Ultrasound data processor

Abstract

An ultrasound data processor for assessing the reliability of ultrasound data to determine blood flow values in a patient. The processor is adapted to apply at least a spectral analysis procedure and a wavelet decomposition procedure to input ultrasound data (or data or signals derived therefrom), and to feed the output(s) of the procedures to a classifier algorithm configured to generate a reliability indicator based on these input information. Both spectral analysis and wavelet decomposition are signal analysis techniques that assess the periodicity characteristics of data and signals, which are particularly suitable for detecting background noise artifacts that do not alter the morphology or shape of the signal but do distort the background, such as background noise artifacts produced by an electrosurgical knife used in a surgical procedure.

Description

TECHNICAL FIELD

[0001] The present invention relates to an ultrasound data processor, in particular for analyzing the reliability of acquired ultrasound data in order to derive hemodynamic measurements. BACKGROUND

[0002] Ultrasound can be used to sense and monitor various hemodynamic parameters of a patient, such as blood flow. For example, various blood flow parameters can be sensed by acquiring pulsed wave Doppler (PWD) ultrasound data and / or B-mode echo data. Ultrasound-derived waveforms can be rendered on a display of a patient monitor. In some applications, the measurements can be acquired continuously as part of continuous monitoring.

[0003] One important parameter that can be measured is the change or trend of the blood flow value over time. In general, observable changes in the blood flow value (or any other hemodynamic parameter) can occur due to clinical factors, such as a change in the patient's hemodynamic condition or an irregular heartbeat of the patient, or due to artifacts in the ultrasound data leading to inaccurate hemodynamic parameter calculations.

[0004] Ultrasound signals are known to be susceptible to artifacts, such as artifacts created by external environment, motion, and beam steering, which reduce the reliability of the derived waveforms and parameters. The presence of external artifacts affects the accuracy of the flow measurements and, thus, the reliability of trends in the flow measurements.

[0005] As an example, in the case of perioperative care, this care encompasses all stages of a surgical procedure. It usually takes place in a surgical center attached to a hospital. One of the most common and unavoidable external artifacts in ultrasound-based measurements in a perioperative environment is an external artifact introduced by an electrosurgical knife (or electric power knife) used by a surgeon during a surgical operation. An example of an electrosurgical knife is an electrosurgical scalpel. Surgical electrosurgical knives use electric current or radiation to cut, coagulate, desiccate, or cauterize tissue. They are frequently used during surgical operations because they can help prevent blood loss when cutting.

[0006] It would be desirable to have means for reliably distinguishing between changes in ultrasound-based hemodynamic parameter measurements caused by clinically relevant factors, e.g. hemodynamic condition, irregular heartbeat, and changes in ultrasound-based hemodynamic parameter measurements caused by signal artifacts, e.g. electrosurgical knife noise in the ultrasound data. SUMMARY

[0007] According to examples in accordance with an aspect of the present invention, there is provided an ultrasound data processor,

[0008] The processor comprises an input / output for receiving as input ultrasound data, the ultrasound data comprising at least pulsed wave Doppler (PWD) ultrasound data,

[0009] The processor is configured to:

[0010] apply a first analysis procedure to the ultrasound data or a derivative thereof to generate a first set of one or more analysis parameters, the first analysis procedure comprising a Fourier spectral analysis of the data;

[0011] apply a second analysis procedure to the ultrasound data or a derivative thereof to generate a second set of one or more analysis parameters, the second analysis procedure comprising a wavelet decomposition,

[0012] input the first and second sets of analysis parameters into a classifier, wherein the classifier is configured to generate, based on the input analysis parameters, at least one reliability classification for the input ultrasound data, the at least one reliability classification being indicative of a reliability of at least part of the input data for using the data to determine a patient blood flow measurement.

[0013] As mentioned above, it is an object of embodiments of the present application to enable to determine background noise artefacts in the data, such as background noise artefacts caused by electrical knife interference. These reduce the quality of the ultrasound signal, but do not necessarily change the morphology or overall shape of the signal. Embodiments of the present application propose to use spectral analysis, e.g. Fourier spectral analysis, and wavelet analysis, both of which allow to analyse periodic properties of the signal. These techniques are best suited to detect background distortions in the signal, rather than artefacts that cause morphological features of the waveform, and thus enable to identify relevant artefacts.

[0014] In particular, both Fourier analysis and wavelet analysis techniques analyse periodic patterns in the signal or periodic properties of the data. Background artefacts of interest tend to not change the shape of the acquired signal. However, the inventors have realised that based on an evaluation of changes in the periodic properties or patterns of the signal, background artefacts are detectable.

[0015] The first and second analysis procedures can be applied to the input ultrasound data itself, or to data or signals derived from the input ultrasound data. For example, the input PWD data can be a PWD signal, and from this a PWD spectrogram can be determined. This can be provided as input to the first and second analysis procedures. In a further example, an envelope of the PWD spectrogram can be extracted, which for example represents a time series of the maximum Doppler frequency or velocity detected at each time point, and this envelope signal is provided as input to the first and second analysis procedures. These represent merely non-limiting examples.

[0016] The classifier can be a classifier algorithm. The classifier can be a machine learning algorithm, such as a decision tree classifier.

[0017] The reliability classification is determined based on outputs of the first analysis process and the second analysis process. Both of these analysis processes analyse periodic patterns in the ultrasound data or data derived from the ultrasound data. Based on this, the classifier is configured (e.g. trained) to derive a classification about the reliability of the ultrasound data for calculating blood flow measurements. It can thus give an indication about whether observed changes in blood flow values calculated using the ultrasound data are due to clinically relevant factors or clinically irrelevant factors. In some examples, the classification can be quantitative (e.g. a reliability score), or can be qualitative (e.g. ‘reliable’ / ‘unreliable’).

[0018] The processor can be configured to generate a data output at the input / output indicating the reliability classification. In some examples, this can be communicated or transmitted to an external device.

[0019] According to an advantageous embodiment, the processor can be configured to receive and process the input ultrasound data in real time. In other words, the reliability classification(s) can be generated in real time with ultrasound data acquisition. This allows the results to be presented to a user in real time, enabling flow measurements to be interpreted in the context of their determined reliability.

[0020] According to one or more embodiments, the processor can be further configured to process the input PWD ultrasound data to derive one or more blood flow measurements. Thus, in this embodiment, the same processor that is used to assess the reliability of the data is used to determine blood hemodynamic parameters. This reduces components, and can also in some embodiments allow the two results to be more easily synchronised or otherwise linked or associated with each other. In some examples, the blood flow measurements can be derived continuously or cyclically.

[0021] According to one or more embodiments, the processor can be configured to determine a reliability of each of the one or more determined blood flow measurements based on the reliability classification for the ultrasound data (based on which the flow measurement(s) are calculated). For example, if the reliability classification of the ultrasound data is low, it can be determined that the reliability of the flow measurements is low, and vice versa.

[0022] According to one or more embodiments, the input ultrasound data can further comprise ultrasound echo data, such as B-mode ultrasound data, and wherein the processor can be configured to apply the first analysis process and the second analysis process to the ultrasound echo data to also determine a reliability classification for the ultrasound echo data

[0023] The ultrasound echo data can then be used to generate ultrasound image data, for example for display on a patient monitor together with flow measurements derived using the PWD data.

[0024] The processor can be configured to compute an overall classification for the ultrasound data based on a combination of the reliability classifications for the PWD data and the echo data.

