings and nomenclature used to describe various waveforms.

[0049] The system may also be configured to detect energy in one or more other frequency bands, at least one of which is at least partially beyond the imaging band, via one or more channels connected to the transducer. The one or more other frequency bands are referred to as "detection bands." A waveform that is at least partially beyond the imaging band is referred to as "out-of-band." A detection band may be within the imaging band. A waveform that is entirely within the imaging band and has frequency components within the entire imaging band or within a sub-band of the imaging band is referred to as "in-band." A noise detection band may be out-of-band or in-band. When using the transducer in an imaging mode (i.e., when the transducer detects imaging energy), at least one detection band may be selected such that the signal-to-noise ratio within the detection band is substantially less than the signal-to-noise ratio of the imaging band. For example, the detection band may be outside the full width at half maximum band of the imaging band, or outside any other band that corresponds to a threshold below the maximum intensity of the signal used.

[0050] As used herein, an "imaging waveform" refers to a waveform (analog or digitally sampled) obtained from an imaging transducer receive circuit when the imaging transducer is receiving or is assumed to be receiving imaging energy.

[0051] As used herein, the phrase "in-band imaging waveform" refers to an imaging waveform (analog or digitally sampled) that is within an imaging band. The in-band imaging waveform includes imaging energy and is expected to include unwanted noise energy. In embodiments of the present disclosure, the in-band imaging waveform is processed to remove the noise energy to generate a noise-free image.

[0052] As used herein, the phrase "detection band imaging waveform" refers to a waveform obtained from an imaging transducer receive circuitry and that resides within one or more detection bands. A detection band imaging waveform may be out-of-band or in-band. For example, an in-band noise detection imaging waveform may be used to verify that noise is present within an imaging band. More specifically, an "in-band noise detection imaging waveform" may be used to verify that a noise source that has a noise component outside the imaging band also has a noise component within the imaging band. A detection band imaging waveform that is at least partially outside the imaging band is referred to as an "out-of-band noise detection waveform."

[0053] See Figure 1A. When the transducer 10 is not transmitting energy to the medium and is not detecting imaging energy from the medium, the system can be configured in a "noise characteristic discrimination mode."

[0054] As used herein, the phrase "noise characteristic discrimination waveform" refers to a waveform acquired when the imaging transducer is not receiving imaging energy.

[0055] As used herein, the phrase "in-band noise characteristic discrimination waveform" refers to a waveform that resides within the imaging band and is acquired from an imaging transducer receive channel when the imaging transducer is not receiving imaging energy.

[0056] As used herein, the phrase "detection band noise characteristic discrimination waveform" refers to a waveform that is within the noise detection band and is acquired from an imaging transducer receive channel when the imaging transducer is not receiving imaging energy. A detection band noise characteristic discrimination waveform that is at least partially outside the imaging band is referred to as an "out-of-band noise characteristic discrimination waveform." A detection band noise characteristic discrimination waveform that is entirely within the imaging band is referred to as an "in-band noise characteristic discrimination waveform."

[0057] As used herein, the phrase "baseline noise characterization waveform" refers to a waveform acquired when the imaging transducer is not receiving imaging energy and the selected noise source is assumed to be OFF (i.e., not producing noise), such that the baseline characterization waveform provides a baseline for the selected noise source. A baseline noise characterization waveform within an imaging band is referred to as an "in-band baseline noise characterization waveform." A baseline noise characterization waveform within a noise detection band is referred to as a "detection band baseline noise characterization waveform."

[0058] See FIG. 1A. An optional reference receiver circuit 11 may be provided. The reference receiver circuit may include a receiver circuit configured to not receive reflected ultrasound signals during imaging and capable of detecting noise-like noise energy coupled to one or more of the transducer receiver circuits during imaging. The reference receiver circuit may use components of one or more of the transducer receiver circuits (e.g., the reference receiver circuit and the transducer receiver circuit may use different channels of an amplifier or ADC).

[0059] In one embodiment, the system can be configured to detect noise in the imaging band via one or more reference receive channels, which are optionally coupled to a reference ultrasound transducer (not shown) that is acoustically isolated or deactivated so as not to convert reflected ultrasound waves but to detect noise received by the imaging transducer receive circuitry. The one or more imaging transducers 10 and the one or more reference transducers need not point in a common spatial direction.

[0060] The signal received by one or more reference transducer receive circuits or reference electrical receive circuits (on a reference receive channel) is referred to as the reference waveform. The reference waveform is primarily noise and not imaging energy.

[0061] As used herein, the phrase "reference waveform" refers to a waveform obtained from one or more reference receive channels. The reference waveform may be filtered to be within and / or outside the imaging band.

[0062] As used herein, the phrase "reference noise detected waveform" refers to a reference waveform obtained from a reference receive channel when the imaging transducer is not receiving or is assumed to be receiving no imaging energy.

[0063] As used herein, the phrase "reference noise characteristic discrimination waveform" refers to a reference waveform obtained from a reference receive channel when the imaging transducer is not receiving imaging energy.

[0064] The system can optionally be configured to suppress noise using a combination of a detection band waveform and a reference waveform. As used herein, the phrase "noise detection waveform" refers to either the reference waveform or the detection band waveform. When the system is in an imaging mode and the imaging transducer receive circuitry is receiving or is assumed to be receiving imaging energy, the noise detection waveform is referred to as the "noise detection imaging waveform." When the system is in a noise characteristic discrimination mode and the imaging transducer receive circuitry is not receiving imaging energy, the noise detection waveform is referred to as the "noise detection characteristic discrimination waveform."

[0065] Although Figure 1A shows a single transducer element, it should be understood that the embodiment shown in Figure 1A is merely a non-limiting example configuration and transducers having multiple electrostatic elements can be used. For example, in one embodiment, multiple transducer elements may be controlled as a phased or linear or 2D array. Additionally, transducers may include, but are not limited to, transducers that transmit imaging energy to generate multi-dimensional 2D cross-sectional images or 3D images (including 4D imaging datasets with time-lapse 3D images), including transducers used for Doppler assessment of flow, transducers used as ultrasonic beacons (e.g., as described in U.S. Patent Publication No. 2016 / 0045184, entitled "Active localization and visualization of minimally invasive devices using ultrasound," which is incorporated herein by reference in its entirety), or ultrasonic transducers used to detect the position of moving elements (e.g., as described in U.S. Patent Publication No. 2012 / 0197113, entitled "Ultrasonic probe with ultrasonic transducers addressable on common electrical channel," which is incorporated herein by reference in its entirety).

[0066] The transducers are not limited to those that transmit and receive imaging energy as in FIG. 1A, but include transducers that receive ultrasound energy from a medium excited by other means, such as optical energy (photoacoustic imaging) or another ultrasound transducer. Additionally, while FIG. 1A shows an arrangement for imaging a spatial region corresponding to multiple scan lines in different directions, the scan lines may be unidirectional, such as in M-mode imaging or Doppler imaging devices, pulsed wave or continuous wave Doppler to assess flow.

[0067] In one embodiment, one transducer receive channel is configured to receive imaging energy in an imaging band while simultaneously receiving other energy in one or more noise detection bands, the other energy including at least out-of-band noise. In another embodiment, one or more imaging transducer receive channels can be used to receive imaging energy in the imaging band and one or more transducer receive channels can be used to receive other energy in one or more noise detection bands, at least some of which include out-of-band noise. In another embodiment, one or more reference receive channels can be used to receive noise energy (i.e., reference noise detection waveforms) while isolating it from the imaging energy in the imaging band. The reference receive channels can be filtered in a similar manner to the imaging transducer receive channels by using an imaging bandpass filter. In alternative embodiments, the reference receive channels may not be filtered at all or may use filters other than imaging bandpass filters to improve noise estimates in the imaging bands.

[0068] The control and processing hardware 100 includes, for example, one or more processors 110, memory 115, a system bus 105, one or more input / output devices 120, and a number of optional devices (eg, a communications interface 135, a data acquisition interface 140, a display 125, an external storage device 130).

[0069] It should be understood that the exemplary system depicted in FIG. 1A is a non-limiting example and is not intended to be limited to the components depicted. For example, the system may include one or more separate processors and memory devices. Additionally, one or more components of the control and processing hardware 100 may be provided as external components coupled to a processing device. For example, as depicted, the optional transmit beamformer 26 and the optional receive beamformer 27 may be components of the control and processing hardware 100 (shown in dashed lines) or may be provided as one or more external devices.

[0070] Some aspects of the present disclosure may be implemented, at least in part, by software. When executed by a computer system, the software configures the computer system to become a special purpose computer capable of implementing the signal processing and noise suppression method (or variations thereof) of the present disclosure. That is, the present technology may be implemented in a computer system or other data processing system as a response to a processor. The processor may be, for example, a microprocessor, a CPU, or a GPU that executes sequences of instructions in a memory. The memory may be, for example, a ROM, a volatile RAM, a non-volatile memory, a cache, a magnetic disk, an optical disk, a cloud processor, or other remote storage device. Furthermore, the instructions may be downloaded to the computer device over a data network, for example in the form of a compiled and linked version. Alternatively, the logic for implementing the process may be implemented in another computer and / or machine-readable medium, for example, in discrete hardware components such as a large scale integrated circuit (LSI), an application specific integrated circuit (ASIC), or firmware such as an electrically erasable programmable read only memory (EEPROM), a field programmable gate array (FPGA), or the like.

[0071] Computer readable media can be used to store software and data that, when executed by a data processing system, performs various methods. The executable software and data can be stored in a variety of places, such as ROM, volatile RAM, non-volatile memory, and / or cache. Portions of this software and data can be stored in any of these storage devices. In general, a machine readable medium has any mechanism for providing (i.e., storing and / or transmitting) information in a form that can be accessed by a machine (e.g., a computer, a network device, a personal digital assistant, a manufacturing tool, any device having one or more processor sets, etc.).

[0072] Examples of computer readable media include, but are not limited to, read only memory (ROM), random access memory (RAM), flash memory devices, floppy and other removable disks, magnetic disk storage media, optical storage media (e.g., compact discs (CDs), digital versatile disks (DVDs), etc.), network storage, cloud storage, and other recordable and non-recordable media. The instructions may be embodied in digital and analog communications links, such as electrical, optical, acoustic and other forms of propagated signals, carrier waves, infrared signals, digital signals, and the like. As used herein, the phrases "computer readable material" and "computer readable storage media" refer to all computer readable media except transitory propagating signals.

[0073] Many of the embodiments herein use adjustment of the denoising filter based on the noise measured in the environment. In some embodiments, one or more of the waveforms, data, filter parameters, and other relevant information described in the following examples can be sent to a network for remote evaluation to analyze and / or optimize the denoising filter. After optimization, the denoising filter algorithm and / or parameters can be sent to the system to improve the denoising effect.

[0074] 1A, the control and processing hardware 100 includes an image processing module 145 and a noise suppression module 150. The image processing module 145 may be configured or programmed to perform known image processing methods (e.g., scan conversion).

[0075] Although some of the embodiments are described in terms that provide real-time noise removal, it should be understood that noise removal may be a post-processing step, for example, data on the transducer receive channel or the reference receive channel may be digitized and stored before or after filtering, envelope detection, shifting, shape / phase or delay adjustment, signal characteristic identification, attenuation, subtraction, or other steps described by embodiments of the present invention.

[0076] FIG. 1B illustrates example steps by which the control and processing hardware 100 and receive channels process the in-band imaging waveforms detected from the imaging transducer receive channels before generating an image. The detected waveforms from the imaging transducer receive circuits are amplified (201) and filtered (202) before analog-to-digital conversion (203). After digitization, the detected waveforms are filtered using a bandpass filter 200 (which may use multiple passbands and stopbands) to retain signals within the imaging band. The envelope of the filtered waveform is generated by an envelope detector (210). The resulting envelope detected waveform is optionally decimated or expanded (220) and provided to an image processing module to generate an image (230).

[0077] With reference to FIG. 1A, the control and processing hardware 100 includes one or more noise suppression modules 150. The noise suppression module 150 has instructions for processing the detected data (e.g., raw RF data, envelope data, image data) to remove the effects of noise according to a noise suppression algorithm described below. As described below, the noise suppression algorithm described herein (shown as noise suppression module 150 in FIG. 1A) is used to remove or suppress noise in the process flow steps shown in FIG. 1B based on processing one or more noise detection waveforms. In the case of a system using an array transducer, noise suppression can be performed either before or after beamforming (or both). In the embodiments described below, noise suppression of image data (including but not limited to raw waveforms, sampled waveforms, envelope waveforms, Fourier transformed signals, and processed image data) is performed based on signal energy measurements (power, amplitude, intensity, and other measurements of signal strength) or waveform pattern measurements of noise detection waveforms (e.g., out-of-band noise detection imaging waveforms detected on a reference receive channel). The noise detection waveform, in many embodiments, is substantially free of imaging energy and includes noise that is correlated or coexists with the noise in the imaging band. One or more relationships between the imaging band noise and the noise detected by the noise detection waveform can be used to correct the imaging band signal to remove or reduce (e.g., suppress) the imaging band noise.

[0078] <Implementation of a reference receiver circuit for in-vivo imaging> Referring to Figure 1C, an example ultrasound imaging system is shown, comprising an internal body imaging probe 350 connected to control and processing hardware 100 via a subject interface module (PMI) 300. The internal body ultrasound imaging device can be configured to receive acoustic imaging energy from a one-dimensional, two-dimensional, or three-dimensional region, optionally by mechanical or electronic scanning.

