阅读说明：本技术 胎儿超声处理单元 (Ultrasonic treatment unit for fetus ) 是由 C·F·弗兰克 于 2020-04-22 设计创作，主要内容包括：提供了用于在超声胎儿监测中使用的超声处理单元和方法。所述处理单元被配置为接收对应于胎儿区域内的一个或多个初始试验深度窗口的多普勒超声数据。从所述多普勒超声数据中提取针对每个试验深度窗口的超声信号。与现有技术的基于信号强度来寻求最优深度范围的方法不同,在本公开内容中,针对每个深度信号计算所述信号的定义的统计结构度量,其中,所述统计结构度量对应于所述信号的固有统计属性或特性。然后基于选择被估计为使从新窗口导出的超声信号的所述统计结构度量最大化的窗口来选择新记录窗口以用于采集胎儿心率信号。(An ultrasound processing unit and method for use in ultrasound fetal monitoring are provided. The processing unit is configured to receive doppler ultrasound data corresponding to one or more initial trial depth windows within the fetal region. An ultrasound signal for each trial depth window is extracted from the doppler ultrasound data. Unlike prior art methods that seek an optimal depth range based on signal strength, in the present disclosure, a defined statistical structural metric of the signal is computed for each depth signal, where the statistical structural metric corresponds to an inherent statistical property or characteristic of the signal. A new recording window is then selected for acquiring a fetal heart rate signal based on selecting a window that is estimated to maximize the statistical structure metric of ultrasound signals derived from the new window.)

1. An ultrasound processing unit for isolating individual heart rate sources within received doppler ultrasound data in fetal monitoring, the unit configured to:

receiving (32) input Doppler ultrasound data corresponding to at least one trial depth region within a uterine region of a subject, the region having a defined height and depth;

extracting (34) an ultrasound signal from the ultrasound data corresponding to the at least one trial depth region and determining (36) a defined statistical structural metric of the signal, the statistical structural metric corresponding to an inherent statistical attribute or characteristic of the signal; and is

Applying (38) a selection algorithm to determine a depth and height of a new recording region within the subject from which ultrasound signals are collected for measuring fetal heart rate, the selection algorithm being configured to select the new recording region based on the determined statistical structure metric for the trial region and based on maximizing the statistical structure metric of signals acquired from the new recording region.

2. The processing unit of claim 1, wherein the selection algorithm comprises: ultrasound signals are collected from a plurality of trial depth regions, and the defined statistical structural metric is determined for each trial region signal.

3. The processing unit of claim 2, wherein the selection algorithm comprises: a comparison of the statistical structure metric derived for each trial region signal.

4. The processing unit of claim 3, wherein the comparison comprises comparing the statistical structure metric for ultrasound signals different from each other and/or comprises comparing the statistical structure metric for each trial region with a further metric derived for a reference depth region signal.

5. The processing unit of claim 2, wherein the selection algorithm comprises an iterative process comprising: signals are iteratively collected from successive trial depth regions, a statistical structure metric is derived for each signal, and the derived metric is compared to a metric derived for a window of one or more previous trials.

6. The processing unit of any one of claims 1-5, wherein the statistical structure metric for a given signal comprises: one or more characteristics of a probability density function for the signal; a measure of instability with respect to one or more characteristics of a power spectral density function of the signal, and/or a variance of the signal.

7. The processing unit of any of claims 1-6, wherein the statistical structure metric comprises an approximation of one or more of: kurtosis or its derivatives, negative entropy, autocovariance.

8. The processing unit of any one of claims 1-7, wherein the ultrasound processing unit is operatively couplable, in use, with an ultrasound transducer unit for acquiring the Doppler ultrasound data and for adjusting acquisition settings of the transducer unit to adjust a depth region from which ultrasound signals are acquired.

9. An ultrasound device (70), comprising:

the sonication unit of any one of claims 1-8; and

one or more ultrasound transducers (76) operatively coupled to the ultrasound processing unit for providing the Doppler ultrasound data to the ultrasound processing unit, and optionally having adjustable acquisition settings for fixing a depth window of the acquired ultrasound data.

10. The ultrasound apparatus of claim 9, wherein the apparatus comprises an ultrasound probe unit containing the ultrasound processing unit and the one or more ultrasound transducers.

11. A patient monitoring system comprising:

the sonication unit of any one of claims 1-8; and

a connection interface for connecting, in use, to an ultrasound transducer unit.

12. The patient monitoring system of claim 11 further comprising an ultrasound transducer unit coupled to the connection interface.

13. The patient monitoring system of claim 11 or 12, further comprising a controller adapted to control acquisition of ultrasound data by a connected transducer unit when in use.