[0025] According to one or more embodiments, the input PWD ultrasound data can represent a PWD signal as a function of time, and wherein the processor can be configured to decompose the signal into a plurality of time portions and generate a respective reliability classification for each time portion.

[0026] According to one or more embodiments, the processor can be further configured to process each of the time portions of the PWD ultrasound signal to derive a respective blood flow measurement based on each time portion.

[0027] The processor can be configured to determine a reliability of each of the respective blood flow measurements based on the reliability classification determined for the respective time portion.

[0028] According to one or more embodiments, the processor can be further configured to apply a data augmentation procedure to ultrasound data for which the classifier determines a low reliability classification. The low reliability classification can be predefined, for example, a low classification can correspond to a predefined range of quantitative reliability scores, or a predefined set of one or more qualitative reliability classifications.

[0029] Optionally, the data augmentation procedure can comprise applying one or more filters.

[0030] According to one or more embodiments, the processor can be adapted to derive a PWD spectrogram based on the input PWD data, and wherein the first analysis procedure and the second analysis procedure are applied at least to the PWD spectrogram data. In some examples, the first analysis procedure can comprise deriving a frequency domain representation of the PWD spectrogram.

[0031] Additionally or alternatively, according to one or more embodiments, the processor can be adapted to process the input PWD data to derive a PWD spectrogram based on the input PWD data, and further extract a PWD envelope signal from the spectrogram, the envelope signal representing a time series of the maximum Doppler frequency or velocity detected at each time point, and wherein the first analysis procedure and the second analysis procedure are applied at least to the representation of the PWD envelope signal. In some examples, the first analysis procedure can comprise deriving a frequency domain representation of the envelope signal.

[0032] According to one or more embodiments, the first analysis procedure can comprise determining a ratio of the spectral energy of a peak portion of a frequency domain representation of the ultrasound data or data derived from the ultrasound data to the spectral energy of a central portion of the frequency domain representation. As an example, a PWD spectrogram or a frequency domain representation of an envelope of a PWD spectrogram can be used.

[0033] According to one or more embodiments, the reliability classification generated by the classifier can comprise a reliability score for the ultrasound data.

[0034] Examples in accordance with another aspect of the application provide a system. The system comprises a patient monitor comprising a display unit and an input / output. The system further comprises an ultrasound data processor in accordance with any of the examples or embodiments outlined above or described below.

[0035] The patient monitor is configured to receive, from the data processor, the at least one reliability classification generated for the input ultrasound data. It can be further configured to receive input indicative of one or more blood flow measurements computed based on the same input ultrasound data. It can be configured to generate, on the display device, a display output representing the blood flow measurements and further representing the reliability classification for the data.

[0036] The patient monitor can receive respective blood flow measurements computed by the data processor for each of a plurality of time portions of the input ultrasound signal (as discussed above).

[0037] The patient monitor can be configured to display a trend of the blood flow measurements over time.

[0038] The patient monitor can be further configured to display a visual indicator indicative of the respective reliability classification determined by the data processor for each blood flow measurement (as discussed above).

[0039] Examples in accordance with another aspect of the application provide another system.

[0040] The further system comprises an ultrasound data processor according to any of the examples or embodiments outlined above or described below.

[0041] The further system comprises a wearable patch comprising an integrated ultrasound sensing device arranged to acquire ultrasound data including PWD ultrasound data.

[0042] The input / output of the ultrasound data processor is communicatively coupled to the wearable patch to receive the acquired ultrasound data from the ultrasound sensing device.

[0043] Examples in accordance with a further aspect of the application provide an ultrasound data processing method.

[0044] The method comprises receiving: input ultrasound data, the ultrasound data comprising at least pulsed wave Doppler (PWD) ultrasound data;

[0045] applying a first analysis procedure to the input ultrasound data or data derived from the ultrasound data to generate a first set of one or more analysis parameters, the first analysis procedure comprising a Fourier spectral analysis of the data;

[0046] applying a second analysis procedure to the input ultrasound data or data derived from the ultrasound data to generate a second set of one or more analysis parameters, the second analysis procedure comprising a wavelet decomposition; and

[0047] inputting the first and second sets of analysis parameters to a classifier, wherein the classifier can be configured to determine at least one reliability classification for the input ultrasound data based on the input analysis parameters, the at least one reliability classification being indicative of a reliability of at least a portion of the input ultrasound data for determining a patient blood flow measurement.

[0048] Examples in accordance with a further aspect of the application provide a computer program product comprising computer program code executable on a processor or computer, wherein the code is configured to cause the processor to perform a method according to any of the examples or embodiments outlined above or described below.

[0049] These and other aspects of the application will be apparent from and elucidated with reference to the embodiments described hereinafter. BRIEF DESCRIPTION OF DRAWINGS

[0050] For a better understanding of the application, and to show how it can be implemented in practice, reference will now be made, purely by way of example, to the accompanying drawings, in which:

[0051] Figures 1-4A set of examples of PWD ultrasound data and resulting blood flow estimates for different clinical scenarios are shown;

[0052] Figure 5 An example processor according to one or more embodiments is illustrated;

[0053] Figure 6 Processing steps performed by the example processor according to one or more embodiments are outlined schematically;

[0054] Figure 7 Stages of an example spectral analysis procedure applied to ultrasound data are illustrated;

[0055] Figure 8 Stages of an example wavelet analysis procedure applied to ultrasound data are illustrated;

[0056] Figure 9 Stages of another example spectral analysis procedure applied to ultrasound data are illustrated;

[0057] Figure 10 Stages of another example wavelet analysis procedure applied to ultrasound data are illustrated;

[0058] Figure 11 Processing steps of an example processor according to one or more embodiments are outlined;

[0059] Figure 12 Processing steps of another example processor according to one or more embodiments are outlined;

[0060] Figure 13 A pre-processing program for application to ultrasound data according to one or more embodiments is illustrated schematically;

[0061] Figure 14 An example system according to one or more embodiments is depicted schematically; and

[0062] Figure 15 Another example system according to one or more embodiments is depicted schematically;

[0063] Figure 16 A block diagram of an ultrasound data processing method is outlined. DETAILED DESCRIPTION

[0064] The application will be described with reference to the accompanying drawings.

[0065] It should be understood that the detailed description and specific examples, while indicating exemplary embodiments of apparatuses, systems and methods, are intended for purposes of illustration only and are not intended to limit the scope of the present application. These and other features, aspects, and advantages of the apparatuses, systems and methods of the present application will become better understood from the following description, appended claims, and accompanying drawings. It should be understood that the drawings are only schematic and are not drawn to scale. It should also be understood that the same reference numerals are used throughout the drawings for like or similar items.

[0066] The present application provides an ultrasound data processor for assessing the reliability of ultrasound data to determine blood flow measurements of a patient. The processor is adapted to apply at least a spectral analysis procedure and a wavelet decomposition procedure to input ultrasound data (or data or signals derived from the ultrasound data), and to feed the output(s) of the procedures to a classifier algorithm configured to generate a reliability indicator based on this input information. Both spectral analysis and wavelet decomposition are signal analysis procedures that assess the periodicity characteristics of data and signals, which are particularly suitable for detecting background data artifacts such as background noise artifacts generated by an electrotome used in a surgical operation, which do not alter the morphology or shape of the data signal but do distort the background.