[0079] The imaging probe 350 includes an imaging component 353 distal to the proximal end, an electrical and / or optical channel 354 passing through an optical conduit 354 along at least a portion of its length, and a connector 351 at the proximal end. For purposes of this disclosure, the imaging component 353 is generally understood to be a component or set of components of the imaging probe 350 for detecting imaging energy (e.g., acoustic or optical signals) to image an area adjacent to the imaging component. The imaging component optionally includes one or more emitters of imaging energy and includes at least one receiver of imaging energy. For example, the imaging component includes an ultrasound imaging transducer 10 that is both an emitter and receiver of acoustic energy. The ultrasound imaging transducer may be mounted on the imaging component. The imaging component may optionally be attached to or connected to a rotatable conduit or shaft (e.g., a torque cable) 352 housed within a hollow sheath of the intracorporeal ultrasound imaging probe to perform mechanical scanning.

[0080] The optional PIM 300 allows signal transmission via the PIM cable 320 to an appropriate imaging processing unit 100 in any wire or conduit, such as when the imaging probe 350 is not directly connected to the control and processing hardware 100. The PIM includes one or more amplifiers 20 to amplify signals from one or more transducer receive channels. The PIM optionally includes a motor drive unit 301 to provide rotational motion to the rotatable conduit 354. The motor drive unit 301 includes slip rings, rotary converters, and other components that connect the probe 350 signals to the control and processing hardware 100, allowing the imaging conduit to rotate and not the PIM cable 320 to rotate. The PIM 300 optionally includes a pullback mechanism 302 or a mutual push-pull mechanism that allows the imaging component 353 to move longitudinally. The longitudinal movement of the imaging component occurs with the longitudinal movement of an outer shaft surrounding the imaging conduit, or occurs within a relatively stable outer shaft.

[0081] Many electrical components in the imaging system can pick up unwanted energy from environmental noise sources. Examples of such components include the imaging components 353, the imaging conduit 352, the motor drive unit 301, and the PIM cable 320. One or more reference receive circuits that detect noise that is correlated with the noise detected by the imaging receive circuits are useful for suppressing in-band noise. Below is an example of a reference receive circuit for noise reduction of the imaging signal of an ultrasound imaging system.

[0082] FIG. 1D shows an embodiment in which the imaging transducer is replicated in the imaging probe 350 to one or more non-imaging reference transducers 361. The reference transducer has its own electrical channel 360 passing through the optical conduit 352 and connector 351. The reference transducer 361 can be coated with epoxy or other acoustically attenuating material 362 to acoustically isolate it from the received imaging energy. Alternatively, the piezo elements can be depoled to make them inactive or replaced with an acoustically responsive substrate. This implementation can be extended to an array transducer where there are multiple ultrasound transducer elements configured to receive acoustic imaging energy. One or more elements of the array can be made acoustically unresponsive, thereby not reflecting the acoustic imaging energy and functioning as a reference transducer receiving circuit.

[0083] 1E illustrates an embodiment in which an imaging probe includes two or more imaging transducers, each having a separate electrical channel 370 and 354. The two or more imaging transducers have sensitivity to receive acoustic imaging energy in substantially non-overlapping spectral bandwidths. For example, a first transducer is configured to receive acoustic energy around a frequency of 10 MHz, and a second transducer is configured to receive acoustic energy around a frequency of 40 MHz. The 40 MHz band of the 10 MHz transducer serves as a reference noise channel for the 40 MHz transducer, and similarly, the 10 MHz band of the 40 MHz transducer serves as a reference noise channel for the 10 MHz transducer.

[0084] FIG. 1F shows an embodiment in which the imaging transducer receive channels are replicated by a reference receive circuit in the imaging probe. The reference receive circuit may optionally comprise a combination of resistors, inductors, capacitors, and / or other components configured to match the electrical impedance of the reference electrical circuit 381 to that of the imaging transducer receive circuit or to make the noise sensitivity of the reference electrical circuit similar to that of the transducer receive channel. The advantage of this embodiment is that it does not require an actual ultrasound transducer, thereby reducing the cost, ease of manufacture, and ease of miniaturization of the components of the reference electrical circuit 381. Furthermore, parts of the reference receive circuit can serve other purposes, such as transmitting energy to drive actuators (e.g., but not limited to, magnetic actuators), carrying signals (including but not limited to, temperature, pressure, current, etc., resulting from magnetic fields for position detection), etc. Using the reference receive circuit for purposes other than collecting reference noise detection waveforms to reduce noise in the imaging signal in this way can facilitate miniaturization, lower cost, and / or improve functionality.

[0085] FIG. 1G shows an embodiment in which the reference receive circuit 391 is terminated within the PIM, which is subject to noise received by the motor drive unit and the PIM cable, as well as the imaging receive channel. Only a portion of the transducer receive circuit is replicated. Reference noise detection may optionally be used in conjunction with detection band (out-of-band or in-band) noise detection to further reduce noise in the in-band imaging waveform from the imaging transducer receive channel.

[0086] An embodiment in which a detection band waveform is used as the noise detection waveform is less expensive to manufacture than an embodiment in which a reference circuit or a reference transducer is used to generate the reference waveform, since the former does not require a physical implementation of a reference channel, such as when an imaging probe or part thereof is not repeated between subjects. A noise detection waveform from a reference channel may be more effective in imaging systems for noise rejection, since it can provide information about the noise within the imaging band, whereas an out-of-band noise detection waveform does not provide a direct estimate of the in-band noise, but instead relies on the noise to be able to estimate the in-band characteristics based at least in part on the out-of-band characteristics.

[0087] The noise estimate obtained using the detection band waveform or the reference noise detection waveform can be used to reduce the noise in one or more imaging channels. For example, in a phased array transducer with multiple electrostatic elements, a reference receive channel or an out-of-band noise detection waveform can be used to estimate the noise that can be collected by all or a subset of the electrostatic elements. The same noise estimation scheme can then be applied to signals collected from all or a subset of the electrostatic elements.

[0088] <Noise measurement> The following sections of this disclosure describe several embodiments that perform denoising of in-band image data based on out-of-band noise detection, reference channel noise detection, or both out-of-band noise detection (which may also be supplemented by in-band noise detection) and reference channel noise detection.

[0089] As described in conjunction with the examples below, noise measurements and / or noise characteristics can be determined from measurements of the noise detection waveform to increase the signal to noise ratio within the imaging band. Non-limiting examples of noise assumptions are as follows: Noise detection and in-band imaging waveform energy measurements (amplitude, root mean square amplitude, average power); Energy measurement in two or more noise detection bands; · the temporal, spectral, and / or time-frequency characteristics of the noise detection waveform; including spatial or space-time patterns in an image generated using the noise detection waveform, features or parameters that represent or characterize the patterns; including temporal, spectral, and / or time-frequency characteristics of the noise detection imaging waveform and concurrent patterns in the in-band imaging waveform, features representative of these patterns; including spatial or space-time patterns in an image generated using the noise detection waveform and concurrent spatial patterns in an image generated from the in-band imaging waveform from the imaging transducer channel, features or parameters representative of or characterizing the patterns; · Filter parameters are determined or controlled using parameters obtained based on waveform characteristics (e.g., energy) and spectral spacing of detected harmonic peaks in the noise detection waveform.

[0090] In some embodiments, the noise characteristics can be estimated when the imaging transducer is not receiving imaging energy (e.g., after ultrasound energy from the most recent emission of an ultrasound pulse is assumed to have dissipated from the foliage). Alternatively, the noise characteristics can be estimated during imaging when imaging energy is assumed to be detected (e.g., when the transducer is in image acquisition mode). In embodiments where the noise characteristics are measured in the absence of imaging energy, the noise characteristics can be updated intermittently to accommodate and compensate for time-dependent changes in the noise characteristics.

[0091] <Noise removal using measurements where no imaging energy exists> Noise reduction embodiments involve measuring and using energy from a noise detection waveform when the imaging transducer is receiving imaging energy, while alternative embodiments use measurements from a noise detection waveform taken during periods when no imaging energy is present, or a combination of both.

[0092] 5A illustrates an embodiment in which noise characterization is performed while the transducer receive channels are not receiving imaging energy, and the resulting noise characterization is used to denoise in-band imaging waveforms acquired while the transducer receive channels are not receiving imaging energy. The noise characterization step is typically performed prior to acquisition and processing of denoised image data, but by appropriately recording the image data, the information gathered in the noise characterization can be used a posteriori on the recorded image data.

[0093] In accordance with the method of this embodiment, during a first time window during which it is assumed or known that at least one transducer receive circuit is not receiving imaging energy, energy is detected at the transducer receive circuit such that the waveform detected by the imaging transducer receive channel 13 is considered noise 405. The detected waveform is filtered at 200 and 410 to generate an in-band noise characteristic discrimination waveform 407 and a detected band noise characteristic discrimination waveform 408.

[0094] The in-band noise characterization waveform and the detected band noise characterization waveform are processed to characterize the noise, as shown at 420. For example, the noise characterization 420 can be used to generate characterization parameters 430 that characterize the noise. Examples of suitable noise characterization parameters are described in the following embodiments.

[0095] Optionally, energy can be detected at the transducer receive circuits in a separate baseline noise characterization stage when it is expected or known that at least one transducer receive circuit is not receiving imaging energy and a particular noise source is known to be selectively OFF and the waveform detected by the imaging transducer receive channel 13 is considered a baseline for the selected noise source. The detected baseline noise characterization waveform is filtered (200 and 410) to generate an in-band baseline noise characterization waveform 407 and a detected band baseline noise characterization waveform. It should be appreciated that the noise parameters 430 may include parameters obtained in the baseline noise characterization stage.

[0096] The characteristic discrimination noise parameters can be calculated before or during the imaging session, or can be retrieved from a pre-stored database located on a local or remote storage device (network drive, cloud, etc.).

[0097] After characterizing the noise in the absence of imaging energy, the characterization parameters 430 can be used to perform noise removal on the in-band imaging waveform 437 detected while the transducer is receiving imaging energy. The waveform detected from the imaging transducer receive channel during imaging (435) includes imaging energy and noise, which is filtered (200 and 410) to generate the in-band imaging waveform 437 and the detected band imaging waveform 438. Thus, the in-band imaging waveform 437 includes the detected imaging energy and noise, and the detected band imaging waveform 438 includes information about noise that may be present in the in-band imaging waveform 437. The characterization parameters 430 obtained in the noise characterization stage are used to detect and / or estimate (440) noise in the in-band imaging waveform and perform noise suppression (500) of the in-band imaging waveform. Suitable examples of noise suppression using the noise characterization parameters are described in the following embodiments.

[0098] 5B illustrates another embodiment in which a reference receive channel (e.g., the reference receive channel depicted in FIG. 1A) is used to detect noise to generate a reference noise characteristic identification waveform 406. The reference noise characteristic identification waveform is filtered (202) to generate a filtered reference noise characteristic identification waveform 409. In one embodiment, the reference channel filter is an imaging bandpass filter. Alternatively, the reference channel filter may be other than an imaging bandpass filter if the noise estimate is due to input of out-of-band noise. The reference noise characteristic identification waveform 409, and optionally the in-band noise characteristic identification waveform 407, are processed in step 420 to provide noise characteristic identification parameters 430.

[0099] During imaging, the reference waveform 436 is detected and optionally filtered to generate a filtered reference noise detected waveform 439. The characteristic parameters 430 obtained in the noise characteristic identification stage are used to detect and / or estimate (440) noise in the in-band imaging waveform 437 and perform noise suppression (500) of the in-band imaging waveform 437. In another embodiment, both the detected band imaging waveform 438 (shown in FIG. 5A ) and the filtered reference noise detected waveform 439 are processed to provide information about noise that may be present in the in-band imaging waveform 437.

[0100] In embodiments, noise suppression can be achieved by processing the in-band imaging waveform using one or more of the following methods: subtracting the estimated noise from the in-band signal; multiplying the in-band signal by an attenuation factor to attenuate the estimated noise energy; or filtering the in-band signal. For example, the amount of subtraction is proportional to the amount of power detected in the out-of-band noise detection imaging waveform. Alternatively, the attenuation factor is inversely proportional to a measurement corresponding to the amount of noise in the in-band imaging waveform, thereby attenuating the noise-related portions of the in-band imaging waveform.

[0101] The noise characteristic identification (e.g., as shown at 420 in FIGS. 5A-5B) may be performed once or, alternatively, may be performed continuously multiple times. For example, the noise characteristic identification may be performed intermittently (e.g., periodically or irregularly) to accommodate and compensate for time-dependent changes in the noise characteristics.

[0102] When noise characterization is being performed, the noise characterization waveforms collected for noise characterization may be digitized and collected in multiple individual arrays (e.g., arrays each long enough to store image data along one scan line of ultrasound imaging) or may be collected in a more continuous manner as one or more data streams stored in a larger array, circular buffer, or other data structure.

[0103] In an embodiment, noise characterization may be user initiated (eg, by pressing a button), for example, at the start of an imaging session or when the user notices or suspects degradation of image quality.

[0104] In another embodiment, the noise characteristic identification may be automatically triggered or user prompted upon detection of the absence of imaging energy. For example, a period of absence of imaging energy may be detected when the relative energy between the in-band imaging waveform and the noise detection imaging waveform remains unchanged and within a specified range over a period of time. A period of absence of imaging energy may be detected, for example, when the energy of the in-band imaging waveform after noise correction (501) is below a specified threshold, indicating the absence of imaging energy.