14. A method of sonication for isolating individual heart rate sources within received doppler ultrasound data in fetal monitoring, the method comprising:

receiving (32) input Doppler ultrasound data corresponding to at least one trial depth region within a uterine region of a subject, the region having a defined height and depth;

extracting (34) an ultrasound signal from the ultrasound data corresponding to the at least one trial depth region and determining (36) a defined statistical structural metric of the signal, the statistical structural metric corresponding to an inherent statistical attribute or characteristic of the signal;

applying (38) a selection algorithm to determine a depth and height of a new recording region within the subject from which ultrasound signals are collected for measuring fetal heart rate, the selection algorithm being configured to select the new recording region based on the determined statistical structure metric for the trial region and based on maximizing the statistical structure metric of signals acquired from the new recording region.

Technical Field

The invention provides an ultrasound processing unit, in particular an ultrasound processing unit for identifying individual heart rate signals within doppler ultrasound data.

Background

Electronic Fetal Monitoring (EFM) includes different methods of recording the vital signs (e.g., pulse rate) of a fetus during pregnancy and childbirth. A common method used by EFM systems is to detect Fetal Heart Rate (FHR) by doppler ultrasound.

This is typically performed by: a doppler ultrasound transducer is placed on the maternal abdomen and the position of the transducer is adjusted so that the fetal pulse rate source (e.g., the fetal heart or fetal aorta) is inside the volume covered by the ultrasound pulses emitted by the transducer. The transmitted ultrasonic pulses are reflected from the moving internal structure. By detecting the frequency shift in the reflected signal (doppler shift), a fetal pulse signal is generated. The fetal pulse signal may then be analyzed by the EFM system and the detected fetal pulse rate may be displayed and/or recorded.

Ultrasound (US) doppler transducers used for this purpose typically utilize an unfocused, approximately cylindrical ultrasound beam field. The extent of the beam volume is defined by a characteristic receive time window. During this window, the US transducer is arranged to acquire reflected signals from any moving anatomical structure.

In particular, ultrasound transducers typically alternate between two modes in rapid succession: a transmit mode comprising generating an ultrasound pulse; and a receive mode comprising receiving a reflection of a previously transmitted ultrasound pulse. There may be a delay or pause period in between.

By modifying the delay between the end of the transmit phase and the start of the receive phase and by modifying the duration of the transmit phase or the receive phase, the reception of reflected ultrasound can be limited to a specific depth range. This can for example be used to limit reception to depths that contain one or more sources of pulse signals and to exclude depth ranges that do not contain any sources of pulse signals but contribute to the noise level of the recorded signal.

The depth range of the recorded signal is also referred to as the "window". The window is typically a subset of the total depth range ("total ultrasound field of view") from which the transducer is able to receive signals.

The adjustment process for the depth window is illustrated in fig. 1, fig. 1 schematically depicts US observations of fetal regions at different depths. Fig. 1(a) schematically depicts a deep viewing zone, and fig. 1(b) depicts a shallow viewing zone. For plot (a), the timing and duration of the receive window of the ultrasound transducer 12 is adjusted relative to the timing and duration of the US transmissions so that US reflections from greater depths are detected. As a result, a deeper observation volume 14 is obtained. In contrast, in fig. 1(b), the receive window is adjusted so that US reflections from shallower depths are detected, resulting in a shallower depth volume 14.

Typically, an algorithm for adjusting the size and depth of the ultrasound window will seek to minimize the size of the window as much as possible, since a smaller receive window results in less noise in the recorded signal. For example, US 4984576 describes one such example depth selection algorithm.

Prior art EFM units, such as such EFM units, may use an iterative approach to determine the optimal depth range to use. In particular, the iterative method may seek a depth range that contains the greatest amount of signal strength, while excluding regions that contribute noise or weaker signal components. For example in fig. 1, the shallower depth range of fig. 1(b) provides a stronger signal because it includes the fetal heart, but excludes any noise that may be present at depths below the fetal heart.

Iterative methods can typically use two separate receive windows with independently adjustable start and end depths. One window is used for actual signal acquisition. Another window is used to investigate whether increasing the signal strength by including a larger depth range and/or decreasing the depth range will significantly decrease the signal strength. In general, the receive window range is preferably as small as possible, since a smaller window generally results in reduced noise.

The prior art depth selection algorithms (e.g., these depth selection algorithms) can work optimally when there is only a single source of pulse rate within the total ultrasound field of view. However, depending on the orientation and size of the beam field, there may be more than one source of pulse signals in the field of view: one pulse signal source corresponds to the maternal pulse rate and one pulse signal source corresponds to one (or in the case of multiple pregnancies) fetal pulse rate source.