[0067] To illustrate the problem the embodiments are intended to solve, Figures 1-4 A set of ultrasound signal traces and corresponding clinical parameter traces is shown for a set of different clinical scenarios. For each of Figures 1-4 For each of the, plot (a) shows a spectrogram of an acquired pulse wave Doppler (PWD) ultrasound data set for a patient. Plot (b) shows an envelope of the PWD spectrogram signal trace (a). The envelope tracks the maximum detected blood flow Doppler frequency (or velocity) at each point in time. It is thus a time series of maximum frequencies (or velocities) at each point in time. The envelope can thus be understood as a "1D time series waveform of maximum velocities versus time".

[0068] Plot (c) shows a derived heart rate estimate as a function of time, which is based on the PWD derivation of plot (b). The plot shows the estimated beat-to-beat blood flow through the carotid artery (y-axis; units: ml / min) as a function of time (x-axis; units: seconds), which is based on the PWD signal trace of plot (b).

[0069] Figure 1 A PWD data as well as heart rate and blood flow estimates are shown for a normal (steady) clinical scenario, where there is no significant background noise in the data and the patient is in a clinically normal state.

[0070] Figure 2PWD data and heart rate and blood flow estimates are shown for a clinical scenario in which a patient is experiencing irregular heartbeats. In this scenario, there is no significant background noise in the ultrasound data (trace (a)). Irregular heartbeats are detectable in the PWD signal trace extracted from the PWD spectrogram (trace (a)) and are shown in the heart rate trace (c). The irregular heartbeats cause detectable irregularities in the blood flow through the carotid artery (trace (d)).

[0071] Figure 3 and 4 A scenario is shown in which a patient is in a clinically normal hematological state (e.g. no irregular heartbeats) but in which there are background artefacts in the PWD ultrasound data (figure (a)) due to interference from a surgical electrosurgical knife.

[0072] With regard to Figure 3 , the background noise is clearly visible in the PWD ultrasound data (graph (a)) as a series of horizontal bars superimposed on the PWD data. The effect on the extracted PWD signal trace can be seen in the signal trace (b). This causes erroneous estimates of the heart rate values in the parts of the PWD signal trace affected by the background noise artefacts as shown in the signal trace (c). This therefore causes an erroneous estimate of the patient’s blood flow changes (signal trace (d)) which falsely indicates that the patient is experiencing significant irregularities in the blood flow (through the carotid artery). A clinician observing the data trends on a patient monitor can therefore falsely conclude that the patient is in a hematological abnormal state, leading to potentially unnecessary clinical intervention.

[0073] A similar effect can be seen in Figure 4 , where Figure 4 background noise artefacts in the PWD ultrasound data are also shown (graph (a)). This causes distortion in the PWD signal trace extracted from the PWD data (trace (b)) which therefore causes erroneous estimates of the heart rate in the signal regions affected by the noise (trace (c)) and erroneous fluctuations in the blood flow estimates (trace (d)).

[0074] Figure 5 Components of an example ultrasound data processor according to one or more embodiments of the application are schematically illustrated.

[0075] The processor 22 comprises an input / output (I / O) 26 for receiving ultrasound data 24 as input, the ultrasound data comprising at least pulsed wave Doppler (PWD) ultrasound data. The ultrasound data can be received in real-time from an ultrasound transducer unit, for example, or in other examples can be received from a data store.

[0076] In Figure 5In the example of FIG. 1, the data processor 22 further comprises an integrated circuit (IC) 28 adapted to perform processing of the ultrasound data received at the I / O 26. The I / O is communicatively coupled with the IC 28.

[0077] Figure 6 The processing of the input ultrasound data 24 applied by the IC 28 of the processor 22 is schematically illustrated.

[0078] The IC 28 of the processor 22 is configured to apply a first analysis procedure 34 to the ultrasound data 24 or data derived therefrom to generate a first set of one or more analysis parameters. The first analysis procedure comprises a spectral analysis of the data (e.g. a Fourier spectral analysis). The first analysis procedure can comprise deriving a frequency domain representation of the input ultrasound data or data derived therefrom. The output of the first analysis procedure forms the first set of one or more analysis parameters. This will be explained by way of further examples below.

[0079] The IC 28 of the processor 22 is further configured to apply a second analysis procedure 36 to the ultrasound data or data derived therefrom to generate a second set of one or more analysis parameters. The second analysis procedure comprises a wavelet analysis applied to the ultrasound data or data derived therefrom. The output of the wavelet decomposition can provide the second set of one or more analysis parameters. This will be explained by way of further examples below.

[0080] The IC is further configured to input the first set of analysis parameters and the second set of analysis parameters to a classifier 38, wherein the classifier is configured to determine, based on the input analysis parameters, at least one reliability classification 30 for the input ultrasound data, which is indicative of a reliability of at least a portion of the input ultrasound data 24 for using the data to determine a patient blood flow measurement.

[0081] The classifier 38 can be a classifier algorithm. It can be stored locally in the processor 22, e.g. in a memory comprised by the processor, or it can be stored externally to the processor and retrieved or accessed by the processor. The classifier can be a machine learning algorithm, e.g. a decision tree classifier.

[0082] The reliability classification is determined based on the output of the first analysis procedure 34 and the second analysis procedure 36. Both of these analysis procedures analyse periodic patterns in the ultrasound data or data derived therefrom. Based on this, the classifier 38 is configured or trained to generate a classification 30 about the reliability of the ultrasound data for computing blood flow measurements. It can thus give an indication about whether observed changes in blood flow values computed using the ultrasound data are due to clinically relevant factors or clinically irrelevant factors. In some examples, the classification can be quantitative (e.g. a reliability score), or it can be qualitative (e.g. reliable / unreliable).

[0083] As mentioned above, the classifier can be a machine learning algorithm. A machine learning algorithm is any self-training algorithm that processes input data in order to produce or predict output data. Here, the input data comprises the first set of analysis parameters and the second set of analysis parameters, and the output data comprises the reliability classification.

[0084] Suitable machine learning algorithms for use in the present application will be apparent to the skilled person. Examples of suitable machine learning algorithms include decision tree algorithms and artificial neural networks. Other machine learning algorithms such as logistic regression, support vector machines or naive Bayes models are suitable alternatives.

[0085] The structure of an artificial neural network (or simply neural network) is inspired by the human brain. A neural network is composed of layers, each layer comprising a plurality of neurons. Each neuron comprises a mathematical operation. In particular, each neuron can comprise a different weighted combination of a single type of transformation (e.g. the same type of transformation, sigmoid, etc. but with different weights). In the process of processing input data, the mathematical operation of each neuron is performed on the input data to produce a numerical output, and the output of each layer in the neural network is sequentially fed to the next layer. The last layer provides the output.

[0086] Methods of training machine learning algorithms are well known. Typically, such a method comprises obtaining a training data set comprising training input data entries and corresponding training output data entries. An initialised machine learning algorithm is applied to each input data entry to generate a predicted output data entry. An error between the predicted output data entry and the corresponding training output data entry is used to modify the machine learning algorithm. This process can be repeated until the error converges, and the predicted output data entries are sufficiently similar (e.g. ±1%) to the training output data entries. This is commonly referred to as a supervised learning technique.

[0087] For example, in the case where the machine learning algorithm is formed by a neural network, the weights of the mathematical operation of each neuron can be modified until the error converges. Known methods of modifying a neural network include gradient descent, backpropagation algorithms, etc.