[0105] The noise characteristic identification step 420 may also be used to alert a user or system that the noise profile has changed in a manner that may adversely affect the noise reduction algorithm (e.g., if a new noise source is detected, the noise suppression module 500 may erroneously suppress in-band signals or reduce the effectiveness of in-band noise suppression). For example, a noise detection waveform (e.g., the out-of-band noise detection waveform 438 of FIG. 5A or the reference noise detection waveform 439 of FIG. 5B) may be processed to determine that the noise characteristics have changed. For example, when noise suppression is being performed, a noise monitoring module may optionally be used to monitor the characteristics of the noise detection waveform (e.g., peak energy, power, frequency content loop, skew, kurtosis, histogram, and other metrics). If the characteristics of the noise detection waveform change (e.g., peak energy exceeds a threshold), the noise monitoring module may communicate (e.g., via a message, interrupt, alarm, and other) with other parts of the system to alert them that the noise content has changed.

[0106] In another embodiment, an error value in the noise characterization stage can be evaluated, where noise suppression is performed on the in-band imaging waveform when no imaging energy is present, and the error value is the energy of the in-band imaging waveform after noise correction. If the error value exceeds a specified threshold, an alert is generated. The alert may prompt a recharacterization of the noise, or the system may ignore one or more out-of-band noise detection bands or one or more reference receive channels in the noise removal algorithm.

[0107] The noise characteristic identification 420 can also be used to determine noise sources in the environment. The noise sources can be determined, for example, by a pattern recognizer as shown in step 570 of FIG. 7A (discussed in more detail in embodiment 4). Information about the noise sources can be used, for example, to access a database (local or over a network) to select parameters for noise suppression or to determine the sequence of noise suppression methods to use. For example, periodic noise can be removed first (discussed in embodiment 6, in more detail in FIGS. 8A-8D). It may be desirable to remove noise with low periodicity next.

[0108] As another example, noise characteristic identification can be used to detect the type of electroanatomical mapping system being used in the ablation process, or to detect the activation and deactivation of an ablation catheter that uses radio frequency energy to perform ablation to treat arrhythmias. This can be accomplished, for example, by a pattern recognizer as shown in FIG. 7A, step 570. The one or more noise detection waveforms can further detect the duration, relative intensity, and frequency of the ablation energy being used. This information is useful for an internal imaging system, as it allows imaging data sets to be tagged with information about when a noise source (e.g., an ablation catheter) was activated.

[0109] Noise suppression (e.g., 500 in FIGS. 5A-5B) may be performed once, multiple times, intermittently, or continuously. For example, noise suppression may be user-initiated. Alternatively, noise suppression may be performed intermittently (e.g., periodic or non-periodic) to compensate for time-dependent noise sources. Alternatively, noise suppression may be controlled by an external device that outputs noise. For example, noise suppression may be activated or deactivated under control of the RF ablation generator, thereby providing noise suppression as RF energy propagates.

[0110] The following embodiments describe one imaging waveform and one noise detection waveform, it is understood that these embodiments can be extended to multiple imaging waveforms and / or multiple noise detection waveforms.

[0111] <Embodiment 1: Noise suppression based on envelope detection out-of-band noise suppression (options: amplitude, shape, and delay correction)> Referring to FIG. 2A, an example method is shown for using the out-of-band noise detection waveform 438 to perform noise suppression of the in-band imaging waveform 437 via a suppression operator 525. In one embodiment, the suppression operator is a subtractor that subtracts the estimated noise from the in-band imaging waveform 437. In another embodiment, the suppression operator is an attenuator that attenuates the in-band imaging waveform by an attenuation factor derived from the estimated noise. One or more transducer receive channels are used to detect imaging energy, the detected energy including both the imaging band and the noise detection band. The waveforms can be digitally sampled, separated (or copied), and filtered to obtain the sampled in-band imaging waveform 437 and the sampled out-of-band noise detection imaging waveform 438. The sampled waveforms can be detected as a set of samples received within a time window. For example, in ultrasound imaging, the listening window is immediately or slightly after the ultrasound transducer pulses, allowing energy exposure to the adjacent environment. The pulsed output corresponds to one or more pulse transmissions.

[0112] In the embodiment shown in Figure 2A, the imaging waveform 435 is filtered (digitally or analog) using an imaging bandpass filter 200 and a noise detection bandpass filter 410 having frequencies that extend outside the imaging band. Envelope detection is then performed on the in-band imaging waveform 437 and the out-of-band noise-detected imaging waveform 438, as shown at 210 and 411.

[0113] In the embodiment shown in FIG. 2A, the out-of-band noise detection imaging waveform 438 is used to suppress noise in the in-band imaging waveform 437. Prior to suppression, the amplitude of the envelope detected out-of-band noise detection imaging waveform 438 is optionally scaled (as shown at 510) to compensate for differences in noise power within the imaging band and the noise detection band. In one embodiment, the amplitude adjustment factor is determined based on the power spectrum of the noise determined in the absence of imaging energy (i.e., the noise characterization stage). In another embodiment, the amplitude adjustment factor may be selected or modified by an operator to achieve a desired level of noise suppression, or may be determined after cross-correlation 580 (described below).

[0114] Prior to suppression processing, the envelope detected out-of-band imaging waveform may be expanded, compressed, or shaped in time using other linear or non-linear time scaling functions (as shown in 510) to compensate for differences in noise waveform shape between the imaging band and the noise detection band.

[0115] It is also useful to apply delay correction to the envelope-detected out-of-band noise-detected imaging waveform before noise suppression, as shown at 510. For example, the two bandpass filters 200 and 410 do not transform the input waveform in the same way. The characteristics of the bandpass filter or the noise itself may introduce offsets to the noise as they propagate through the bandpass filter. Without delay correction, the noise would shift before suppression introducing errors, which would have a detrimental effect on the noise-suppressed signal.

[0116] In one embodiment, the delay adjustment can be performed by calculating the cross-correlation between the in-band and out-of-band imaging waveforms and aligning the waveforms at the point where the cross-correlation is maximum. In other words, the cross-correlation can be used to determine a time delay correction value to correct for the relative time shift of the envelopes of the in-band and out-of-band imaging waveforms.

[0117] The time delay correction values ​​and amplitude correction values ​​can be calculated using multiple sampled in-band imaging waveforms and (simultaneous) sampled out-of-band noise detection imaging waveforms (referred to as an "array"), or in one or more time windows of the sampled in-band imaging waveforms and (simultaneous) sampled out-of-band noise detection imaging waveforms.

[0118] In one embodiment, the out-of-band noise can be queried as multiple noise detection bands, and the dependence of power on frequency among the multiple noise detection bands can be used to select an appropriate amplitude adjustment to estimate the noise power present in the imaging band for noise suppression in step 525. For example, the average noise power in the multiple noise detection bands can be fitted (e.g., linearly fitted) to a functional dependence on frequency to estimate the noise power in the imaging band. This functional dependence on frequency can be determined when no imaging energy is present (i.e., noise characteristic identification stage 420).

[0119] Alternative: Noise suppression based on using frequency-shifted out-of-band noise detection waveforms (options: amplitude, shape, and delay correction) 2B illustrates another embodiment of a subtractive or attenuating noise correction method in which the out-of-band noise detection imaging waveform is frequency shifted prior to delay and amplitude adjustment. As shown at 530, a frequency shift operation (e.g., by multiplying by a complex exponential) is performed on the out-of-band noise detection imaging waveform to shift the spectrum of the out-of-band noise detection imaging waveform so that it is within or overlaps the imaging band.

[0120] The frequency shift calculation 530 can be performed so that the center frequency of the frequency-shifted noise detection waveform matches or is approximately equal to the center frequency of the imaging band. For example, if the imaging band is 7-13 MHz, the center frequency fc1 is 10 MHz. If the noise detection band is 15-25 MHz, the center frequency fc2 of the noise detection band is 20 MHz. Thus, the frequency shift calculation can be performed so that the out-of-band noise detection imaging waveform is shifted by fc2-fc1=10 MHz. Alternatively, the frequency shift can be performed so that the center frequency of the frequency-shifted out-of-band noise detection imaging waveform matches or is approximately equal to the frequency within the imaging band (where some of the in-band noise is expected or known to exist).

[0121] After frequency shifting, another stage of bandpass filter 203 is implemented to filter out the effect of sum frequency (in the above example, sum frequency is fc2+10MHz=30MHz). Frequency shifting is useful in the envelope detection embodiment shown in FIG. 2A because it provides better noise correlation between imaging band and noise detection band, improving noise suppression. In the embodiment shown in FIG. 2B, after noise suppression, envelope detection 210 can be applied to the output signal 501 to obtain a noise-suppressed signal envelope 520.

[0122] <Another embodiment based on using a reference noise signal> Referring to FIG. 2C, an example method is shown in which in-band noise detected via a reference receive channel (using a reference receive circuit) is used to suppress noise in an in-band imaging waveform via a suppression operator (i.e., subtractor or attenuator). One or more imaging transducer receive channels are used to receive imaging energy, and one or more reference receive channels are used to receive noise energy (i.e., a reference waveform). This noise energy is assumed to be correlated with the noise energy received by the imaging transducer receive channels.

[0123] The waveforms are digitally sampled to obtain a sampled in-band imaging waveform and a sampled filtered reference noise detection waveform. Alternatively, noise suppression may be performed with analog electronics. For example, the input from the reference receiver circuit to the analog signal summer may be inverted in the delay, scale and shape adjuster block 510, thereby subtracting the estimated noise. As an alternative embodiment of analog signal suppression, the suppression process may be implemented as an amplifier with a time-varying gain, which is modulated by the noise received by the reference receive channel.

[0124] In the embodiment shown in FIG. 2C, the input waveforms from the imaging transducer receive channel and the reference receive channel are filtered (digitally or analog) using an imaging bandpass filter 200 and an optional reference channel filter 202 to provide a filtered reference noise detection waveform along with the in-band imaging waveform, respectively. As described above, the reference channel filter 202 is similar to the imaging bandpass filter 200. An optional envelope detection is performed on the filtered signal, as shown at 210 and 411. In the embodiment of FIG. 2C, the reference noise detection waveform (as measured by the reference receive channel) is used to suppress noise in the in-band imaging waveform. Before being subtracted from or attenuating the in-band imaging waveform (or its envelope), the amplitude of the filtered reference noise detection waveform (or its envelope) is optionally scaled (as shown at 510) by an amplitude adjustment factor to compensate for differences in noise power in the filtered reference noise detection waveform and the in-band imaging waveform. In one embodiment, the amplitude adjustment factor can be determined based on the power spectrum of the noise determined when no imaging energy is present (i.e., the noise characterization stage). In another embodiment, the amplitude adjustment factor can be operator selected to provide a desired level of noise suppression, or can be determined after cross-correlation 580.

[0125] It is useful to apply a delay correction to the envelope-detected and filtered reference noise detection waveform before subtraction, as shown at 510. In one embodiment, the delay adjustment is accomplished by calculating the cross-correlation between the in-band imaging waveform and the filtered reference noise detection waveform, and aligning the waveforms at the point where the cross-correlation is maximum. As with the previous embodiment, the time delay and amplitude corrections can be calculated using multiple in-band imaging waveforms and the (concurrent) reference noise detection waveforms, or for one or more windows of the in-band imaging waveform and the (concurrent) reference noise detection waveform.

[0126] <Embodiment 2: Noise suppression using a reference noise detection waveform from a reference receiving channel as an input to an adaptive filter> FIG. 3A illustrates an example of a noise correction method using an adaptive filter in an dynamic noise control (ANC) scheme by applying noise suppression correction to an in-band imaging waveform based on a reference noise detection waveform. The reference noise detection waveform is correlated to noise detected by the imaging transducer receive circuitry. The reference noise detection waveform is filtered (202) and the waveform from the imaging transducer receive channel is filtered (200). In a preferred embodiment, the in-band imaging waveform and the reference noise detection waveform are filtered in the same band (e.g., 7-13 MHz in an example in vivo echocardiographic imaging system).

[0127] An adaptive filter is a linear filter with a transfer function controlled by variable parameters, by means of which its parameters are adjusted according to an optimization algorithm. Adaptive filters are usually digital Finite Impulse Response (FIR) or Intermittent Impulse Response (IIR) filters. A dynamic noise control (ANC) method is applied to the primary input. The primary input receives a signal (S) from a signal source that is corrupted by the presence of noise (N) that is uncorrelated with the signal. The reference input receives a noise (Nr) that is uncorrelated with the signal but somehow correlated with the primary input noise (N). The reference noise is passed through the adaptive filter to produce an output noise (N) that is an estimate of the primary input noise (N). estimate) to generate the noise-suppressed signal (S estimate ) estimate. The adaptive filter dynamically adjusts its coefficients to estimate the output power E[S estimate 2 ]. Because the signal S is uncorrelated with N and Nr, and the noise N is correlated with the noise Nr, the signal-to-noise ratio can be maximized by minimizing the total output power. A minimization algorithm (e.g., the stochastic least squares (LMS) or deterministic iterative least squares (RLS) algorithm) can be used to find the filter coefficients that minimize the output noise power.

[0128] In the example ANC scheme shown in Figure 3A, a reference noise is measured via a reference receive channel using a reference receive circuit that is isolated from the imaging energy. The main input is obtained by applying an imaging bandpass filter 200 to the input waveform from the imaging transducer receive channel 13. The reference input Nr is obtained by applying a reference channel filter 202 to the reference noise detection waveform obtained via the reference receive channel.

[0129] <Another embodiment: Dynamic noise cancellation using out-of-band noise> In contrast to the form of dynamic noise control described above, which detects and processes the reference noise detection waveform and the in-band imaging waveform within overlapping frequency bands (and common frequency bands), Figures 3B and 3C show an example of a noise correction method in which an adaptive filter 540 applies noise suppression correction to the in-band imaging waveform using an out-of-band noise detection imaging waveform.