Such iterative methods described above cannot distinguish between different pulse signal sources captured in the field of view of the transducer unit, e.g. two fetal hearts; or a maternal heart and a fetal heart. Instead, it typically includes all detectable pulse rate sources in its calculations, and attempts to include all of these pulse rate sources within the selected depth window. As a result, adjusting the procedure may result in acquiring doppler ultrasound signals that are a mixture of two or more sources of pulse rate. Such mixtures are difficult to analyze, particularly if the two pulse rate sources have similar amplitudes or powers. This may result in the fetal heart rate recording being lost or the wrong heart rate being recorded.

A reduced Fetal Heart Rate (FHR) value or a wrong FHR value may lead to a delayed diagnosis of impaired fetal health or an incorrect diagnosis of such a condition.

It would therefore be advantageous to provide an improved doppler ultrasound processing method that is capable of distinguishing between different heart rate sources within input data.

Disclosure of Invention

The invention is defined by the claims.

According to an example of an aspect of the present invention there is provided an ultrasound processing unit for isolating individual heart rate sources within received doppler ultrasound data in fetal monitoring, the unit being configured to:

receiving input doppler ultrasound data corresponding to at least one trial depth region within a uterine region of a subject, the region having a defined height and depth;

extracting an ultrasound signal from the ultrasound data corresponding to the at least one trial depth region and determining a particular statistical structural metric of the signal;

applying a selection algorithm to determine a depth and height of a new recording region within the subject from which ultrasound signals are collected for measuring fetal heart rate, the selection algorithm configured to select a window based on the determined statistical structural metric for the trial region and based on maximizing the statistical structural metric of signals acquired from the new recording region. The algorithm thus seeks to select a new recording depth zone containing only a single (fetal) heart rate source.

Embodiments of the present invention are based on techniques used in statistical signal processing. Embodiments are based on adjusting a recording depth window from which Ultrasound (US) signals are acquired according to an analysis of one or more statistical properties of the signals.

The statistical structural metric preferably corresponds to an inherent statistical property or characteristic of the ultrasound signal.

Statistical signal processing is one such method: which treats the signal as a random process, using its statistical properties to perform signal processing tasks or derive information. Statistical techniques are widely used in signal processing applications. For example, in a different area of image processing, a probability distribution of noise caused when an image is captured may be modeled, and a technique is constructed based on the model to reduce noise in the resulting image. See, for example: "Statistical signal processing: detection, evaluation, and time series analysis" by Scharf, Louis L (Boston: Addison-Wesley, 1991).

Embodiments of the invention include determining a statistical structural metric. The statistical structural metric is preferably an intrinsic property of the ultrasound signal. Thus, the statistical structure metric is preferably determined based only on a reference to the ultrasound signal (e.g., based on processing of the ultrasound signal) and not with reference to any external entity (e.g., any external signal or information or reference).

There are different options for the particular metric employed, as will be explained further below. In general, a statistical structure metric refers to a statistical property or characteristic of a signal. It refers to the degree of statistical structure (degree of statistical structure) of the signal. The statistical structure metric essentially refers to a non-stochastic measure or degree of statistical patterning of the signal. For example, the terms "statistical structural metric" and "non-stochastic metric" of a signal may be used interchangeably.

By way of further explanation, a.The books "Independent component analysis" by j.karhunen and e.oja discuss three related but not identical concepts related to the statistical "structure" of the signal, for example.

The first structural concept relates to the degree of dispersion of the probability distribution function of signals from equivalent normal distributions (or "gaussian distributions") having the same mean and variance. This type of statistical structure may be referred to as "non-gaussian".

The second statistical structure metric applies to the signal in time series, i.e., samples taken at known points in time and ordered accordingly. In this case, the statistical structure (or non-randomness) corresponds to the extent to which the value at a certain point can be at least partially predicted from known previous values. Such a structural metric may be referred to as a "structure given by linear autocovariance". Such statistical structural metrics are also captured by power spectral density, for example.

The third statistical structure metric also applies to signals corresponding to a time series, but in particular these signals exhibit an unstable variance, i.e. the variance of these signals changes slowly over time. Here, the statistical structure metric corresponds to a degree to which variance variation can be predicted based on a previous sample group. Here, although it is impossible to predict the value of a single sample from the previous value, it is possible to track the variation of the variance by considering a larger contiguous sample group.

Thus, these statistical structure metrics each relate to the degree of statistical patterning or non-randomness of the signal. However, each statistical structural metric is effectively independent of the other statistical structural metrics. Each statistical structure metric does not imply nor preset the existence of other statistical structure metrics. A signal will typically exhibit one, two or all three types of statistical structure.