[0088] The training input data entries correspond to exemplary first and second analysis parameters. The training output data entries correspond to the reliability classification. The reliability classification of the training data can be determined manually by an expert, for example, or can be determined empirically, for example, based on a comparison between resulting blood flow values derived from input ultrasound data determining the analysis parameters and true blood flow values measured, for example, using a separate blood flow measurement device.

[0089] The processor 22 can be configured to generate a data output at the input / output 26 indicative of the reliability classification. In some examples, this can be communicated or transmitted to an external device.

[0090] The input / output 26 provides a communication interface, for example, with an external device, such as a patient monitor, or a user interface of the patient monitor.

[0091] As mentioned above, the first analysis procedure 36 comprises a spectral analysis or decomposition of the ultrasound data. This can comprise applying a Fourier transform to the ultrasound data or data derived or extracted therefrom to obtain a frequency domain representation of the ultrasound data or data derived therefrom. The analysis parameters can correspond to features or properties of this frequency domain representation, for example, statistical properties, such as a spread of the spectral energy. This will be explained further in more detail below.

[0092] As mentioned above, the second analysis procedure 34 applied by the processor 22 comprises a wavelet analysis. Wavelet decomposition is a well-known signal analysis technique, and the person skilled in the art will be aware of various means for implementing a wavelet analysis.

[0093] In the context of embodiments of the present application, the wavelet decomposition is to derive wavelet coefficients, the absolute and relative values of which can provide parameters that can be used to distinguish between reliable and unreliable data. In some instances, statistical parameters related to the wavelet coefficients can be derived, which can be used by the classifier 38 to determine the reliability classification.

[0094] According to a first set of example embodiments, the processor 22 can be adapted to process the input PWD data 24 to derive a PWD spectrogram based on the input PWD data, and to further extract a PWD envelope signal from the spectrogram, the envelope signal being an envelope of the PWD spectrogram. The first analysis procedure 34 and the second analysis procedure 36 can then be applied at least to a representation of the PWD envelope signal.

[0095] The PWD spectrogram means a time-frequency representation of a pulse wave Doppler signal obtained by an ultrasound transducer device at a particular location on a body location. For example, the spectrogram is derived from the raw input PWD ultrasound data based on applying a short-time Fourier transform (STFT) to the input data. The spectrogram represents the distribution of Doppler frequencies included in the raw PWD signal data over time. It represents the amplitude or magnitude of each Doppler frequency at each time point. The spectrogram is typically represented in a 2D graphical form, wherein time is represented on the x-axis, frequency is represented on the y-axis, and wherein the intensity, color or shade value of each pixel represents the magnitude of the corresponding frequency component at the corresponding time point. Of course, the spectrogram can be represented in other ways, as will be known to the person skilled in the art.

[0096] The PWD envelope signal is an envelope of the PWD spectrogram, corresponding to the upper trace of the PWD signal spectrogram, as for example in Figures 1-4In each case, signal trace (a) shows the acquired PWD signal, and signal trace (b) shows the extracted PWD envelope signal.

[0097] More specifically, the envelope signal tracks the maximum detected blood flow Doppler frequency (or velocity) at each point in time. It is thus a time series of maximum frequency (or velocity) values at each point in time. The envelope can thus be understood as a "1D time series waveform of maximum velocity versus time". To extract the envelope signal, an edge detection algorithm (e.g. canny edge detection) can be used according to one or more examples. This allows to track the upper edge of the PWD spectrogram, and this trace is used for the envelope signal.

[0098] One example embodiment following the above described method will now be outlined in more detail.

[0099] Figure 7 The processing steps of an exemplary first analysis procedure applied to the envelope of the PWD spectrogram are illustrated. Figure 8 The steps of an exemplary second analysis procedure applied to the PWD envelope signal are illustrated.

[0100] With respect to Figure 7 This shows for each of three different clinical scenarios a set of phases of data processing applied to the ultrasound data in the exemplary first analysis procedure. The plots of row (a) and row (b) correspond to scenarios in which there is no significant background noise in the signal, and the PWD data reliably represents the clinical situation of the patient. Row (a) corresponds to a clinically normal patient. Row (b) corresponds to a patient with irregular heartbeats (ectopic heart). The plots of row (c) correspond to scenarios in which there is background noise in the data, e.g. caused by surgical electrocautery interference.

[0101] The plots of column (i) show the PWD envelope signal for each case, corresponding to the extracted envelope of the PWD spectrogram for each case. The y-axis corresponds to blood flow velocity in cm / s. The x-axis corresponds to time in seconds.

[0102] The PWD spectrogram for each case can be derived from the (raw) input PWD ultrasound data, e.g. by the processor 22. Alternatively, the input ultrasound data to the processor can take the form of the PWD spectrogram. The plots of column (ii) show the frequency domain representation of each PWD envelope signal of column (i). This can be derived, e.g. by applying a Fourier transform (e.g. a Fast Fourier Transform (FFT)) to the envelope signal. The y-axis of each plot of column (ii) corresponds to the spectral magnitude. The x-axis corresponds to the frequency. The frequency x-axis is centered around zero, and the plot includes both positive and negative frequency components (corresponding to positive and negative blood flow velocity components).

[0103] The plot of column (iii) illustrates a selected range of each of the frequency domain representations of column (ii), which is indicated in Figure 7 each case. This corresponds to a central range or portion of the frequency domain representation in each case. For the purpose of illustration, this is presented in Figure 7 the plots in column (iii) and can not be explicitly extracted by the processor during the analysis procedure.

[0104] As can be seen in the plots of column (iii), the spectral energy distribution in the selected region differs for the different scenarios. A common distinction between the spectral distribution of the low background noise data (rows (a) and (b)) and the spectral distribution of the high background noise data (row (c)) is that for the higher noise data there is a concentration of spectral energy near the center of the spectral distribution, i.e. at the low frequency components. This arises from the fact that background noise artifacts, such as those generated by electrocautery interference, tend to be present in the PWD data only at a selected range of spectral frequencies, which are typically low, whereas the clinically relevant portion of the PWD data tends to have a more uniformly distributed set of frequency components within the PWD data. This feature can be used to identify ultrasound data that can contain background noise artifacts and thus can not be as reliable for the determination of hemodynamic parameter estimates.

[0105] According to one or more embodiments, therefore, the first analysis procedure can comprise determining a ratio of the spectral energy of a peak portion of the frequency domain representation of the PWD ultrasound data or PWD envelope signal to the spectral energy of a central portion of the frequency domain representation.

[0106] According to one example, for instance, the ratio F pc may take the following form:

[0107]

[0108] where Xk is the frequency domain representation of the envelope of the PWD ultrasound spectrogram, P is the frequency at which the maximum spectral amplitude occurs, N = N fft / 2, where N fft is the number of frequency points or bins used to compute the frequency domain representation, and L is the number of adjacent data points (around the maximum point P) n is used to compute the spectral energy at the maximum point P. For instance, L defines a selected "half-width" (in terms of the number of frequency bins) of the central portion of the frequency distribution and the peak portion of the frequency distribution. L can be selected as desired in order to capture the appropriate portion of the distribution around the peak frequency point P and the central frequency point N. For instance, N can correspond to the Nyquist frequency. In some examples, L can be selected to be less than a certain fraction of N, for instance, less than ¼ of N, less than 1 / 6 of N, less than 1 / 8 of N, or less than 1 / 10 of N.

[0109] Figure 8 The data processing stage of the second analysis procedure is illustrated, including a wavelet analysis applied to the PWD envelope signal derived from the PWD spectrogram.