[0130] The primary input is obtained by applying an imaging bandpass filter 200 to the input waveform from the imaging transducer receive channel to obtain an envelope 210. In the embodiment shown in Figures 3B and 3C, an out-of-band noise detected imaging waveform is obtained by applying a detection band filter 410 to the input waveform from the imaging transducer receive channel.

[0131] In FIG. 3B, the out-of-band noise detected imaging waveform is demodulated via envelope detection 411 (in a similar manner to the embodiment of FIG. 2A) to obtain the reference input for the ANC.

[0132] In FIG. 3C, the out-of-band imaging waveform is frequency shifted (530) into the imaging band (e.g., 7-13 MHz) in a manner similar to the embodiment of FIG. 2B and filtered using an imaging bandpass filter 203 to obtain the reference input for the ANC.

[0133] <Embodiment 3: Frequency-shift based noise suppression using detection band waveform as input to in-band waveform tunable filter> FIG. 4 illustrates an example of a noise correction method for filtering an in-band imaging waveform using a dynamic filter 550. The dynamic filter is controlled by a filter update algorithm 560 that updates the filter coefficients after processing an out-of-band noise detected imaging waveform that includes out-of-band noise (optionally an in-band noise detected imaging waveform that includes noise within all or part of the imaging band). As in the above embodiment (FIGS. 2B and 3B), the input waveform is filtered by the imaging bandpass filter 200 and the noise detection bandpass filter 410 separately, as shown in FIG. 4, to generate an in-band imaging waveform 437 and at least one out-of-band noise detected imaging waveform (including out-of-band noise 438). The one or more out-of-band noise detected imaging waveforms are processed by the filter update algorithm 560.

[0134] The filter update algorithm analyzes the out-of-band noise and evaluates the signal characteristics, for example by performing a Fourier transform on the waveform array and identifying the spectral maximum and the frequency at which it occurs. The filter update algorithm can use this information to control the coefficients of a dynamic digital filter (550) that filters the imaging waveform.

[0135] In one embodiment, the method can be used to noise suppress a signal that contains high frequency noise. For example, the high frequency noise can come from a switching rectangular pulse source where the spacing of the spectral lines in the frequency domain depends on the pulse repetition frequency. The pulse width modulated source generates noise in the 3-40 MHz band, the imaging band is 7-13 MHz, and the dynamic filter 550 is a comb filter or a multi-notch filter. The filter update algorithm 560 processes the signal from the noise detection band (e.g., evaluates the spectral line spacing and position found in the 15-25 MHz band) and can use this information to control the stop band of the dynamic in-band filter 550 to remove or suppress harmonic noise.

[0136] In addition to updating the dynamic filter with out-of-band noise, the filter update algorithm 560 and / or the dynamic filter 550 can optionally probe the in-band noise detection imaging waveform to determine the presence of in-band noise in one or more selected sub-bands within the imaging band prior to noise removal or suppression. For example, if the imaging band is 7-13 MHz and the filter update algorithm determines that harmonic noise is present at 15 MHz, 18 MHz, 21 MHz, and 24 MHz (integer multiples of 15-25 MHz), the filter update algorithm sets the dynamic filter to filter signals at 9 MHz and 12 MHz (integer multiples of 3 MHz within the imaging band). In one embodiment, such a filter may be optionally applied only if the signals at 9 MHz and 12 MHz are stronger than expected relative to signals within the imaging band. In another embodiment, a noise characterization step may be performed when no imaging energy is present to determine whether harmonic noise is present within the imaging band.

[0137] <Embodiment 4: Noise suppression based on pattern recognition> 7A and 7B illustrate an embodiment of a noise correction method that uses pattern recognition to detect noise and perform noise suppression on in-band imaging waveforms. In this embodiment, as shown in FIG. 7A, a matched pattern set of detection band noise characterization waveforms (at least one of which is at least partially outside the imaging band) and associated in-band noise characterization waveforms are first identified during a noise characterization period in which no received imaging energy is present. Once the noise detection patterns are correlated with the in-band noise patterns via the noise characterization stage, these correlations are used during imaging to perform noise suppression of the in-band imaging waveforms based on pattern recognition of one or more detection band imaging waveforms.

[0138] According to a first stage of the method, in a noise characterization stage, energy within both the imaging band (in-band) and the noise detection band is detected when no imaging energy is present (e.g., in a non-imaging noise characterization stage), thereby obtaining correlated measurements of in-band and detection band noise characterization waveforms. Samples from the in-band noise characterization waveform and the detection band noise characterization waveform are recorded as an array pair. An array pair is a simultaneously recorded sampled in-band waveform and a corresponding sampled detection band noise detection waveform.

[0139] The detection band noise characteristic identification array and the in-band noise characteristic identification array may be windowed (566 and 565). The windows may be, for example, sliding windows, and may optionally overlap. Optionally, the windows may be centered around a peak noise amplitude or may be time-fixed relative to a noise amplitude threshold. The windows may be conditioned to reduce adverse effects caused by the windowing, for example by applying a Hamming window, a Blackman window, or other window functions well known in the art of signal processing. The array data (or a window thereof) may be processed (570) to identify the presence of one or more noise patterns.

[0140] See FIG. 7A. One or more detection band noise characteristic identification arrays are processed to identify waveform patterns corresponding to the patterns of in-band noise characteristic identification waveforms. A pattern recognizer 570 extracts features from the detection band characteristic identification arrays and classifies the features into noise "classes" using a predictive model. The extracted features can be, for example, statistical features (including but not limited to variance, standard deviation, power, skewness, kurtosis, etc.) in the time domain, frequency domain (e.g., peak frequency), and time-frequency domain (e.g., wavelet coefficients). The features to be extracted can be preselected using feature selection algorithms such as forward selection and backward selection.

[0141] The extracted features are fed to a predictive model in step 570. The predictive model can be trained to identify patterns in the detection band waveform and assign the patterns to noise classifications. For example, machine learning methods can be used to train the predictive model to recognize patterns in the detection band noise characteristic identification array using the extracted features. The predictive model can include unsupervised learning models (e.g., k-means clustering) or supervised learning models (e.g., linear classifiers, artificial neural networks, nearest neighbor classifiers). A learning model can be used when prior information about noise sources is known. For example, the sequence of noise sources and noise patterns may be known in advance, and noise classification labels can be assigned to the waveform patterns in the detection band noise characteristic identification waveform. The predictive model in step 570 can accept input classification weights or prior probabilities. The higher the prior probability or classification weight, the higher the likelihood of recognition.

[0142] The database of step 575 may store in-band noise characteristic discrimination waveform patterns that are known to co-exist with the detected band noise characteristic discrimination waveform patterns. For example, the database may store samples or averages of in-band noise characteristic discrimination waveform patterns paired with the co-existing detected band noise characteristic discrimination patterns and noise classification labels. The detected band noise characteristic discrimination waveform patterns and their co-existing in-band noise characteristic discrimination waveform patterns may be determined, for example, on a window-by-window basis. Further noise characteristic discrimination may be performed in the time domain or may be performed in the spatial domain after image generation in steps 230 and 231. In this case, spatial features are also extracted in step 570. The database may be in any suitable format for use by the computer, for example, a look-up table.

[0143] Once the detection band noise patterns have been correlated with the in-band waveform noise patterns via the noise characteristic identification stage, these correlations can be used during imaging to perform noise suppression of the in-band imaging waveforms based on pattern identification of one or more detection band imaging waveforms.

[0144] See FIG. 7B. An in-band imaging waveform is obtained by applying an imaging bandpass filter 200 to the waveform detected by the imaging transducer receive channels while the transducer is receiving imaging energy. Optional envelope detection 210 is performed. A detection band imaging waveform is obtained by applying a noise detection bandpass filter 410 to the waveform detected by the imaging transducer receive channels. Optional envelope detection 411 is performed. The detection band imaging waveform and the in-band imaging waveform may be windowed (566 and 565), similar to the windowing step of the noise characteristic identification stage. The waveforms may be sampled and presented as an array.

[0145] Similar to the feature extraction step of the noise characteristic identification stage, features can be extracted from the detection band imaging array (or a window thereof). A pattern recognizer 570 trained in the noise characteristic identification stage (described above) uses the extracted features and optionally classification weights to identify the presence of one or more patterns in the detection band imaging waveform.

[0146] In conjunction with applying the pattern classification algorithm 570, the period (i.e., repetition frequency) of the noise pattern can be used to adjust the prior probability of the pattern classification (e.g., a Bayesian classifier) ​​or adjust the weight of the classification (e.g., that of a support vector machine). The higher the prior probability or weight of the classification, the more likely it is to be recognized. If the repetition interval of a pattern is known, the pattern is expected to be present at a given time with high probability. The prior probability or weight of the classification may be adjusted higher at that time to increase the likelihood that the pattern classifier will recognize the noise pattern. This repetition interval may be determined in the noise characteristics identification stage and stored in the database 578 or retrieved to a pre-stored database (local, network, cloud storage).

[0147] The patterns of the detected band imaging waveforms identified in step 570 as corresponding to one or more noise classifications are used to generate noise corrections for the in-band imaging waveforms (e.g., in-band imaging arrays). These corrections can be generated based on finding the correlating in-band patterns in step 575, and the detected band noise characteristic discrimination waveform patterns, matched sets of features of the in-band noise characteristic discrimination waveform patterns and noise classification labels, and the noise classification labels are stored in a searchable database or other classification mechanism.

[0148] In the example method of Figure 7B, an in-band noise correction is generated for each window and subtracted for each window from the in-band imaging waveform 525, optionally after a delay and / or amplitude and / or shape adjustment 510 to time-align the in-band noise pattern obtained from the database 575 or other classification mechanism with the in-band waveform.

[0149] In one embodiment, in a noise characteristic identification stage (no received imaging energy is present), the detection band noise characteristic identification array is first processed to extract and store one or more features. The time interval for detecting a given noise pattern may also be determined and stored in this stage (578). The corresponding correlation time pattern of the in-band noise characteristic identification array is also stored (575) (e.g., stored in a look-up table).

[0150] According to this example, in the imaging stage performing noise removal, the detection band imaging array of one or more detection band imaging waveforms is mostly noisy and is processed through the same feature extraction process. A weight vector (assigning a weight or prior probability to each noise classification pattern) may be optionally obtained. The repetition frequency of each pattern can be retrieved from a database 578 generated in the noise characteristic identification stage or from a pre-stored database. The weight of each classification can be dynamically adjusted to depend on the repetition frequency of the pattern, the time when the pattern was previously detected, and the certainty with which the pattern was previously detected. The features (and optionally the classification weights) extracted from the detection band imaging waveform are fed to a trained prediction model (trained in the noise characteristic identification stage). The prediction model identifies the noise patterns in the detection band imaging waveform and assigns them a classification. The corresponding correlated in-band noise patterns of the noise classification are obtained from the database 575. The database 575 stores, for example, in the noise characteristic identification stage, the features of the in-band noise waveform patterns and the detection band waveform patterns for each noise classification (e.g., in-band time waveforms stored in a look-up table). The in-band noise waveform pattern extracted from the classification comparison is, for example, the average of all co-existing in-band noise patterns of the current noise classification, or the in-band noise pattern whose co-existing detection band pattern features are closest to the features of the current detection band imaging array (e.g., as determined by a nearest neighbor calculation). This co-existing pattern is subtracted from the input after amplitude and delay adjustments, thereby obtaining a noise-suppressed in-band imaging waveform.

[0151] 7A and 7B show an embodiment in which a single detection band waveform (with at least some energy above the imaging band) is generated by a single detection bandpass filter 410 that directs the input to the pattern recognizer 570. Alternatively, multiple detection band waveforms may be generated by multiple detection bandpass filters (with at least one detection band waveform being out of band).

[0152] In addition to at least one detection band waveform carrying out-of-band noise, one or more detection band waveforms may also carry noise in all or part of the imaging band (i.e., in-band noise detection waveforms). Such in-band data is useful for the pattern recognizer 570 to verify that the noise predicted by the out-of-band noise detection waveforms is actually present within the imaging band (either at the noise characterization stage or during imaging).

[0153] For example, the pattern recognizer can use the in-band noise detection imaging waveform to identify noise sources. For example, imaging sub-bands and corresponding noise sources can be identified if the sub-bands of the imaging band are substantially different from one or more other sub-bands of the imaging band or if the net energy within the imaging band is substantially different. For example, an in-band noise detection waveform in the 7-13 MHz imaging band can be obtained using a peak filter centered at 8 MHz, and an out-of-band noise detection waveform can be obtained using a detection band pass filter with a pass band of 15-25 MHz. If the 8 MHz band in-band noise detection waveform detects increasing energy relative to the energy of the 15-25 MHz out-of-band noise detection waveform, the pattern recognizer can adjust its weights to detect the particular noise source (i.e., noise classification). If relying solely on out-of-band information for the imaging band, the system can select a correlated in-band noise pattern from database 573 to remove the 8 MHz peak.

[0154] Although the above embodiments have been described in the context of detecting temporal patterns in raw or envelope processed signals, it should be understood that the above algorithms are also applicable to embodiments using image data. For example, image data (e.g., B-mode image data) can be processed to determine spatial noise patterns instead of processing time domain (e.g., RF or envelope detection) signals. These alternative embodiments are illustrated by the dotted lines in FIG. 7B, where decimation (220 and 221) and B-mode image line generation (230 and 231) are performed prior to 570 and 575.