In some examples, the derived signal structure metric may relate to or correspond to one major type of statistical structure, or may relate to or represent all types of structures present in the signal, for example.

The invention is based on the following insight: statistical structural measures of the signal provide a better method to determine parameters of the ultrasound reception window (depth zone) than signal amplitude/intensity/power and allow the Electronic Fetal Monitoring (EFM) system to adjust the depth window height and depth parameters such that only one pulse signal source is included in the ultrasound reception window, even if there are multiple independent pulse rate sources in the possible ultrasound field of view.

The statistical structure of a signal can be expressed in terms of different specific properties or characteristics of the signal or characteristics of different representations. For example, the statistical structure of the signal can be expressed in terms of the instability of the probability density function, the linear autocovariance, the power spectral density function or the variance of the signal. These are merely examples, and additional examples will be outlined below.

Using a criterion to analyze the experimental ultrasound receive window signal based not on signal strength but on a statistical measure of the signal structure solves the above-mentioned problems in prior art selection algorithms, in which the algorithm tends to select a depth window comprising a plurality of independent pulse rate sources. As discussed above, this problem arises because known algorithms use signal strength as a key parameter, which can result in windows with high signal strength, but where this may be due to the presence of multiple pulse sources.

Using other options instead of (preferably inherently) statistical properties, in particular the degree of statistical structure or non-randomness of the acquired signal, avoids this problem. It allows the electronic fetal monitoring system to set the ultrasound reception window parameters such that only one source of pulse signal is included in the window (even if there are several sources in the total field of view) and to record the doppler ultrasound signal from which the pulse rate can be unambiguously derived.

The flow performed by the selection algorithm may have different options.

In one set of embodiments, the processing unit may be configured to collect ultrasound signals from a plurality of trial depth regions, and wherein the selection algorithm determines a statistical structural metric for each ultrasound signal.

The processing unit may be configured to perform a comparison process to reach the height and depth of the new recording window. The same statistical structural metric is computed for each signal so that the metrics can be directly compared.

The comparison process may include comparing the statistical structure metrics to each other for different ultrasound signals. The depth area of the signal having the largest metric value may then be selected as the new recording area or window.

Alternatively, the comparison process may comprise comparing each of the metrics with a metric derived for the reference window signal. Again, based on the comparison, the window whose signal achieves the maximum value of the statistical structure metric compared to the reference may be selected as the new recording area or window.

Additionally or alternatively, the selection algorithm may be configured to perform an iterative process comprising: signals from successive trial depth regions are iteratively collected, a statistical structure metric is derived for each signal, and the derived metric is compared to metrics derived for windows of one or more previous trials. This method differs from the above-described method in that: the signals are collected one at a time from different depth regions, and the analysis is performed after the collection of each signal. In the above method, a plurality of signals are collected from different depth regions, and then a comparison process is performed using, for example, the entire set.

The selection algorithm may include: a new recording region is selected from the one or more trial depth regions from which ultrasound data was collected.

Although the above-mentioned example is of ultrasound signal collection from multiple trial depth zones, in some cases the selection of a new depth zone can be based on data from only one trial depth zone.

For example, a predefined threshold for the statistical structure metric may be pre-stored. A statistical structure metric derived for the ultrasound signal extracted for the at least one trial depth region may be compared to a predefined threshold. The selection algorithm may be configured to select the trial depth region depending on whether the statistical structure metric for the trial depth region meets, exceeds, or falls within some defined proximity of a predetermined threshold.

The predefined threshold may be pre-computed as a threshold estimated to maximize a statistical structural metric of the ultrasound signal. In other words, if the statistical structure metric for a trial depth region meets or falls within a defined proximity region of the threshold, then the statistical structure metric is most likely to be or be close to the maximum statistical structure metric. This may be based on empirical testing, for example.

The predefined threshold may be based on statistical structure metrics pre-acquired for the reference depth region. For example, the above-mentioned reference depth region may be a depth region for which a statistical structure measure is predetermined, and wherein the statistical structure measure derived for the at least one trial depth region is compared with this pre-stored measure for selecting the depth and height of the new recording region.

The foregoing merely represents a non-limiting selection of possible examples.

The statistical structure metric computed for each signal may have different options.

The statistical structural metric for a given signal may include: one or more characteristics of a probability density function for the signal; one or more characteristics of a linear cross-covariance for the signal, one or more characteristics of a power spectral density function for the signal, and/or a measure of instability of a variance of the signal.

Each of the above options will be described in more detail in the next section.