[0110] Again, different rows illustrate different clinical scenarios. Rows (a) and (b) correspond to data without significant background noise, and rows (c) and (d) correspond to scenarios where the data does contain background noise artefacts, e.g. caused by interference from an electrosurgical knife. More specifically, row (a) corresponds to a clinically normal patient. Row (b) corresponds to a patient with an irregular heartbeat (a holter heart). Rows (c) and (d) correspond to different scenarios where the data contains background noise artefacts.

[0111] The plots of column (i) illustrate the PWD envelope signal for each case, corresponding to the extracted envelope of the PWD spectrogram for each case (as discussed in more detail above).

[0112] In this example, the second analysis procedure includes applying an approximation (low-pass) and detail (high-pass) wavelet filter to the PWD envelope signal to derive approximation CA i and detail CD i wavelet coefficients, where i indexes the number of decomposition levels of the coefficients. The wavelet decomposition can be applied with any desired number N of decomposition levels. The wavelet coefficients can be continuous wavelet coefficients.

[0113] The wavelet coefficients obtained for PWD data including background noise artefacts differ in a reliable manner from the wavelet coefficients derived for PWD data without background noise. These differences can be used to identify ultrasound data that can contain background noise artefacts and that can therefore be unreliable for deriving blood flow measurements.

[0114] As an example, and with reference to Figure 8 Various different analysis parameters can be derived based on the wavelet coefficients derived for the PWD envelope signal. One or more of these can be fed to the classifier 38 or used to derive a reliability classifier for the corresponding ultrasound data.

[0115] One example analysis parameter can be the ratio of the absolute standard deviations of different wavelet coefficients, e.g.:

[0116]

[0117] where,

[0118]

[0119] where, L i is the wavelet coefficient CD ilength (e.g., in the time domain). The wavelet coefficients can be continuous wavelet coefficients plotted as a function of time.

[0120] Another example analysis parameter can be the ratio of power spectral density:

[0121]

[0122] wherein,

[0123]

[0124] wherein, L i is the length of the wavelet coefficient CD i (e.g., along the time axis).

[0125] The plots of column (ii) represent example approximation (i.e., low-pass) wavelet coefficients derived from the wavelet decomposition of the PWD envelope in each case. The plots of column (iii) represent example detail (i.e., high-pass) wavelet coefficients derived from the wavelet decomposition of the PWD envelope in each case. Both the approximation coefficients and the detail coefficients are plotted as continuous waveforms as a function of time. The x-axis in each of the plots of (ii) and (iii) corresponds to time, and the y-axis corresponds to wavelet amplitude.

[0126] Wavelet decomposition procedures are well known in the art, and a person of skill in the art will immediately recognize means for performing the decomposition procedure. Further details regarding example wavelet decomposition procedures can be found, for example, in Daubechies, I. Ten Lectures on Wavelets, CBMS-NSF Regional Conference Series in Applied Mathematics. Philadelphia, PA: SIAM Ed, 1992. Details can also be found in the following paper: Mallat, S. G. “A Theory for Multiresolution Signal Decomposition: The Wavelet Representation,” IEEE Transactions on Pattern Analysis and Machine Intelligence. Vol. 11, Issue 7, July 1989, pp. 674-693.

[0127] The plots of column (iv) show example derived WPSD values for each of five different patients, calculated according to the formula described above. The y-axis corresponds to WPSD values, and the x-axis corresponds to patient number.

[0128] The plot of column (v) illustrates the characteristics of each plot in column (iv) plotted on the same plot. All data points above the line shown on the plot are wavelet features derived from PWD data without background artefacts (rows (a) and (b)), and all data points below the line are features derived from PWD data with background artefacts (rows (c) and (d)). Thus, it can be seen from this how the classifier can use the features derived and illustrated in the plots of column (iv) to distinguish between data containing background noise (and thus being unreliable) and data not containing background noise (and thus being reliable).

[0129] According to one or more embodiments of the second set, the processor can be adapted to derive a PWD spectrogram based on the input PWD data, and wherein the first and second analysis procedures are applied to at least the PWD spectrogram data. At least the first analysis procedure can comprise deriving a frequency domain representation of the PWD spectrogram.

[0130] Thus, in this set of embodiments, the first and second analysis procedures are applied to the PWD spectrogram itself, rather than to the envelope signal extracted therefrom. The PWD spectrogram represents the distribution of Doppler frequencies in the input PWD data as a function of time. It represents the amplitude or magnitude of each Doppler frequency at each point in time. Thus, it comprises data varying in two dimensions (frequency and time). The PWD spectrogram can thus be referred to herein as a 2D PWD spectrogram.

[0131] The spectrogram is typically represented in the form of a 2D plot, in which time is represented on the x-axis, frequency is represented on the y-axis, and in which the intensity, colour or shade value of each pixel represents the magnitude of the relevant frequency component at the relevant point in time. Of course, the spectrogram can be represented in other ways, as will be known to the person skilled in the art.

[0132] One example embodiment following the above-described method will now be outlined in more detail.

[0133] Figure 9 The stages of the first analysis procedure applied to the PWD spectrogram are illustrated. Figure 10 The second analysis procedure applied to the PWD spectrogram is illustrated.

[0134] With regard to Figure 9The plots of column (i) and column (ii) correspond to scenarios in which there is no significant background noise in the signal, and the PWD data reliably represents the clinical situation of the patient. Column (i) corresponds to a clinically normal patient. Column (ii) corresponds to a patient with an irregular heartbeat (a holter heart). The plots of column (iii) correspond to scenarios in which there is background noise in the data, e.g. caused by surgical electrotome interference.

[0135] The plots of row (i) show the PWD spectrograms for each case. The PWD spectrograms for each case can be derived, e.g. by the processor 22, from the (raw) input PWD ultrasound data. Alternatively, the input ultrasound data to the processor can take the form of PWD spectrograms.

[0136] The plots of row (ii) show the frequency domain representation of each PWD spectrogram of row (i). As mentioned above, since the spectrograms cover both frequency and time dimensions, the frequency domain representation also extends over these two dimensions. The x-axis corresponds to frequency, the y-axis corresponds to time, and the z-axis corresponds to the amplitude or spectral power. The frequency domain representation can be derived, e.g. by applying a Fourier transform, e.g. a Fast Fourier Transform (FFT), to the PWD spectrogram.

[0137] The plots of row (iii) show the average of the spectrum over all time frames of the 2D PWD spectrogram of row (ii). The x-axis corresponds to frequency, and the y-axis corresponds to the frequency amplitude.

[0138] Figure 9 The circled areas in the plots of (c)(ii) and (c)(iii) indicate spectral components of the data of scenario (c) caused by background noise artifacts in the data, e.g. associated with electrotome interference.

[0139] As can be seen in plot (iii), the spectral energy distribution of the noise-free scenarios (a) and (b) differs from the spectral energy distribution of the noise- containing scenario (c) in that it includes additional spectral energy components (shown in the circled areas) away from the central peak. This arises from the fact that background noise artifacts, such as those generated by electrotome interference, tend to be present in the PWD data only at a selected range of spectral frequencies, whereas the clinically relevant part of the PWD data tends to have a more uniformly distributed set of frequency components within the PWD data. This feature can be used to identify ultrasound data that can contain background noise artifacts, and thus can not be as reliable for the determination of hemodynamic parameter estimates.