[0155] Alternatively, spatial patterns can be detected using 2D image windows when processing images in the spatial domain, e.g., B-mode image data. In B-mode data, texture features are extracted in the spatial domain (e.g., gray-level co-occurrence matrix), frequency domain (e.g., Fourier spectrum measurements), or spatial frequency domain (e.g., energy of 2D wavelet coefficients).

[0156] Please refer to Figure 7D. An alternative embodiment is shown in which a reference noise detection waveform is substituted for the detection band imaging waveform of Figure 7B when performing pattern recognition during imaging. Similarly, in Figure 7C, a reference noise characteristic discrimination waveform is used in the initial pattern recognition stage performed when no imaging signal is present. Thus, the above algorithm or method shown in Figures 7A and 7B is adjusted in this embodiment by replacing the detection band noise characteristic discrimination waveform (and corresponding array measurement) with the reference noise characteristic discrimination waveform (shown in Figure 7C) and replacing the detection band imaging waveform (and corresponding array measurement) with the reference noise detection waveform (shown in Figure 7D).

[0157] <Embodiment 5: Noise suppression based on relative energy measurement> In this embodiment, noise suppression is performed by selectively attenuating a window portion of the in-band imaging waveform based on a condition evaluated according to measurements from one or more detection band imaging waveforms (at least one of which is an out-of-band imaging waveform). The attenuation may be, for example, by subtracting a derived subtraction value from the envelope of the windowed in-band imaging waveform and / or by multiplying the windowed in-band imaging waveform or its envelope by an attenuation factor. The subtraction value and / or the attenuation factor are determined from measurements of the noise detection imaging waveform or the reference noise detection waveform.

[0158] According to the first stage of this example, the noise characterization stage detects energy in both the imaging band (in-band) and the noise detection bands when no imaging energy is present (e.g., in a non-imaging noise characterization stage), thereby obtaining a correlation measurement of in-band noise and detection band noise. At least one noise detection band is out-of-band. Samples of the in-band noise characterization waveform and the detection band noise characterization waveform are recorded as an array pair. An array pair is a simultaneously recorded sampled in-band waveform and a corresponding sampled detection band waveform.

[0159] Optionally, the noise characteristic identification stage may include another stage, called the baseline noise characteristic identification stage, which is performed when it is known or expected that the imaging transducer receive channels are not receiving imaging energy and noise energy. As shown in FIG. 6A, an in-band baseline noise characteristic identification array may be obtained by applying an imaging bandpass filter 200 to the input waveform detected from the imaging transducer receive channels 13 when no imaging energy and noise energy are being received, and optionally detecting (210) the envelope of the filtered waveform. A detection band baseline noise characteristic identification array may be obtained by applying a detection band filter 410 to the input waveform from the imaging transducer receive channels 13, and optionally detecting (411) the envelope of the filtered waveform. The in-band baseline noise characteristic identification array and the detection band baseline noise characteristic identification array are measured when no imaging energy and noise energy are present, and are denoted as Gi and Gn, respectively. A given array pair may optionally be segmented according to multiple time windows, as shown at 565 and 566 in FIG. 6A, to obtain a window array pair (Gi w and Gn w) can be obtained. The window can be, for example, a sliding window, optionally overlapping with adjacent windows. In the baseline noise characterization stage, one or more noise measurements can be calculated from the energy measurements per window. For example, denote the maximum power in the windowed in-band baseline noise characterization array as Ti. Similarly, denote the maximum power in the windowed out-of-band baseline noise characterization array as Tn.

[0160] The noise characteristic identification stage includes a stage where the imaging transducer receive circuit is assumed to receive no imaging energy but noise energy. See FIG. 6A. When no imaging energy is received, the in-band noise characteristic identification array 407 can be obtained by applying the imaging bandpass filter 200 to the input waveform from the imaging transducer receive channel 13, and optionally detecting (210) the envelope of the filtered waveform. The detection band noise characteristic identification array can be obtained by applying the noise detection band filter 410 to the input waveform from the imaging transducer receive channel 13, and optionally detecting (411) the envelope of the filtered waveform. The in-band noise characteristic identification array 407 and the detection band noise characteristic identification array 408 are measured when no imaging energy is present, and are shown as Ci and Cn, respectively (as shown in FIG. 6A).

[0161] Optionally, a given array pair may be segmented according to multiple time windows to form a windowed array pair (Ci w and Cn w, where Tn is a time window, and Tn is a time window that may be obtained by performing a noise characteristic identification step. The window may be, for example, a sliding window, optionally overlapping with adjacent windows. In another embodiment, the window may have a center frequency around a peak amplitude of the noise waveform in one or more of the detection bands and / or imaging bands. In another embodiment, the time window may be fixed in time to a noise onset determined when the noise amplitude in one or more of the detection bands and / or imaging bands exceeds a specified threshold. For example, the threshold may be proportional to the parameters Tn and / or Ti obtained in the baseline noise characteristic identification stage.

[0162] In-band and detection band noise characteristics discrimination array pair Ci w and Cn w 6A to obtain one or more measurements corresponding to the energy of the imaging band and the noise detection band for each time window, thereby characterizing the relative intensity of the noise in the two bands. w and Cn w For each pair, the power of the imaging band and the power of the noise detection band are calculated (570 and 572).

[0163] The noise characterization measurement result may optionally be calculated only from selected windows. The selection condition may be evaluated according to the in-band energy measurement result and optionally the detection band energy measurement result. For example, only windows in which the in-band power exceeds a specified threshold are selected for obtaining the noise characterization measurement result (as shown in FIG. 6B). The threshold is proportional to Ti obtained in the baseline noise characterization stage. In another example, only windows in which the out-of-band power exceeds a specified threshold are selected for obtaining the noise characterization measurement result. The threshold may be derived from Tn obtained in the baseline noise characterization stage.

[0164] It should be understood that in the above examples, the maximum and minimum values ​​refer to the upper and lower percentiles, or the true maximum and minimum values. For example, the 98th and 2nd percentiles could be used in place of the maximum and minimum values. Other statistical thresholds (e.g., 95th and 5th percentiles, 90th and 10th percentiles, 80th and 20th percentiles, etc.) could also be used to represent the maximum and minimum values ​​for characteristic discrimination.

[0165] From the energy measurements for each window in the noise characteristic identification stage, one or more noise measurements can be calculated and used to define the relationship between the imaging band power and the detection band power in the presence of a noise source. For example, one or more pairs of in-band and detection band power values ​​can be selected as inflection points for generating a piecewise linear function to define the relationship between the imaging band power and the detection band power in the presence of a noise source (as shown in FIG. 6B).

[0166] In one example, a piecewise linear function that defines the relationship between the in-band power and the detected band power in the presence of a noise source can be generated based on the maximum and minimum power in the noise characterization stage (shown in FIG. 6B). The maximum and minimum in-band power values ​​from the noise characterization stage are evaluated (e.g., using absolute maximum / minimum values ​​or statistical measurements) and denoted as Pi, respectively. min and Pi max In one embodiment, the in-band power is expressed as Pi min Identify a set of windows that are within a preselected range (e.g., percentile range) for Pn min Similarly, the in-band power is selected as Pi max Identify a set of windows that are within a preselected range for Pn max Select as the power pair (Pn min ,Pi min ) and (Pn max ,Pi maxUsing [[ID=]], a function that defines the estimated relationship between the in-band power and the detected band power can be fitted. The examples are non-limiting examples for selecting the in-band power and the detected band power to provide appropriate fitting points and / or for providing a functional relationship between the in-band power and the detected band power, and it should be understood that other methods can be used.

[0167] Optionally, the ratio of the power of the in-band noise characteristic identification waveform to the power of the detected band noise characteristic identification waveform can be calculated for each window, and the maximum ratio R between multiple windows off (An example of the use of this quantity is when determining whether to apply noise correction during imaging, which will be described below) can be obtained.

[0168] One or more relationships f(pn) between the power of the in-band noise characteristic identification waveform (Pi) and the power of the detected band noise characteristic identification waveform (Pn) can be obtained. For example, as shown in FIG. 6B, f(Pn) is a piecewise linear function whose slope, intercept, and / or inflection point are calculated at the noise characteristic identification stage for (Pn min , Pi min ) and (Pn max , Pi max ). In another example, f(Pn) is a non-linear polynomial or a combination of a linear polynomial and one or more non-linear polynomials. In another example, f() is a set of values defined for a range of one or more Pn values. For example, f(Pn) is Pia when a1 ≤ Pn < a2, Pib when b1 ≤ Pn < b2, and so on. [a1, a2] and [b1, b2] are non-overlapping intervals of Pn. A window set of the detected band noise characteristic identification array whose power is between a1 and a2 can be identified, and Pia in the interval [a1, a2] can be determined using the corresponding window of the in-band noise characteristic identification array. For example, Pia is a representative power value (e.g., maximum, minimum, average, median, other statistical values) calculated from the in-band power measurement results of all windows where the detected band power Pn is within the interval [a1, a2].

[0169] The noise measurements obtained in the noise characteristic identification stage can be used to perform noise suppression of the in-band imaging waveforms obtained during imaging (as shown in Figures 6C and 6D). The input waveforms are filtered (200 and 410) and, optionally, envelope detected (210 and 411) to provide the in-band imaging waveform 407 and the detection band imaging waveform 408. The waveforms are sampled to obtain the in-band imaging array and the detection band imaging array.

[0170] In this embodiment, the term "intraband imaging array" refers to a sampled intraband imaging waveform. The term "detection band imaging array" refers to a sampled detection band imaging waveform. A set of arrays is recorded, each corresponding to a given scan line. For example, a first intraband imaging array corresponds to a first scan line, a second intraband imaging array corresponds to a second scan line, etc. The intraband and detection band imaging arrays are each referred to as Qi θ and Qn θ where θ is an index that identifies a given acquisition period (e.g., corresponding to one scanline).

[0171] The arrays can be windowed using the same window characteristics used to window the noise characteristic discrimination array in the characteristic discrimination stage, as shown at 565 and 566 in FIG. 6C. The in-band imaging array and the detection band imaging array are segmented in time according to the window, and Qi θ,w and Qn θ,w The symbol w is an integer that indicates the window number. For example, Qi 1,10 is the 10th window portion of the in-band imaging array corresponding to the 1st scan line. For each window portion of the in-band imaging array and the detector band imaging array, calculate power or other suitable energy measurements (570 and 572). The in-band and detector band power values ​​are calculated for each window and are denoted as P(Qi θ,w ) and P(Qn θ,wThese energy measurements are used to suppress noise based on measurements taken in the noise characterization stage.

[0172] In one embodiment, as shown in FIG. 6C, an intraband imaging array Qi θ,w The noise is suppressed by subtracting (525) the power value from the envelope of the noise detection band power P(Qn θ,w ) based on the function f() obtained in the noise characteristic identification stage, the intra-band imaging array window Qi θ,w The noise energy P^i Nθ,w For example, Qi θ,w Noise energy P^i in Nθ,w is P^i Nθ,w =f(P(Qn θ,w )) can be estimated (575). The estimated in-band noise P^i Nθ,w can be optionally scaled by multiplying it by a scaling factor β (0≦β≦1). The scaled noise estimate βP̂i Nθ,w Optionally, clamp to an upper limit (e.g. 0.8×P(Qi θ,w ), 0.9×P(Qi θ,w ), 1.0×P(Qi θ,w ), but not limited to, to obtain the subtraction value. θ,w The subtraction value is subtracted (525) from the envelope of the noise suppression waveform to obtain the noise suppression waveform envelope.

[0173] In another example shown in FIG. 6D, an intraband imaging array Qi θ,w By multiplying the elements of Qi by the attenuation coefficient (526), ​​noise can be suppressed. θ,w The in-band power of P(Qi θ,w ) can be calculated as the noise detection band power P(Qn θ,w ) based on the function f() obtained in the noise characteristic identification stage, the in-band imaging window Qi θ,w Noise energy P^i in Nθ,w For example, Qi θ,w Noise energy P^i inNθ,w is P^i Nθ,w =f(P(Qn θ,w )) can be estimated (575). The estimated in-band noise P^i Nθ,w can be optionally scaled by multiplying it by a scaling factor β (0≦β≦1). The damping factor is [P(Qi θ,w )-βP^i Nθ,w ] / P(Qi θ,w ) can be chosen to be proportional to

[0174] In an embodiment, the scaling factor β may be selected in the range of 0 to 1. Attenuation of ultrasound energy reduces the imaging energy over time, and the determination of β is based on the window depth within the waveform (which therefore corresponds to the depth within the imaged tissue). The parameter β may be optionally user controlled. The attenuation coefficient may be optionally clamped to an upper limit (e.g., but not limited to, 0.95, 0.9, 0.8, etc.). Additionally or alternatively, the attenuation coefficient may be clamped to an upper limit (e.g., but not limited to, 0, 0.01, 0.05, 0.1, etc.). The attenuation coefficients are then scaled to an array Qi θ,w or multiplied with its envelope to obtain a noise suppression array.

[0175] <Pattern Recognition for Classification-Specific Noise Suppression> During the noise characteristic identification stage (e.g., FIG. 6A), the system can optionally be configured to group the array pairs Cnw and Ciw into one or more categories (called classes). Referring again to the pattern recognizer in step 570 of FIG. 7A, one or more detection band noise characteristic identification waveforms can be processed by the pattern recognizer to identify one or more noise pattern classes and assign classes to time windows corresponding to the identified noise patterns. A set of windows of the detection band noise characteristic identification waveforms belonging to class k is selected, denoted as {w_k}. The in-band and out-of-band power measurements of the windows in the set {w_k} are used to find a class k-specific functional relationship f k () can be derived.