The statistical structural metric may include an approximation of one or more of: kurtosis or its derivatives, negative entropy, autocovariance. These options will be described in more detail in the next section.

The ultrasound processing unit is operatively couplable in use with an ultrasound transducer unit for acquiring the doppler ultrasound data and for adjusting acquisition settings of the transducer unit, thereby adjusting a depth region from which ultrasound signals are acquired.

The acquisition settings may include start and stop timings and durations of transmit pulses, and receive windows of ultrasound transducers comprised by the ultrasound transducer unit.

Examples according to further aspects of the present invention provide an ultrasound apparatus comprising: an ultrasound processing unit according to any of the examples or embodiments outlined above or described below or according to any claim of the present application; and one or more ultrasound transducers operatively coupled to the ultrasound processing unit for providing the doppler ultrasound data to the ultrasound processing unit, and optionally having adjustable acquisition settings for fixing a depth window of the acquired ultrasound data.

Alternatively, gating the data to select a particular depth window may be performed by the processing unit rather than locally at the ultrasound transducer unit.

The apparatus may comprise an ultrasound probe unit containing the ultrasound processing unit and the one or more ultrasound transducers. The probe unit may be a hand-held ultrasound probe. The processing unit is integrated in the probe and the signal processing is thus performed locally at the probe.

The probe unit may for example have a housing, within which the ultrasound processing unit and the one or more ultrasound transducers are comprised.

Examples according to further aspects of the invention provide a patient monitoring system comprising: an ultrasound processing unit according to any of the examples or embodiments outlined above or described below or according to any claim of the present application; and a connection interface for connecting, in use, to the ultrasound transducer unit.

The connection interface may be a wired connector or may be a wireless connection interface for connecting to a wireless ultrasound probe.

The patient monitoring system may further comprise an ultrasound transducer unit coupled to the connection interface.

For example, the transducer unit may be an ultrasound probe.

The ultrasound transducer unit may be a mere transmitting/receiving unit, i.e. comprising one or more ultrasound transducers for transmitting and sensing ultrasound signals. Here, the ultrasound processing unit includes all signal processing components (including components for digitizing and demodulating the analog signals output by the transducer unit) to separate the signals for one or more different depth regions. In this case, the analog signal is transmitted from the transducer unit to the ultrasound processing unit via the communication interface.

In other examples, the ultrasound transducer unit may additionally include local or field signal processing components for performing digitization and demodulation of signals and enabling separation of depth channels. In this case, the resulting digital data representing the separated signal channels is transmitted from the ultrasound transducer unit to the ultrasound processing unit.

The patient monitoring system may comprise a base station or base unit to which at least the ultrasound transducer unit (e.g. ultrasound probe) can be connected. The base station may include a display for displaying the results of the processing performed by the ultrasound processing unit.

The patient monitoring system may further comprise a controller adapted to control the acquisition of ultrasound data by the connected transducer unit when in use.

The controller may control the transmit and receive circuitry of the ultrasound transducer unit to acquire ultrasound signals representing different depths. The controller may control the duration of the transmit pulse and the receive window and the timing between the transmit pulse and the receive window. The controller may control the gating of the input doppler signal data over a defined time window to separate signals corresponding to one or more different depth regions within the tissue of the subject.

An example according to a further aspect of the invention provides a method of ultrasound processing for isolating individual heart rate sources within received doppler ultrasound data in fetal monitoring, the method comprising:

receiving input doppler ultrasound data corresponding to at least one trial depth region within a uterine region of a subject, the region having a defined height and depth;

extracting an ultrasound signal from the ultrasound data corresponding to the at least one trial depth region and determining a statistical structural metric of the signal;

applying a selection algorithm to determine a depth and height of a new recording region within the subject from which ultrasound signals are collected for measuring fetal heart rate, the selection algorithm configured to select a window based on the determined statistical structural metric for the trial region and based on maximizing the statistical structural metric of signals acquired from the new recording region.

The method may be a computer implemented method, for example for implementation by a suitable processor, controller or computer.

The statistical structural metric preferably corresponds to an inherent statistical property or characteristic of the signal.

Implementation options and detailed information for each of the above-described steps may be understood and interpreted in light of the description and illustrations provided above by the apparatus aspects of the invention (i.e., the ultrasound processing unit aspects).

Any of the example, option or embodiment features or details described in relation to the apparatus aspect of the invention (in relation to the ultrasound processing unit) may be applied or combined or incorporated into the method aspect of the invention.

These and other aspects of the invention are apparent from and will be elucidated with reference to the embodiment(s) described hereinafter.