[0140] According to the present example, the first analysis procedure can comprise deriving a frequency domain representation of the 2D PWD spectrogram data (see Figure 9, and also by averaging all time frames of the representation (see Figure 9 , line (iii)). The first analysis procedure can also include determining a ratio of spectral energy of a central portion of the time-averaged frequency-domain representation of the PWD spectrogram to spectral energy of a peak portion.

[0141] As an example, the ratio can take the following form:

[0142]

[0143] where Xka is the average of the spectra of all 2D PWD frequency-domain representation frames, P is the frequency where the maximum spectral amplitude occurs in Xka, N = N fft / 2, where N fft is the number of frequency bins used to compute the frequency-domain representation, and L is the number of adjacent data points, n, used to compute the spectral energy. For example, L defines a selected "half-width" (in terms of the number of frequency bins) of a central portion of the frequency distribution and a peak portion of the frequency distribution. L can be selected as desired in order to capture an appropriate portion of the distribution around the peak frequency point P and the central frequency point N. For example, N can correspond to the Nyquist frequency. In some examples, L can be selected to be less than a particular proportion of N, less than ¼ of N, less than ½ of N, less than ¾ of N, or less than 10% of N.

[0144] As an example, the result of this ratio can provide an analysis parameter that is subsequently provided to the classifier algorithm 38 for use in deriving a reliability classification.

[0145] Figure 10 Figures illustrate applying an example second analysis procedure, including wavelet decomposition, to a (2D) PWD spectrogram.

[0146] Again, different rows illustrate different clinical scenarios. Rows (a) and (b) correspond to data without significant background noise, and rows (c) and (d) correspond to scenarios where the data does include background noise artifacts, e.g., caused by electrosurgical interference. More specifically, row (a) corresponds to a clinically normal patient. Row (b) corresponds to a patient with an irregular heartbeat (a holter heart). Rows (c) and (d) correspond to different scenarios where the data includes background noise artifacts.

[0147] The plots of column (i) illustrate the PWD spectrogram for each case (as discussed in more detail above).

[0148] In this example, the second analysis procedure includes applying approximation (low-pass), horizontal detail (high-pass), vertical detail, and diagonal detail wavelet filters to the PWD spectrogram to respectively derive an approximation CA iand horizontal details CH i vertical details CV i and diagonal details CD i wavelet coefficients, where i indexes the decomposition level of the coefficients. This is a 2D wavelet decomposition process. The wavelet decomposition can be applied with any desired number N of decomposition levels. The wavelet decomposition procedure applied in this case is described in more detail for example in the following paper: Mallat, S. G. “A Theory for Multiresolution Signal Decomposition: The Wavelet Representation,” IEEE Transactions on Pattern Analysis and Machine Intelligence. Vol. 11, Issue 7, July 1989, pp. 674-693. It is also described in the following book: Meyer, Y. Wavelets and Operators. Translated by D. H. Salinger. Cambridge, UK: Cambridge University Press, 1995.

[0149] For applying the wavelet decomposition, the PWD spectrum is processed in the form of a 2D image comprising pixels, each pixel having a pixel value representing the amplitude of a given Doppler frequency component at a given time. This approach is well known in the field of signal processing, in particular for example in the field of speech or audio processing, where speech spectrum is often visualized and processed in this way.

[0150] As mentioned above, the wavelet coefficients obtained for PWD data comprising background noise artifacts differ in a reliable way from the wavelet coefficients derived for PWD data without background noise. These differences can be used to identify ultrasound data that can contain background noise artifacts and thus can be unreliable for deriving blood flow measurements.

[0151] As an example, and with reference to Figure 10 Various different analysis parameters can be derived based on the wavelet coefficients derived for the PWD envelope signal. One or more of these can be fed to the classifier 38 or used to derive a reliability classification for the corresponding ultrasound data.

[0152] One example analysis parameter can be a wavelet power ratio WPR, which can be defined as:

[0153]

[0154] where i indexes the wavelet decomposition level.

[0155] Referring to Figure 10 The plots in column (ii) represent, in each case, example approximation (i.e. low pass) wavelet coefficients derived from a wavelet decomposition of the PWD envelope. The plots in column (iii) represent, in each case, example vertical detail (i.e. high pass) wavelet coefficients derived from a wavelet decomposition of the PWD envelope. The plots in column (iv) represent example horizontal detail coefficients. The wavelet coefficients are continuous coefficients, and are plotted as continuous waveforms as a function of time in each plot of columns (ii), (iii) and (iv). The y-axis corresponds to wavelet coefficient magnitude in each case, and the x-axis corresponds to time.

[0156] The plots in column (v) show the wavelet power ratio, WPR, computed for each of the scenarios of rows (a)-(d) for decomposition levels 1, 2 and 3.

[0157] The plots in column (v) show the wavelet power ratio for each of the plots in column (iv) plotted on the same plot. All data points above the line shown on the plot are WPR results derived from PWD data without background artefacts (rows (a) and (b)), and all data points below the line are WPR results derived from PWD data with background noise artefacts (rows (c) and (d)). Thus, it can be seen from this how the classifier can use the analysis parameters derived and shown in the plots of column (iv) to distinguish between data containing background noise (and thus being unreliable) and data not containing background noise (and thus being reliable).

[0158] The wavelet power ratio can provide an analysis parameter output of a second analysis procedure, which is subsequently fed to the classifier algorithm 38 for deriving a reliability classification.

[0159] Two exemplary first and second analysis procedures that can be applied to input ultrasound data to derive first and second analysis parameters for input to a classifier algorithm have been described above in accordance with one or more embodiments. In a first example, an envelope of a spectrogram of PWD data is extracted, and spectral and wavelet analysis is applied to this ID envelope signal. In a second example, a 2D PWD spectrogram (time-frequency domain) is derived from the input PWD data, and spectral and wavelet analysis is applied to this.

[0160] In accordance with one or more embodiments, the first and second analysis procedures can be applied to the (2D) PWD spectrogram and the (ID) PWD envelope extracted from the PWD spectrogram as described above. The analysis coefficients derived from both processes can be fed to the classifier 28 for deriving a reliability classification.

[0161] This is in Figure 11The first analysis procedure 34 comprising a spectral analysis and the second analysis procedure 36 comprising a wavelet analysis are applied to both the PWD spectrum and the extracted PWD envelope signal. A first set of analysis parameters and a second set of analysis parameters are derived in each case and provided as input to a classifier 38. Based on these inputs, the classifier is configured to derive a reliability classification 30, such as a score or ranking of the input ultrasound data. The reliability classification is indicative of the reliability of the data used to derive the blood flow measurements. The classifier can be a machine learning algorithm, e.g. a decision tree classifier.

[0162] According to one or more embodiments, the input ultrasound data can further comprise ultrasound echo data, e.g. B-mode ultrasound data, and wherein the processor is configured to apply the first analysis procedure and the second analysis procedure to the ultrasound echo data to further determine a reliability classification for the ultrasound echo data. The ultrasound echo data can then be used to generate ultrasound image data, e.g. for display on a patient monitor together with the flow measurements derived using the PWD data.

[0163] The processor can be configured to calculate an overall classification for the ultrasound data based on a combination of the reliability classifications for the PWD data and the echo data.