[0176] The system can be configured to use a pattern recognizer to identify noise patterns when the imaging transducer circuitry is receiving imaging energy. Using a method similar to that described in step 570 of FIG. 7B, one or more detection band imaging waveforms can be processed by a pattern recognizer to identify one or more noise pattern classifications and assign a classification to the time window corresponding to the identified noise pattern. The estimated in-band noise power can be derived from the identified classification-specific functional relationship f(). For example, for a window belonging to classification 1, the function f1() can be used to calculate the estimated in-band noise (i.e., P̂i Nθ,w =f1(P(Qn θ,w Similarly, for windows belonging to category 2, the estimated in-band noise can be derived using function f2(). f1(), f2(), f3(), etc. are obtained in the noise characteristic identification stage. As mentioned above, the window Qi θ,w The noise inside the Qi θ,w By subtracting the subtraction value from the envelope of, or the damping factor and Qi θ,w The subtraction value or attenuation coefficient is P^i Nθ,w As a specific example, the noise classified by the pattern recognizer as Class 1 is from the electro-anatomical mapping system, and the noise classified as Class 2 is from the ablation energy generator. Thus, when the system recognizes that the noise arrived from the mapping system, it can estimate the in-band noise generated by the electro-anatomical mapping system based on the power of the noise detection band using f1(). Also, when the system recognizes that the noise arrived from the ablation generator, it can estimate the in-band noise generated by the ablation generator based on the power of the noise detection band using f2().

[0177] In another example, the system outputting the noise may be monitored to determine the functional relationship to use for noise estimation and suppression. For example, the control of the ablation generator may be monitored to enable a binary gate signal when the ablation generator is generating energy (and noise). This gate signal may be used to determine the time period during which the function f2() should be used for in-band noise estimation.

[0178] <Selective implementation of noise suppression> The energy measurements can optionally be used to estimate whether an intraband imaging array is experiencing low signal to noise ratio, or in other words, to classify a window as having or not having noise suppression correction (through subtraction or multiplication of an attenuation factor).

[0179] In one embodiment, the decision on whether to apply noise suppression correction to the window w can be made based on the power detected in the noise detection band. P(Qn θ,w ) exceeds the specified threshold, noise correction is θ,w The threshold can be obtained in the noise characteristic identification stage.

[0180] In another example, the decision on whether to apply noise suppression correction for a given window can be based on the ratio of imaging band power to noise detection band power. For example, in the noise characterization stage, for each window of the pair of in-band noise characterization array and detection band noise characterization array, the ratio of imaging band power to detection band power is obtained. The representative maximum ratio (R Off ) can be used as a threshold to determine whether or not to apply noise suppression correction when the imaging transducer receive circuitry is receiving imaging energy, as described above.

[0181] The decision whether to apply noise suppression correction to a window w of the in-band imaging waveform is determined by the ratio of the imaging band power to the noise detection band power (ROn w In one embodiment, R On w γR Off γ is the relaxation parameter and R Off is the representative maximum ratio calculated in the noise characteristic identification stage. R On w is γR Off If R is greater than , the signal-to-noise ratio of the in-band signal is assumed to be large enough, and no noise suppression correction is applied. On w is γR Off If it is less than or equal to, then the signal to noise ratio is estimated to be small enough and noise suppression correction is applied.

[0182] The value of γ is used to adjust the sensitivity of the signal-to-noise ratio, and in some cases the value of γ can be used as an adjustment factor when weaker signal parts are suppressed. Lowering the value of γ lowers the threshold for applying noise suppression, which reduces the number of windows to be suppressed, leaving more imaging energy (and noise) in the final output.

[0183] In an embodiment, γ can be selected between zero and a unity value. In ultrasound, the imaging energy decreases over time due to attenuation of the ultrasound energy, and the determination of γ depends on the window depth in the waveform (i.e., the depth in the tissue being imaged). The parameter γ may be optionally user controlled. For example, if the tissue or other structures in the image are undesirably or excessively attenuated, the user may decrease the value of this parameter to reduce the effectiveness of noise suppression.

[0184] In the window identified for noise suppression, any suitable noise removal or suppression method can be used to suppress the noise of the intra-band imaging array within the window. It should be understood that various noise suppression corrections can be applied, including but not limited to a subtraction correction as shown in FIG. 6C and / or a multiplication of an attenuation factor as shown in FIG. 6D.

[0185] Noise suppression may result in erroneous noise removal for each window. For example, windows with low imaging energy may be adversely affected by noise suppression based on erroneously determining the signal-to-noise ratio in the window (i.e., incorrect window classification). This may result in small image "holes" in the peripheral uniform areas of the image, or residual noise pixels in the peripheral low-noise areas of the image. In one embodiment, the status of neighboring windows (i.e., spatial neighbors) may be used to determine whether a window determined to be suitable for noise suppression should actually apply the process. If a given window is determined by the above method to be suitable for noise suppression, neighboring windows in one or more neighboring arrays (i.e., arrays corresponding to neighboring scan lines) may be used to evaluate whether noise suppression should be performed by the attenuation coefficient of the given window.

[0186] For example, for a given scanline, if a given window is determined to be unsuitable for noise suppression and one or more adjacent windows are determined to be suitable for noise suppression, the window can be flagged as potentially misclassified. The state of a given window can be overridden, the given window identified as suitable for noise suppression, and noise suppression applied to the given window. Conversely, if a given window is determined to be suitable for noise suppression and adjacent windows in an adjacent array are determined to be unsuitable for noise suppression, the initial determination of the state of the given window is overridden and the given window is flagged as unsuitable for noise suppression and potentially misclassified. Samples of the misclassified window can be replaced by samples of one or more noise-free adjacent windows (i.e., replaced or interpolated by samples of adjacent windows determined to be unsuitable for noise suppression according to the method described above), optionally after performing delay and amplitude adjustments.

[0187] FIG. 6E shows an example implementation of this method. Neighboring windows in neighboring arrays are examined to determine whether the classification of the current window is consistent with the surrounding windows in neighboring arrays. Since two windows (array 2, windows 4 and 5) classified as mainly noisy (marked "N") are surrounded by windows (marked "S") classified as having a sufficiently high signal-to-noise ratio, the two "N" windows are replaced (e.g., by copying or interpolation with amplitude and / or shape adjustment) with samples from the neighboring "S" windows, as shown in FIG. 6F, and noise suppression by noise estimation is not performed on these windows. This example uses two neighboring windows on either side of the array when checking the status of a given window, but any number of windows may be used in other embodiments.

[0188] One or more temporally adjacent windows before and after a given window can also be used when evaluating whether the classification of the given window should be changed. For example, in Figure 6E, window 4 of array 5 is initially marked as "S" and in Figure 6F is reclassified as "N" because the preceding and following windows in the array are marked as "N".

[0189] Although this embodiment performs signal processing before image processing, this embodiment can be applied to image data processing rather than time domain signal processing. For example, multiple intra-band image arrays and detection band image arrays representing multiple adjacent scan lines can be acquired, and frames of intra-band images and detection band images can be acquired by post-processing. The intra-band and detection band image pixels can be denoted by Bi, θ,d and Bn θ,d where θ represents the scan line and d represents the depth. Using the detection band image, the attenuation (i.e., what is subtracted or multiplied) value of each pixel in the in-band image can be evaluated. These attenuation values ​​can be obtained pixel by pixel from the corresponding pixels in the detection band image (i.e., Bn θ,d Using the value in θ,dcan be obtained by attenuating pixel intensities at Bn θ,d Using the values ​​of the surrounding 3 × 3 adjacent positions (e.g., polar or Cartesian coordinates), θ,d attenuating pixel intensity at

[0190] Please refer to Figures 6G, 6H, and 6J. Another embodiment of using a reference receiving channel when performing noise characteristic identification (Figure 6G), another embodiment of determining an appropriate subtraction value for noise suppression during imaging (Figure 6H), and another embodiment of determining an attenuation coefficient (Figure 6J) are shown. In Figure 6G, when performing noise characteristic identification, a reference noise characteristic identification waveform 409 is used instead of the detection band noise characteristic identification waveform of Figure 6A. The reference channel filter is, for example, an imaging bandpass filter. Alternatively, if noise estimation is performed from an input of out-of-band noise, the reference channel filter is not an imaging bandpass filter. In Figure 6H, a reference noise detection waveform 439 is used instead of the detection band imaging waveform of Figure 6B to determine a subtraction value and apply noise suppression by subtraction. Similarly, in Figure 6I, a reference noise detection waveform 439 is used instead of the detection band imaging waveform of Figure 6B to determine and apply an attenuation coefficient for noise suppression. Thus, the method of Figures 6A to 6F can be applied to this embodiment by replacing the detection band noise characteristic identification waveform 408 (and corresponding power measurements) with the reference noise characteristic identification waveform 409 of Figure 6G, and by replacing the detection band imaging waveform 438 (and corresponding array and power measurements) with the reference noise detection waveform 439 shown in Figures 6H and 6I.

[0191] <Embodiment 6: Noise suppression of quasi-periodic noise sources> In this embodiment, noise suppression is performed by estimating and subtracting in-band noise, which is assumed to arise from a quasi-periodic noise source or a quasi-periodic sequence of noise sources. The in-band noise is estimated based on measurements made in a noise characterization stage when the imaging transducer receive circuitry is not receiving imaging energy.

[0192] According to the first stage of this embodiment, waveforms within the imaging band and within the noise detection band are detected and sampled when no imaging energy is present (e.g., when the ultrasound transducer is not receiving imaging energy), thereby obtaining a pair of simultaneously existing in-band and detection band (out-of-band) noise characteristic identification arrays.

[0193] As shown in FIG. 8A, an in-band noise characteristic discrimination waveform 590 can be obtained by applying an imaging bandpass filter 200 to an input waveform detected from the imaging transducer receive channel 13, and optionally detecting 210 the envelope of the filtered data. A detected band noise characteristic discrimination waveform 595 can be obtained by applying a noise detection bandpass filter 410 to an input waveform detected from the imaging transducer receive channel 13 (at least one noise detection band includes signals from outside the imaging band), and optionally detecting 411 the envelope of the filtered data. The in-band noise characteristic discrimination waveform and the detected band noise characteristic discrimination waveform are measured in the absence of imaging energy and can be sampled to obtain an in-band noise characteristic discrimination array and a detected band noise characteristic discrimination array, denoted as Ci and Cn, respectively. It is assumed that Ci (and Cn) capture one or more periods of a periodic noise source.

[0194] Once the in-band and detection band noise characteristic discrimination arrays are obtained in the noise characteristic discrimination stage, the correlation between the detection band characteristic discrimination array and the detection band imaging array can be used to estimate adjustment parameters for subtracting the in-band characteristic discrimination array from the in-band imaging array.

[0195] As shown in Figure 8B, during imaging, the input waveforms from the imaging transducer receive circuitry 13 are filtered (200 and 410) to provide in-band and detection band imaging waveforms. These waveforms are sampled (before or after envelope detection) to obtain the in-band imaging array 537 and detection band imaging array 538. The in-band and detection band imaging arrays are denoted as Qi and Qn, respectively.

[0196] According to the method, each intra-band imaging array is processed for noise suppression by subtracting noise from the intra-band imaging array Qi using the intra-band noise characteristic discrimination array Ci. However, in order to perform noise suppression by subtraction, the intra-band noise characteristic discrimination array Ci needs to be time-aligned with the intra-band imaging array Qi so that the noises are simultaneously present. This alignment is possible in the case of a periodic noise source that generates noise in the imaging band that is correlated with the noise in the noise detection band.

[0197] Time alignment can be achieved, for example, by segmenting the intra-band and detector band imaging arrays Qi and Qn into multiple time windows 565 and 566 (as described in the above embodiment). The windows are preferably long enough to capture one or more periods of the periodic noise source. The imaging arrays, time segmented according to the windows, are then aligned in time to the Qi w and Qn w The symbol w is an integer indicating the window number.

[0198] In one embodiment, the time alignment can be realized on a window-by-window basis. In this embodiment, the time alignment is performed by dividing the detection band noise characteristic discrimination array Cn and the detection band imaging array Qn in each window. w This can be achieved by calculating the cross-correlation between the in-band noise characterization array Cn and the in-band imaging array Qi and selecting the relative time delay τ corresponding to the maximum cross-correlation (580). Due to the simultaneous existence relationship between the noise in the imaging band and the noise in the noise detection band, this time delay τ is applied to convert the in-band noise characterization array Cn to the in-band imaging array Qi wA scaling factor may be applied to the aligned in-band noise characteristic identification array.

[0199] After alignment, the window part of the in-band noise characteristic discrimination array ( Ci w ) is used as the intra-band imaging array Qi w By subtracting from the noise suppression band, the imaging array Qi w This is optionally done after taking the absolute value after subtraction or applying a floor function to remove negative values. This process is repeated for each window that is to be noise suppressed.

[0200] In one embodiment, when adjacent windows overlap, the intra-band imaging array Qi w From in-band noise characteristic identification array Ci w When subtracting the aligned window segments, a scaling factor may be applied. For example, the subtraction (and optionally modulo) is Qi w =Qi w -α Ci w where α is a windowing scaling factor, α=1-β, and β is an overlap factor. For example, when β=0.75, the scaling factor is α=0.25. It should be understood that this example is provided to illustrate an example method for scaling the subtraction component, and other functional forms may be adopted.