Drawings

For a better understanding of the present invention and to show more clearly how it may be carried into effect, reference will now be made, by way of example only, to the accompanying drawings, in which:

fig. 1 illustrates a known method for acquiring fetal heart rate data based on iteratively adjusting the depth of the acquired signal;

FIG. 2 is a block diagram of steps performed by an example processing unit in accordance with one or more embodiments of the invention; and is

FIG. 3 illustrates an example ultrasound system in accordance with one or more embodiments.

Detailed Description

The present invention will be described with reference to the accompanying drawings.

It should be understood that the detailed description and specific examples, while indicating exemplary embodiments of the devices, systems, and methods, are intended for purposes of illustration only and are not intended to limit the scope of the invention. These and other features, aspects, and advantages of the apparatus, systems, and methods of the present invention will become better understood with regard to the following description, appended claims, and accompanying drawings. It should be understood that the figures are merely schematic and are not drawn to scale. It should also be understood that the same reference numerals are used throughout the figures to indicate the same or similar parts.

The invention provides an ultrasound processing unit and a method for use in ultrasound fetal monitoring. The processing unit is configured to receive doppler ultrasound data corresponding to one or more initial trial depth regions (depth windows) within the fetal region. An ultrasound signal for each trial depth region is extracted from the doppler ultrasound data. At least one particular statistical structure metric for the signal is calculated for each depth signal. A new recording window is then selected for acquiring a fetal heart rate signal based on selecting a window that is estimated to maximize the statistical structure metric of ultrasound signals derived from the new window.

It is an object of the invention to identify a depth window for recording an ultrasound signal containing only a single fetal heart rate source. This avoids capturing a mixed signal of multiple heart rate sources (which can lead to inaccuracies).

Existing methods known in the art are based on depth windows that seek to maximize signal amplitude or power. However, this is not suitable for isolating a single heart rate source, as areas with multiple sources may tend to generate signals of higher intensity and therefore may tend to be favoured by these known selection algorithms.

The present invention differs in that the new depth window is selected not based on the signal amplitude but on the degree of statistical structure of the signal (in effect the degree to which statistical non-randomness of the signal can be attributed, or the strength of statistical patterning that can be identified in the signal). A single heart rate source will result in a more strongly statistically structured signal than a mixture of multiple heart rate signals. Thus, the statistical structure or patterned or non-random criteria provide a more efficient and reliable criterion for selecting a depth window with an isolated heart rate source than methods that utilize signal amplitude or power.

One example according to a first aspect of the present invention provides an ultrasound processing unit for isolating individual heart rate sources within received doppler ultrasound data in fetal monitoring. The processing unit is configured to perform a plurality of steps when in use.

Fig. 2 shows in block diagram form the steps performed by the ultrasound processing unit. These steps will now be briefly outlined before being explained in more detail.

The ultrasound processing unit is configured to receive 32 input doppler ultrasound data corresponding to at least one trial depth region within a uterine region of the subject, the region having a defined height and depth.

The ultrasound processing unit is further configured to: an ultrasound signal is extracted 34 from ultrasound data corresponding to at least one trial depth region, and a defined statistical structural metric of the signal is determined 36 from the ultrasound signal.

The processing unit is further configured to apply 38 a selection algorithm to determine the depth and height of a new recording area within the subject from which the ultrasound signals are collected for measuring the fetal heart rate. The new recording area forms part of the trial area. The selection algorithm is configured to select a new depth window (which is a sub-portion of the previous window corresponding to the trial region) based on the derived statistical structure metric for the trial window and based on seeking a window in which the collected ultrasound signals are estimated to have the maximized statistical structure metric.

Thus, the selection algorithm may perform a maximization procedure to identify height and depth parameters of the new recorded depth region (which is typically smaller than the trial depth region) that maximize the statistical structural metric of the signal to be derived from the window. This is done based on the calculated statistical structural metric for one or more trial regions.

The method performed by the processing unit is based on deriving at least one predefined statistical structural measure of one or more trial depth regions. It is also based on then selecting a recording depth region that maximizes this same statistical structure metric for the US signal derived therefrom.

As discussed briefly above, there are different options for the particular metric employed. In general, a statistical structure metric refers to a statistical property or characteristic of a signal. It refers to the degree of statistical structure (degree of statistical structure) of the signal. The statistical structure metric essentially refers to a non-stochastic measure or degree of statistical patterning of the signal. For example, the terms "statistical structural metric" and "non-stochastic metric" of a signal may be used interchangeably.