[0164] In Figure 12 Examples are schematically outlined in the following. Here, as in the examples of Figure 11 The first analysis procedure 34 and the second analysis procedure 36 are applied to both the PWD spectrum and the PWD envelope signal, and the resulting analysis parameters are provided to a first classifier 38a to derive a reliability classification. In addition, ultrasound echo data (e.g. B-mode ultrasound data) is also processed with the first analysis procedure 34 comprising a spectral analysis and the second analysis procedure 36 comprising a wavelet analysis. The output analysis parameters from these processes are then fed to a second classifier 38b which is configured to derive a reliability classification for the echo data. A combined reliability classification 31 or quality score can also be generated based on a combination of the reliability classifications for the PWD data and for the echo data. As an example, an overall reliability or quality score can be determined as, e.g., the maximum of the reliability or quality scores of the PWD and echo data, or the average of the scores.

[0165] According to one or more embodiments, the processor can be further configured to process the input PWD ultrasound data to derive one or more blood flow measurements. Thus, in this embodiment, the same processor that is used to assess the reliability of the data is used to determine the hemodynamic parameters. The blood flow measurements can be derived continuously or cyclically. The processor can additionally or alternatively be configured to determine one or more other hemodynamic parameters based on the input ultrasound data.

[0166] Based on extracting the blood flow velocity v from the PWD data and applying the following equation, the blood flow measurements can be derived from the input PWD data:

[0167] Blood flow = v Ar

[0168] where A is the arterial cross-sectional area and r is the heart rate. Further details of the method for deriving blood flow measurements from ultrasound data can be found in the following paper: Blanco, P. Volumetric blood flow measurement using Doppler ultrasound: concerns about the technique. J Ultrasound 18, 201-204 (2015).

[0169] According to one or more embodiments, the processor 22 can be configured to determine, based on the reliability classification(s) for the ultrasound data, a reliability of each of the one or more determined blood flow measurements (on which the flow measurement(s) are calculated). For example, if the reliability classification of the ultrasound data is low, it can be determined that the reliability of the flow measurements is low, and vice versa.

[0170] According to one or more embodiments, the input PWD ultrasound data can represent a PWD signal as a function of time, and wherein the processor is configured to decompose the signal into a plurality of time portions and generate a respective reliability classification for each time portion.

[0171] According to one or more embodiments, the processor 22 can be further configured to process each time portion of the PWD ultrasound signal to derive a respective blood flow measurement based on each time portion.

[0172] The processor 22 can be configured to determine, based on the reliability classification determined for the respective time portion, a reliability of each of the respective blood flow measurements.

[0173] According to one or more embodiments, the processor 22 can be further configured to apply a data augmentation procedure to ultrasound data for which the classifier 38 determines a low reliability classification, e.g. prior to deriving blood flow measurements from the data.

[0174] In Figure 13 An example is schematically illustrated in Fig. 2. From the input PWD data, a 2D PWD spectrogram is derived (as described in more detail above). The 2D PWD spectrogram is then passed through a plurality (n) of filters in parallel. This results in n 2D spectrograms, one for each filter. After each filter, a cyclic shift is applied to the data. The outputs from the total set of filters are then passed through a median filter which combines the n filtered spectrograms into one PWD filtered spectrogram. The output thereof provides the filtered (pre-processed) ultrasound data.

[0175] For processing, the spectrogram is processed in the form of a 2D image comprising pixels, each pixel having a pixel value representing the amplitude of a given Doppler frequency component at a given time. This approach is well known in the field of signal processing, in particular e.g. in the field of speech or audio processing, where speech spectrograms are typically visualized and processed in this way.

[0176] The filter is applied at each pixel level of the spectrogram. It uses a central pixel and a surrounding layer of 8 pixels at most one pixel away in different directions. As an advantageous example, the following filter can be used: [-1,-1], [-1,0], [-1,1], [0,-1], [0,0], [0,1], [1,-1], [1,0] and [1,1].

[0177] According to one or more examples, the classification generated by the classifier for the ultrasound data can comprise a score or a rank. Depending on the score or rank, the data can be discarded (in case of a low rank), a pre-processing data augmentation procedure can be applied (in case of a medium rank), or passed through without any pre-processing (in case of a high quality rank). In some examples, the above filtering process can only be applied to data having a reliability classification score falling within a defined range.

[0178] Examples according to another aspect of the present application provide a patient monitoring system. In Figure 14 An example system 50 is schematically outlined in Fig. 3.

[0179] The system 50 comprises a patient monitor 54 comprising a display unit 56 and an input / output 58. The system further comprises an ultrasound data processor 22 according to any of the examples or embodiments outlined above or described below.

[0180] The patient monitor 54 is configured to receive from the data processor 22 at least one reliability classification 30 generated for the input ultrasound data 24. It may also be configured to receive input indicating one or more blood flow measurement results 40 calculated based on the same input ultrasound data 24. It may be configured to generate a display output on the display unit 56 representing the blood flow measurement results and also representing the reliability classification(s) for the data(s).

[0181] The patient monitor 54 can receive the corresponding blood flow measurement results 40 calculated by the data processor 22 for each of the multiple time segments of the input ultrasound signal (as discussed above).

[0182] The patient monitor 54 can be configured to display the trend of blood flow measurement results over time.

[0183] The patient monitor 54 may also be configured to display visual indicators (as discussed above) indicating the corresponding reliability classification 30 determined by the data processor 22 for each blood flow measurement result 40.

[0184] Input ultrasonic data can be received from an ultrasonic sensing device (e.g., an ultrasonic transducer device). The data can be processed in real time by the processor 22.

[0185] An example of another aspect of the invention may provide another system.

[0186] System 60 includes an ultrasound data processor 22 according to any example or embodiment outlined above or described below.

[0187] System 60 also includes a wearable patch 64, which includes an integrated ultrasound sensing device 66 arranged to acquire ultrasound data including PWD ultrasound data.

[0188] The input / output section of the ultrasonic data processor 22 is communicatively coupled to the wearable patch 64 to receive acquired ultrasonic data from the ultrasonic sensing device. The ultrasonic sensing device may include, for example, an ultrasonic transducer array.

[0189] Despite Figure 15 In the example provided, a patch-based ultrasonic sensing device is shown, but in other examples, different types of ultrasonic sensing devices, such as ultrasonic sensing probes, can be used.

[0190] An example of another aspect of the present invention provides a method for processing ultrasonic data. Figure 16 The example method is outlined in the form of a block diagram.

[0191] The method includes receiving 102 input ultrasound (US) data, the ultrasound data including at least pulse wave Doppler (PWD) ultrasound data.

[0192] The method further comprises applying 104 a first analysis procedure to the ultrasound data or data derived therefrom to generate a first set of one or more analysis parameters, the first analysis procedure comprising a spectral analysis (e.g. Fourier analysis) of the data.

[0193] The method further comprises applying 106 a second analysis procedure to the ultrasound data or data derived therefrom to generate a second set of one or more analysis parameters, the second analysis procedure being a wavelet decomposition.

[0194] The method further comprises inputting the first set of analysis parameters and the second set of analysis parameters to a classifier, wherein the classifier is configured to determine 108, based on the input analysis parameters, at least one reliability classification for the input ultrasound data, the at least one reliability classification being indicative of a reliability of at least a portion of the input ultrasound data in order to determine a patient blood flow measurement.

[0195] Examples in accordance with another aspect of the present application also provide a computer program product comprising computer program code executable on a processor or computer, wherein the code is configured to cause the processor to perform a method in accordance with any of the examples or embodiments outlined above or described below.