[0201] 8C and 8D, another embodiment using a reference receive channel is shown to perform noise characteristic discrimination and to determine appropriate amplitude adjustments for noise suppression during imaging. In FIG. 8C, the reference noise characteristic discrimination waveform 596 is used instead of the detection band noise characteristic discrimination waveform 595 of FIG. 8A when performing noise characteristic discrimination. Similarly, in FIG. 8D, the filtered reference noise detection waveform 439 is used instead of the detection band imaging waveform 595 of FIG. 8B to determine and apply amplitude adjustments. Thus, the method described in FIG. 8A and 8B can be applied to this embodiment by replacing the detection band noise characteristic discrimination waveform 408 with the reference noise characteristic discrimination waveform 409 of FIG. 8C and by replacing the detection band imaging waveform 438 (and corresponding array) with the reference noise detection waveform 439 of FIG. 8D.

[0202] <Embodiment 7: Noise suppression using multiple scans by changing the scan rate> The system can be configured to acquire two or more sets of intra-band imaging arrays corresponding to imaging energy from the same scan line or corresponding to scan lines that substantially overlap in space. The imaging energy in the intra-band imaging array set has a redundant time / depth dependency. By averaging the redundant intra-band imaging arrays (or performing other statistical processing such as evaluating the minimum value), noise can be suppressed, provided that the noise itself is not time-locked relative to the trigger that causes the imaging transducer receiving circuitry to start receiving imaging energy on each scan line. For example, if the pulse repetition frequency of the voltage pulses that excite the imaging ultrasound transducer is 200 μs, periodic noise that repeats every 2 μs will always have components of 0 μs, 2 μs, 4 μs, etc. in each intra-band imaging array. However, if the pulse repetition frequency is adjusted to 199 μs, the first imaging array will have noise components of 0 μs, 2 μs, 4 μs, etc., and the second imaging array will have noise components of 1 μs, 3 μs, 5 μs, etc. Noise can be suppressed by averaging two successive redundant intra-band imaging arrays, optionally after performing envelope detection.

[0203] In this embodiment, the noise of the detection band waveform (out of band) can be used to determine the period of one or more in-band noise sources. The system is prompted to adjust the scan rate to ensure that the imaging scan period is not an integer multiple of the period of the noise sources. For example, an autocorrelation function can be used to detect the periodicity of the detection band imaging waveform while the imaging transducer receive circuitry is receiving imaging energy. Alternatively, the period of one or more noise sources can be determined from the detection band noise characteristic identification waveform or from a reference noise characteristic identification waveform during a noise characteristic identification stage when the imaging transducer is not receiving imaging energy, or can be retrieved from a pre-stored database. The system is prompted to adjust the scan rate to ensure that the imaging scan period is not an integer multiple of the period of the noise sources.

[0204] After determining the optimal scan rate and adjusting the scan rate, the intra-band imaging waveforms acquired from multiple scan lines are sampled after envelope detection to obtain a set of intra-band imaging arrays. The intra-band imaging arrays are designated as Qi θ where θ is the scan line. The samples from the intra-band imaging array are expressed as Qi θ The noise suppression is expressed as [k], where k=1··K is the sample index and K is the number of samples in the array. The system can be configured to suppress noise by averaging (or taking the minimum or other statistical measure) multiple arrays corresponding to adjacent scan lines or scan lines with strong spatial overlap. For example, if the system is configured to group three intra-band imaging arrays, then the samples Qi θ [k] is [Qi θ-1 [k], Qi θ [k], Qi θ+1 [k]]. Optionally, the samples Qi θ [k] is [Qi θ-1 [k], Qi θ [k], Qi θ+1[k]] and after performing other numerical analyses and determining whether the sample should be noise suppressed or not, it is selectively retained. For example, [Qi θ-1 [k], Qi θ [k], Qi θ+1 In [k], if the minimum is greater than half the maximum, the sample values ​​may not be large enough to suppress noise by averaging and are therefore left unchanged.

[0205] This technique is useful in MRI imaging in the presence of periodic noise sources because it ensures that RF excitation pulses are delivered at times that are uncorrelated with the timing or periodicity of local environmental noise sources.

[0206] Another component: Selecting samples to suppress noise when using multiple scans. In this embodiment, noise suppression is performed by adjusting the scan rate as described above and selectively replacing portions of the intra-band imaging array based on statistical measurements (e.g., average or minimum) from the array of multiple adjacent scan lines, assuming there is sufficient overlap with adjacent scan lines in the scan area. Only those segments of the intra-band imaging array that are assessed as noisy using the detection band measurements are replaced.

[0207] As shown in FIG. 8E, the input waveform is filtered (200 and 410) to provide in-band and detection band waveforms, which are sampled after envelope detection to obtain in-band and detection band imaging arrays.

[0208] Data corresponding to multiple scan lines is recorded as an array set for each scan line. The array set is acquired for each scan line. One in-band imaging array is acquired within the imaging band for each scan line, and at least one out-of-band imaging array is acquired for each scan line. The in-band and out-of-band imaging arrays are each acquired using Qi θ and Qn θwhere θ is the scan line. The samples from the intra-band imaging array and the detection band imaging array are expressed as Qi θ [k] and Qn θ We denote it as [k], where k=1·K is the sample index and K is the number of samples in the array.

[0209] The detection band imaging array is segmented into multiple windows, each of which is Qn θ The windowed array contains J samples of Qn θ,w The symbol w is an integer representing the window number, θ is an integer representing the scan line, and Qn θ,w is a sample [Qn θ [kw],Qn θ [kw+1],···Qn θ [kw+J-1], where kw is the index of the first sample in the window.

[0210] Each window Qn in the detection band θ,w is evaluated 600 for the presence or absence of noise. If the noise (e.g., as determined by a measure of waveform energy, such as peak, RMS, etc.) exceeds a threshold, the window is considered to be noisy. The threshold can be selected in a noise characteristics identification stage.

[0211] According to this embodiment, each detection band array window Qn classified as having noise θ,w For the simultaneous in-band samples (i.e., Qn θ [kw],Qn θ [kw+1],···Qn θ [kw+J-1]) is identified as a suitable sample for noise suppression.

[0212] When the windows overlap, for a given sample Qi θ[k] may be associated with more than one window. There are cases where a sample is associated with both a noisy window and a noise-free window. In this case, the system can be configured to pool noise estimates from multiple out-of-band windows before determining whether the sample is suitable for noise suppression.

[0213] A statistical value (e.g., replicates, average, minimum, etc.) from the samples of the array corresponding to adjacent scan lines (calculated in 610) is used to replace the sample deemed suitable for noise suppression (620). For example, the sample Qi θ If [k] is deemed suitable for noise suppression and the system is configured to group arrays from three scan lines, the sample Qi θ [k] is [Qi θ-1 [k], Qi θ [k], Qi θ+1 [k] is replaced by the minimum value of

[0214] Please refer to Figure 8F. Another embodiment is shown in which a reference waveform is used instead of the detection band imaging waveform of Figure 8E when performing noise suppression during imaging. Therefore, the above method described in Figure 8D can be applied to this embodiment by replacing the detection band imaging waveform (and corresponding array) with the filtered reference noise detection waveform.

[0215] <Time domain vs. frequency domain processing> The above embodiments have been described in the context of time domain processing. However, many embodiments herein may use frequency domain or time-frequency domain processing in one or more steps. For example, instead of using the maximum power as well as the minimum power ratio of the imaging band and the noise detection band in steps 570 and 572 of Figures 6A and 6C, a short-time Fourier transform or wavelet transform may be performed on the in-band and out-of-band waveforms for each motion or array. Analysis of the transform coefficients (e.g., mean, root mean square, etc.) may be used to characterize or window detect noise when it is present or absent. When noise is detected, instead of attenuating the time domain signal (526), ​​the transform coefficients of the current window may be attenuated. An inverse transform may then be performed on the attenuated frequency domain signal to obtain a noise-suppressed time domain signal.

[0216] In FIG. 2A (step 510), FIG. 8B, and FIG. 8D, as well as other embodiments where cross-correlation is required between two time-continuous waveforms, a Fourier transform algorithm can be used to efficiently compute the cross-correlation.

[0217] Additionally, as described in Figures 7A-7D, machine learning algorithms can be used to classify noise patterns in step 570. These patterns can be defined by frequency domain and / or time-frequency domain characteristics, which require frequency domain or time-frequency domain processing of time-continuous waveforms.

[0218] <Generalization to areas other than ultrasound> Although the above example systems and methods for image noise suppression have been described in the context of ultrasound imaging, it should be appreciated that the embodiments herein are applicable to a variety of imaging devices, systems, and methods.

[0219] Another example of an imaging system to which noise suppression may be applied according to the above embodiment is a magnetic resonance imaging system. See FIG. 9. Another example of a system for obtaining signals to be noise suppressed from a magnetic resonance (MR) system is shown. The system includes a magnetic resonance scanner 50 that uses a main magnet 52 to generate a main magnetic field B0. The main magnetic field B0 produces polarization in a patient 60 or object. The system includes gradient coils 54 that generate magnetic field gradients. A receive coil 58 detects MR signals from the patient 60. The receive coil 58 may also be used as a transmit coil. Alternatively, a body coil 56 may be used to emit and / or detect radio frequency (RF) pulses. The RF pulses are generated by an RF unit 65, and the magnetic field gradients are generated by a gradient unit 70. The manner in which MR signals are detected using RF pulses and magnetic field gradients, and the method for reconstructing MR images, are known to those skilled in the art.

[0220] The reference receiver circuit may comprise a coil in the same room as the scanner, but not in close proximity to the imaging sample (e.g., a patient) emitting the MRI signal. Electromagnetic noise propagating near the MRI machine is detected by both the imaging receiver coil 58 and the reference receiver circuit. The coil of the reference receiver circuit is positioned and oriented to receive the same noise as the imaging receiver coil, but is far enough away from the imaging sample so that the imaging energy detected by the reference receiver circuit is negligible.

[0221] The reference receiver coil is tuned to have the same bandwidth as the imaging receive coil 58, or has another bandwidth capable of collecting environmental noise signals that are correlated with the noise received by the imaging receive coil.

[0222] The reference receiver circuit further includes a set of receiver coils, for example three coils aligned orthogonally to each other, to collect electromagnetic noise in such a way that a weighted sum of the noise collected by the three coils more closely matches the noise collected in the imaging receiver coils, thereby taking into account the directionality of the dominant electromagnetic noise sources in the MRI environment.

[0223] It should be understood that the MR system may have other units or components that are not shown for clarity, such as, but not limited to, control devices, input devices, sensing devices (e.g., cardiac and / or respiratory synchronization devices), etc. Furthermore, each unit may be realized in a manner other than as separate units as shown. Different components may be incorporated into a unit or different units may be combined. Each unit (shown as a functional unit) may be designed as hardware, software, or a combination of both.

[0224] In the system shown in Figure 9, control and processing hardware 100 acquires magnetic resonance images of a patient 60 in accordance with an appropriate pulse sequence. The control and processing hardware 100 interfaces with a magnetic resonance imaging scanner 50 which receives the acquired images and controls the image acquisition. The control and processing hardware 100 receives image data from an RF unit 65 and processes the image data in accordance with the methods described above.

[0225] The control and processing hardware 100 can be programmed with a set of instructions that, when executed by a processor, causes the system to perform any one or more of the above methods, thereby suppressing noise in signals acquired from a magnetic resonance imaging system. For example, as shown in FIG. 9, the control and processing hardware 100 can be programmed with instructions in the form of a set of executable image processing modules, including, but not limited to, a pulse sequence generation module (not shown), an image acquisition module (not shown), an image processing module 145, and a noise suppression module 150. The pulse sequence generation module, image acquisition module, and image processing module can be implemented using pulse sequence generation, image acquisition, and image reconstruction algorithms, respectively, known to those skilled in the art. RF data is received from RF coils 56 and / or 58, and optionally from one or more reference receive coils. The data is sampled and filtered to obtain an in-band waveform. A reference waveform via a reference receive circuit or a noise detection band waveform that is RF filtered from coils 56 and / or 58 is acquired. One or more of the noise suppression methods described in FIGS. 2-8 can be used for noise suppression (100). The pulse generation module generates a sequence of RF pulses and magnetic field gradients depending on the desired imaging sequence. The image acquisition module stores the MR signals detected by the coils 56 and / or 58 in raw data space. The image processing module 145 processes the acquired noise-suppressed RF data to perform image reconstruction of the MRI image.

[0226] The ability to detect noise that correlates with noise in the bandwidth of the imaging signal (either via a reference receive channel or a noise detection band) provides the ability to estimate in-band noise and improve the imaging signal by removing the estimated noise from the imaging signal.

[0227] This can improve the SNR in cages that shield the MRI from environmental noise, or improve the operation of the MRI system in open / unshielded environments that are more susceptible to noise.

[0228] <Example> The following examples are presented to enable those skilled in the art to understand and practice embodiments of the present disclosure, and should not be construed as limiting the scope of the present disclosure, but are merely illustrative and illustrative.

[0229] <Example 1: Noise suppression of unknown noise source by attenuation coefficient (example of embodiment 5)> This example involves acquiring ultrasound data using an intracardiac echo (ICE) system in the presence of two noise sources. The transducer was configured to detect ultrasound energy at a frequency of 9 MHz. Two bandpass filters were used in parallel to separate the radio frequency (RF) signal into an imaging band of 7-13 MHz and a noise detection band of 15-25 MHz, which is beyond the frequency range of the emitted ultrasound.

[0230] The first noise source is an electroanatomical mapping system (Carto®3). The system has an electromagnetic tracking module and an impedance-based tracking module, and patches are attached to the patient to estimate impedance and device position. These patches can couple a large amount of noise into the imaging band of the ICE images. In this experiment, a cardiac dynamic phantom was used in a saline bath. The electrodes from the impedance patches were immersed in the water bath. The second noise source was due to turning on the ablation generator connected to the Carto3 console. The noise generated from this second noise source was assumed to be noise propagating from the ablation generator to the saline via the Carto3 console and the patch electrodes when the generator was turned on.