By way of a set of non-limiting examples, the statistical measures of the signal structure may be higher order cumulants (e.g., skew or kurtosis or combinations thereof), or may be non-polynomial measures (e.g., negative entropy or approximations thereof), or they may be related to the temporal structure of the signal (e.g., autocovariance or autocovariance). Statistical structure metrics can be used to distinguish unstructured signals (e.g., gaussian noise or noise without temporal structure) from smaller structured signals (e.g., a mixture of multiple pulse rate sources), and then to distinguish unstructured signals (e.g., gaussian noise or noise without temporal structure) from highly structured signals (single pulse rate sources).

According to one or more further non-limiting examples, the statistical structure metric for a given signal may include: one or more characteristics of a probability density function for the signal, one or more characteristics of a power spectral density function for the signal, and/or a measure of instability of a variance of the signal.

Some of these example statistical structural metrics will now be explained in more detail.

Kurtosis is an attribute derived from the probability density function of the input signal. An estimate of the kurtosis of the signal x having length N can be obtained by calculating:

where m is the mean of x and σ is the standard deviation of x. For a signal x with a gaussian distribution, the estimate is close to a value of 3, whereas a signal with a non-gaussian distribution has a value of more than 3 or less than 3.

Thus, for signals with a gaussian distribution, subtracting 3 from the estimate and then squaring would yield a zero quantity, while for most signals with a non-gaussian distribution, subtracting 3 from the estimate and then squaring would yield a non-zero quantity. Thus, this quantity can be used as a statistical structural measure of the signal: the higher the value, the greater the deviation of the signal from a normal distribution, and thus the more non-random or statistically structured the signal.

The term [ kurtosis (x) -3] is sometimes referred to in the literature source as "excess kurtosis". However, the term may also be referred to simply as "kurtosis" in some other sources.

The criterion (kurtosis) is related to the probability density function of the signal x.

A further example of a statistical structure metric is negative entropy. The negative entropy is also related to the probability density function of the signal. The estimate of the negative entropy J of the signal x can be obtained by calculating:

for example, at A.The approximate results are given in the book "Independent component analysis" of j.karhunen and e.oja.

It should be noted that this approximation, as well as the kurtosis estimation outlined above, may be somewhat sensitive to outliers in the signal, since the third or fourth power of the signal x is used in this calculation.

In the above formula, optionally, the first term may be omitted because it is zero for any symmetric probability distribution. This leaves a single term based on the squared value of the excess kurtosis. This may be considered the simplest estimator of negative entropy and may be used as a statistical structure metric in an example embodiment.

Negative entropy has the following properties: the negative entropy is zero for signals with a gaussian distribution and greater than zero for any signal with a non-gaussian distribution. Furthermore, the increase of the non-zero value of the negative entropy is directly related to the degree to which the probability distribution for the signal x deviates from the gaussian distribution. Thus, the metric directly quantifies non-gaussian. Thus, the metric provides a direct statistical structural measure of the signal.

However, in general, the negative entropy can only be estimated for the signal, and often cannot be directly calculated.

It is possible to construct an estimator of negative entropy that is more robust to outliers. For example, at A.An example method for building such an estimator is given in the book "Independent component analysis" of j.karhunen and e.oja.

For example, a simple estimator for negative entropy j (x) is for example:

the negative entropy estimator can be made to any degree of accuracy, but at the cost of increased complexity with increased accuracy. Therefore, choosing an estimator for an application is usually a compromise between the required accuracy and the available computing power.

Another example of a statistical structural metric of a signal is linear autocovariance.

A simple example metric based on linear autocovariance is as follows:

this formula can also be found in the book "Independent component analysis" of Oja et al (see chapter 18 "Methods using time structure").

For this example, the value of the metric c (x) is zero if two adjacent samples of the signal are uncorrelated, and the value of the metric c (x) is non-zero if there is some degree of correlation between the two adjacent samples of the signal. To compare two input signal channels, the two signal channel vectors are first normalized by subtracting their respective mean values from the two signal channel vectors and dividing by their respective standard deviations (which removes any effect of the power of each channel in the result on c (x)).

For an input signal channel containing a mixture of multiple source heart rate signals, the absolute value of c (x) will be lower than the highest absolute value of c (x) for the individual heart rate signal sources.

In the context of the present invention, the selection algorithm may be configured to search for a recording area within the object having the highest absolute value of c (x).

It should be noted, however, that in the case where there are several heart rate sources and have exactly the same corresponding c (x), this approach may not be completely effective (as the several heart rate sources cannot be distinguished from each other). However, this situation rarely occurs.

A further example of a statistical structure metric is the instability of the variance of the signal.