[0196] As noted above, the system utilizes a processor to perform data processing. The processor can be implemented in numerous ways, with software and / or hardware, to perform the various functions described herein. The processor typically employs one or more microprocessors that can be programmed using software (e.g., microcode) to perform the required functions. The processor can be implemented as a combination of dedicated hardware to perform some functions and one or more programmed microprocessors and associated circuitry to perform other functions.

[0197] Examples of circuitry that can be employed in various embodiments of the disclosure include, but are not limited to, conventional microprocessors, application specific integrated circuits (ASICs), and field-programmable gate arrays (FPGAs).

[0198] In various implementations, the processor can be associated with one or more storage media such as volatile and non-volatile computer memory such as RAM, PROM, EPROM, and EEPROM. The storage media can be encoded with one or more programs that, when executed on one or more processors and / or controllers, perform the required functions. Various storage media can be fixed within a processor or controller or can be transportable, such that the one or more programs stored thereon can be loaded into a processor.

[0199] Variations to the disclosed embodiments can become apparent to those of ordinary skill in the art upon reading the description of the disclosed embodiments, and the claims, and the accompanying drawings. In the claims, the term "comprising" does not exclude other elements or steps, and the words "a" or "an" do not exclude a plurality.

[0200] A single processor or other unit can fulfill the functions of several items recited in the claims.

[0201] The mere fact that certain measures are recited in mutually different dependent claims does not indicate that a combination of these measures cannot be used to advantage.

[0202] A computer program can be stored / distributed on a suitable medium, such as an optical storage medium or a solid-state storage medium supplied together with or as part of other hardware, but can also be distributed in other forms, such as via the Internet or other wired or wireless telecommunication systems.

[0203] If the term "comprises" is used in the claims or the specification, it is noted that the term "comprises" is intended to be equivalent to the term "consists of" or "consisting of".

[0204] No reference signs in the claims shall be construed as limiting the scope.

Claims

1. An ultrasonic data processor (22), The processor includes an input / output unit for receiving ultrasound data (24) as input, the ultrasound data including at least pulse wave Doppler (PWD) ultrasound data. The processor is configured to: The first analysis process (34) is applied to the ultrasound data or data derived from the ultrasound data to generate a first set of one or more analysis parameters, the first analysis process including Fourier spectrum analysis of the data; The second analysis procedure (36) is applied to the ultrasound data or data derived from the ultrasound data to generate a second set of one or more analysis parameters, the second analysis procedure including wavelet decomposition. The first set of one or more analytical parameters and the second set of one or more analytical parameters are input into the classifier (38), wherein, The classifier is configured to determine at least one reliability classification (30) for input ultrasound data based on input analysis parameters, the at least one reliability classification indicating at least a portion of the reliability of the input ultrasound data used to determine patient blood flow measurement results.

2. The ultrasonic data processor (22) according to claim 1, wherein, The processor is configured to generate data output indicating the reliability classification at the input / output section.

3. The ultrasound data processor (22) according to claim 1 or 2, wherein, The processor is configured to receive and process the input ultrasound data in real time.

4. The ultrasound data processor (22) according to any one of claims 1-2, wherein, The processor is also configured to process input PWD ultrasound data to derive one or more blood flow measurement results.

5. The ultrasonic data processor (22) according to claim 4, wherein, The processor is configured to determine the reliability of each of the one or more blood flow measurements based on the at least one reliability classification of the ultrasound data.

6. The ultrasound data processor (22) according to any one of claims 1-2, wherein, The input ultrasound data also includes ultrasound echo data, and the processor is configured to apply the first analysis process and the second analysis process to the echo data to further determine a reliability classification for the echo data; Optionally, the processor is configured to calculate an overall classification for the ultrasound data based on a combination of the reliability classifications for the PWD ultrasound data and the echo data.

7. The ultrasonic data processor (22) according to claim 6, wherein, The ultrasound echo data is B-mode ultrasound data.

8. The ultrasound data processor (22) according to any one of claims 1-2, wherein, The PWD ultrasound data represents a PWD signal over time, and the processor is configured to decompose the signal into multiple time segments and generate a corresponding reliability classification for each time segment.

9. The ultrasonic data processor (22) according to claim 8, wherein, The processor is also configured to process each time segment of the PWD signal to derive a corresponding blood flow measurement result based on each time segment, and Optionally, the processor is configured to determine the reliability of each blood flow measurement result in the corresponding blood flow measurement results based on the reliability classification determined for the corresponding time portion.

10. The ultrasound data processor (22) according to any one of claims 1-2, wherein, The processor is also configured to apply a data augmentation process to ultrasound data for which the classifier determines a low-reliability classification, and optionally, the data augmentation process includes applying one or more filters.

11. The ultrasound data processor (22) according to any one of claims 1-2, in, The processor is adapted to derive PWD spectra based on the PWD ultrasound data, wherein the first analysis procedure and the second analysis procedure are at least applied to the representation of the PWD spectra; and / or The processor is adapted to process the PWD ultrasound data to derive a PWD spectrogram based on the PWD ultrasound data, and is further configured to extract a PWD envelope signal from the PWD spectrogram, the PWD envelope signal representing a time series of the maximum Doppler frequency or velocity detected at each time point, and wherein the first analysis process and the second analysis process are applied at least to the representation of the PWD envelope signal.

12. A system comprising: The patient monitor (54) includes a display unit (56) and an input / output unit; The ultrasound data processor (22) according to any one of claims 1-11, The patient monitor is configured to receive from the data processor the at least one reliability classification (30) generated for the input ultrasound data, and is also configured to receive input indicating one or more blood flow measurement results (40) calculated based on the same input ultrasound data, and is configured to generate on the display unit (56) a display output representing the blood flow measurement results and also representing the one or more reliability classifications for the data.

13. The system according to claim 12, in, The ultrasonic data processor (22) is the processor according to claim 9. The patient monitor (54) is configured to receive corresponding blood flow measurements (40) for each time segment of the PWD signal, and The patient monitor is configured to display the trend of the blood flow measurement results over time, and The patient monitor is configured to display a visual indicator that indicates the corresponding reliability classification for each blood flow measurement result.

14. A system comprising: The ultrasonic data processor (22) according to any one of claims 1-11; Wearable patch (64) includes an integrated ultrasound sensing device (66) arranged to acquire ultrasound data including PWD ultrasound data; The input / output section of the ultrasound data processor is communicatively coupled to the wearable patch to receive acquired ultrasound data from the ultrasound sensing device.

15. An ultrasound data processing method, comprising: Receive input ultrasound data, wherein the ultrasound data includes at least pulse wave Doppler (PWD) ultrasound data; The first analysis procedure is applied to the ultrasound data or data derived from the ultrasound data to generate a first set of one or more analysis parameters, the first analysis procedure including Fourier spectrum analysis of the data; The second analysis process is applied to the ultrasound data or data derived from the ultrasound data to generate a second set of one or more analysis parameters, the second analysis process including wavelet decomposition; One or more analytical parameters from the first group and one or more analytical parameters from the second group are input into a classifier, wherein the classifier is configured to determine at least one reliability classification for the input ultrasound data based on the input first group and one or more analytical parameters from the second group and one or more analytical parameters, the at least one reliability classification indicating at least a portion of the reliability of the input ultrasound data for determining patient blood flow measurement results.

16. A computer program product comprising computer program code, said computer program code being executable on a processor or computer, wherein, The code is configured to cause the processor to execute the method according to claim 15.

Images