[0231] Figure 10A shows an ultrasound image acquired when no noise sources are present, while Figures 10B and 10C show the effect of the first and second noise sources, respectively, on image quality.

[0232] Noise suppression of the ultrasound waveform detected by the ultrasound transducer of the ICE console was performed according to the method shown in Figures 6A and 6D.

[0233] In the first baseline noise signature identification stage of this experiment, energy in the imaging band was detected in the absence of received imaging energy and received noise energy. This resulted in the acquisition of an in-band baseline noise signature identification array (denoted as Gi). The in-band baseline noise signature identification array was acquired by sampling a 125 μs in-band waveform at 200 MS / s. Using sliding and overlapping windows (window size = 64 samples, 20% overlap), an in-band power measurement was calculated for each window. The representative maximum power (90th percentile of all windows) was calculated and this value was assigned to a threshold value Ti.

[0234] In the second noise characteristic identification stage of this experiment, when no incoming imaging energy was present (i.e., when the transducer was not being voltage pulsed and was not receiving ultrasonic energy), energy was detected both in the imaging band and in the noise detection band. This resulted in a correlation measurement of the in-band noise and the noise detection band noise (Ci and Cn). The noise detection band was configured as a frequency band of approximately 15 to 25 MHz.

[0235] In this example of an ICE system, 512 waveforms (each with a duration of 125 μs and sampling at 200 MS / s) were acquired. Thus, 512 in-band and detection band noise characteristic discrimination waveform pairs were used to estimate the noise characteristics. For both the imaging band and the noise detection band, sliding and overlapping windows (window size = 64, 20% overlap) were used to estimate the in-band and detection band noise characteristic discrimination arrays (Ci w and Cn w ) window pair is obtained. w is an integer indicating the window number. w and Cn w For each pair, the power of the imaging band and the power of the noise detection band were calculated. Only windows with in-band power greater than a threshold Ti were selected for noise characteristic discrimination.

[0236] Statistical noise power measurements were calculated for noise characterization. A set of windows for the in-band noise characterization array whose power was between the 96th and 99th percentiles of the in-band noise characterization array power values ​​was selected. Within this set, the window with the minimum near-zero noise detection band power (the 20th percentile within the set) was selected as w_max, and the power value P(Cn w_max ) and P(Ci w_max ) were calculated (Pn max and Pi max Similarly, a set of windows was selected for the in-band noise characteristic discrimination array whose in-band power was between 1 and 5 percentiles of the in-band noise characteristic discrimination array power values. Within this set, the window with the minimum near-zero noise detection band power (the 20th percentile within the set) was selected as w_min, and the power value P(Cn w_min ) and P(Ci w_min ) were calculated (Pn min and Pi min (denoted as

[0237] As shown in FIG. 6B, the point (Pn max ,Pi max ) and points (Pn min ,Pi min ) has slope m and y-intercept c. The relaxation parameter β was set to unity. The function f can be written as:

number

[0238] The function f() was used to perform noise suppression of the in-band imaging waveform acquired during imaging (i.e., when the pulser was periodically emitting imaging energy and the transducer receiving circuit was receiving ultrasonic energy).

[0239] The waveforms of the imaging band and the noise detection band are sampled and the pair of the in-band imaging array and the detection band imaging array (Qi θ and Qn θ ) was acquired. θ is an index that identifies the acquisition period corresponding to the scan line. The imaging array is segmented in time according to the window, and Qi θ,w and Qn θ,w where w is an integer indicating the window number. For each window portion of the in-band imaging array and the out-of-band imaging array, the power P(Qi θ,w ) and P(Qn θ,w ) were calculated (Pi θ,w and Pn θ,w (denoted as

[0240] For each window, the out-of-band imaging power Pn θ,w Using the in-band imaging window Qi θ,w Noise energy P^i in Nθ,w We estimated P^i Nθ,w is f(Pn θ,w ) was calculated. β was set to unity, and the parameters m, c, and Pn max、 Pn min Obtained P^i Nθ,w Pi θ,w The damping coefficient is clamped to [Pi θ,w -P^i Nθ,w ] / Pi θ,w The attenuation coefficient was clamped to 0.02 or more. The attenuation coefficient was calculated by the array Qi θ,w Multiplying by , the noise attenuation array Q i θ,w The image generator (230) obtains the following for all 512 scan lines: Q i θ was included and processed.

[0241] The above algorithm was used to suppress noise in the in-band imaging waveform when the first and second noise sources were present. Figures 11A and 11B show images acquired when the first noise source was present, with and without the application of the noise suppression algorithm, respectively. Figures 12A and 12B show images acquired when the second noise source was present, with and without the application of the noise suppression algorithm, respectively. In both cases, the signal-to-noise ratio is clearly improved (about 6 dB improvement). As another example, Figures 11C, 11D, and 11E show images acquired by applying the noise suppression method when the first noise source was present. The relaxation parameter β of the function f() is 0.5 (Figure 11C), unity (Figure 11D), and 1.5 (Figure 11E).

[0242] Example 2: Noise suppression of periodic noise sources using delay correction (embodiment 6) In this example, noise suppression of the ultrasound waveform detected by the ultrasound transducer of the ICE console was performed according to the method shown in Figures 8A and 8B. The data in this example was collected by the ICE console in the presence of an electromagnetic tracker. The control unit of the electromagnetic tracker was observed to produce a quasi-periodic noise pattern in the ultrasound image data, which is shown in Figure 13A.

[0243] In this example for the ICE system, one waveform (125 μsm long, sampling 200 MS / s) was acquired when the imaging transducer was not receiving imaging energy and collecting data in raw / RF mode.

[0244] The data was stored as array pairs (i.e., sampled waveforms). The first array contains the in-band noise signature discrimination waveform (7-13 MHz) sampled after envelope detection, and a corresponding array containing the out-of-band noise signature discrimination waveform in the 15-25 MHz band sampled after envelope detection. These in-band and detection band noise signature discrimination arrays are denoted as Ci and Cn.

[0245] The correlation between the in-band and out-of-band noise was used to perform noise suppression of the in-band imaging waveform during image data acquisition (ie, when the ultrasound transducer is receiving ultrasound energy).

[0246] During imaging, array pairs were acquired with multiple acquisition periods corresponding to the scan line θ. For each array pair, one array was acquired in the imaging band (7-13 MHz) and one array was acquired in the noise detection band (15-25 MHz). Each array was 125 μs long with sampling of 200 MS / s. The in-band and out-of-band imaging arrays were respectively Qi θ and Qn θ and was measured during imaging (i.e., when the transducer was receiving imaging energy).

[0247] Each intra-band imaging array was processed for noise suppression using an intra-band noise characteristic discrimination array to suppress the noise from the intra-band imaging array.

[0248] In-band and out-of-band imaging array Qi θ and Qn θ The imaging array, segmented in time according to the windows, is then divided into multiple time windows. In this example, the window size is 800 samples (4 μs at a rate of 200 MS / s in this example) with 75% overlap between adjacent windows. The imaging array, segmented in time according to the windows, is then divided into multiple time windows, called Qi θ,w and Qn θ,w where w is an integer indicating the window number.

[0249] Qn θ,w The time alignment between the out-of-band noise characteristic discrimination array Cn and the out-of-band imaging array Qn was performed for each window. θ,wThis was achieved by calculating the cross-correlation between and selecting the relative time delay τ corresponding to the maximum cross-correlation. Due to the correlation between the in-band and out-of-band noise, a time delay τ was applied to align the in-band noise characteristic discrimination array to the windowed in-band imaging array. After alignment, the in-band noise characteristic discrimination array was windowed ( Ci w (represented as ).

[0250] Intraband Imaging Array Qi θ,w Before subtracting from the in-band noise characteristic identification array Ci w A scaling factor was applied to Qi θ,w =Qi θ,w -α Ci w According to the above, a noise suppression in-band imaging array Qi θ,w was calculated. α=0.25 is a scaling factor to account for windowing. Negative values ​​were replaced by 0.

[0251] The above method was used to suppress noise in data acquired during imaging in the presence of an Aurora™ electromagnetic tracking system (Northern Digital Inc), which acts as a noise source. Figures 13A and 13B show images acquired in the presence of a noise source. Figure 13A shows the image with and without the noise suppression method applied, respectively. A clear improvement in the signal-to-noise ratio (about 5 dB) was observed when the noise suppression method was implemented.

[0252] It should be understood that the specific embodiments described above are illustrative and that these embodiments are susceptible to various modifications and alternative forms. It should be further understood that the claims are not intended to be limited to the particular forms, but rather to cover all modifications, equivalents, and alternative forms falling within the spirit and scope of the present disclosure.

Claims

1. 1. A method for removing noise from a detected imaging signal in the presence of broadband noise, comprising: obtaining a noise characterization waveform by detecting energy waves with an imaging transducer receive circuit when no imaging energy is received, and processing the noise characterization waveform to generate an in-band noise characterization waveform that resides within an imaging band and an out-of-band noise characterization waveform that resides at least in part within a noise detection band outside the imaging band; processing the in-band noise characterization waveform and the out-of-band noise characterization waveform to determine a relationship between noise in the imaging band and noise in the noise detection band; obtaining one or more imaging waveforms by detecting imaging signals with the imaging transducer receive circuitry; For at least one of the imaging waveforms: processing the imaging waveform to generate an in-band imaging waveform that resides within the imaging band and an out-of-band noise-detection imaging waveform that resides within the noise-detection band; applying a noise removal correction to at least a portion of the in-band imaging waveform using the relationship and the out-of-band noise detected imaging waveform; The method according to claim 1,

2. The method of claim 1 , wherein the denoising correction is determined and applied on a time window basis for at least two time windows of the in-band imaging waveform.

3. The step of detecting an imaging signal includes detecting an imaging signal with the imaging transducer receive circuitry along a plurality of scan lines, the method further comprising: obtaining a plurality of imaging waveforms respectively associated with the plurality of scan lines; generating an image based on the denoised in-band imaging waveforms respectively associated with the plurality of scan lines; having The method of claim 1.

4. The method of claim 1 , wherein the denoising correction is applied within a given time window when a measurement related to an amount of noise in the in-band imaging waveform within the given time window is greater than a threshold.

5. The method of claim 4 , wherein the measure related to the amount of noise in the in-band imaging waveform is based on the relationship and an amount of power in the in-band noise characterization waveform.

6. 5. The method of claim 4, wherein, prior to applying the denoising correction, each time window having the measurements above the threshold is re-evaluated against spatially adjacent time windows corresponding to spatially adjacent scan lines, whereby a sample originally estimated to be associated with the presence of noise is replaced with a sample from a spatially adjacent time window if the adjacent time window is estimated to be associated with the absence of noise.

7. 5. The method of claim 4, wherein, prior to applying the noise removal correction, each time window having the measurement value exceeding the threshold is re-evaluated relative to adjacent time windows in time, whereby a time window initially estimated to be associated with the absence of noise is re-evaluated as associated with the presence of noise if the adjacent time window is estimated to be associated with the presence of noise.

8. The method of claim 4 , wherein the threshold is determined based on a baseline measurement performed in the absence of imaging energy and noise.

9. The method of claim 1 , wherein the relationship between the noise in the imaging band and the noise in the noise detection band is related to a measure of relative energy in the in-band noise characterization waveform and the out-of-band noise characterization waveform.

10. The method of claim 1 , wherein the denoising correction comprises a subtraction value.

11. 11. The method of claim 10, wherein the denoising correction is determined and applied for each time window, and the subtraction value associated with a given time window increases with the amount of energy detected in the out-of-band noise detected imaging waveform in the given time window.

12. The method of claim 1 , wherein the denoising correction comprises multiplication by a damping factor.

13. 13. The method of claim 12, wherein the denoising correction is determined and applied on a time window basis, and the attenuation coefficient associated with a given time window decreases with an amount of noise in the in-band imaging waveform for the given time window, thereby attenuating portions of the in-band imaging waveform associated with noise.

14. The method of claim 1 , wherein the denoising correction is determined and applied for each time window, and the denoising correction applied for a given time window depends on a depth of the given time window within the in-band imaging waveform.

15. The method of claim 1 , further comprising repeating noise characterization to re-establish the relationship between the noise in the imaging band and the noise in the noise detection band.

16. The method of claim 15 , wherein the noise characterization is repeated according to user input.

17. The method of claim 15 , wherein the noise characterization is automatically repeated when an absence of imaging energy is detected.

18. The method of claim 1 , further comprising detecting a change in the noise within the noise detection band by monitoring one or more parameters associated with the out-of-band noise detection imaging waveform.

19. 20. The method of claim 18, further comprising generating an alert or message when a change in at least one parameter is detected.

20. The method further comprises: performing noise suppression on the in-band imaging waveform while imaging energy is absent; generating an error value based on the energy of the in-band imaging waveform after noise correction; generating an alert to repeat the noise characterization if the error value exceeds a preselected threshold; having The method of claim 1.

21. The method of claim 1 , wherein the imaging transducer receive circuitry includes an ultrasound transducer.

22. The method of claim 1 , wherein the imaging transducer receive circuitry comprises a coil for detecting a magnetic field.

Citation Information

Patent Citations

  • Measuring system for CN ratio

    JP1986274495A

  • Ultrasonic diagnostic apparatus

    JP1987270138A

  • Ultrasonic diagnosis device

    JP1990134145A

  • Phasing circuit for received wave and ultrasonic image pick-up apparatus by using it

    JP1990261436A

  • Device and method for suppressing noise

    JP1991266899A