The variance instability of the signal x can be quantified using a fourth order cumulant (which is part of the autocovariance matrix):

the value of the expression is larger for the signal x whose variance varies more in the interval N, and smaller in the case where the variance is closer to a constant.

This expression therefore quantifies this instability of the signal. Typically, a mixture of heart rate signal sources is less stable than individual heart rate signal sources alone. Thus, variance instability provides a direct statistical structural measure of the signal. In an example, the selection algorithm may search for a recording region within the object from which the acquired signal exhibits the greatest amount of variance instability.

To compare different input signal channels, this again requires first normalizing each signal x to zero mean and unit variance over the total sample range.

Typically, the first term of the formula dominates over the two following terms, so the last two terms can be omitted to simplify the calculation.

This formula may also be at a.Karhunen and e.oja in the book "Independent component analysis" (see chapter 18 "Methods using time structure").

By way of explanation, a random process can be said to be stable if its finite dimensional distribution is invariant under time translation. This type of stochastic process can be used to describe a physical system that is in a steady state but still experiences stochastic fluctuations. The concept of stability is: the distribution of stable random processes remains unchanged over time.

The exact calculation of these functions requires an infinite number of signal samples. Thus, in practice approximate results may be used, or quantities may be calculated using a number of available samples containing some information about the underlying probability distribution.

By way of further non-limiting example, provided by way of illustration, for a signal x that has been pre-processed to have a zero mean (i.e., zero baseline) and a unit variance (i.e., amplitude of 1, i.e., normalized signal), examples of statistical measures of the signal include:

kurtosis: kurt (x) E (x)⁴)-3(E(x²))²

Square kurtosis: (kurt (x))²＝[E(x⁴)-3(E(x²))²]²

Approximate result of negative entropy:

alternative approximations of negative entropy:

additional alternative approximations of negative entropy result: [ E (log (cosh (x)) -c)₂]²

Wherein E () represents the expected value operator, and c₁、c₂A constant representing a value corresponding to the first term in the case where x is a gaussian distribution.

Preferably, the selection algorithm comprises: ultrasound signals from a plurality of trial depth regions are collected and at least one statistical structural metric is determined for each trial region signal.

In some embodiments, the selection algorithm may include comparing the statistical structure metrics derived for each trial area signal. The metrics for different trial regions may be compared to each other, or the metrics for different trial regions may all be compared to one or more metrics derived for one or more reference depth regions.

By comparing the statistical structure metric(s) of the trial depth window with the statistical structure metric(s) of the reference depth window, the ultrasound processing unit is able to determine how much the height and depth parameters of the trial region should be adjusted in order to maximize the statistical structure of the signal in the recording window.

In the case of testing a plurality of depth regions and comparing their signal structure measures with one another, different statistical structure values and spectra of the corresponding depth parameters which result in these different statistical structure values are provided. This allows to derive a relation between the statistical structure metric and each of the height and depth parameters. From this relationship, a set of parameters that may result in an ultrasound signal with the largest signal structure can be derived (e.g., extrapolated or interpolated).

Alternatively, the selection algorithm may comprise an iterative process in which ultrasound signals are collected one at a time from successive trial depth regions. For each new trial depth region, a statistical structural metric is derived for the ultrasound signal. The statistical structure metric may then be compared to the statistical structure metric of the signal for one or more previously trialed depth regions, and based on the result, the algorithm determines whether and in what way the height and depth parameters for the next trialed region should be adjusted in order to improve the statistical structure metric of the signal derived therefrom.

Based on the comparison with one or more previous regions, the selection algorithm may determine, for each new trial region, whether the current region has maximized the statistical structure metric. This may be based, for example, on plotting or otherwise calculating the persistent change in the statistical structural metric as a function of the incremental trial region index. When it is detected that the statistical structural metric has reached the turning point (e.g. the second derivative of the calculated function is zero or approximately zero), the selection algorithm may simply use the height and depth parameters of the current trial depth region for the new recording region.

Alternatively, the selection algorithm may detect for each new trial region whether the statistical structure measure for that region is within a certain threshold proximity of the statistical structure measure for the previous trial region, in which case the height and depth parameters for the new region are selected for the final recorded depth region.

Embodiments of the present invention relate to controlling a depth region within an object from which ultrasound signals are extracted. The general flow of performing this operation will now be briefly outlined.

Doppler ultrasound data is obtained using an ultrasound transducer unit. This may be controlled by the ultrasound processing unit or may be performed separately. In a preferred example, the ultrasound processing unit of the present invention is operatively couplable to the ultrasound processing unit when in use, and is adapted to control the transducer unit to acquire ultrasound data. The processing unit is preferably operable to adjust the acquisition or other acoustic settings of the ultrasound transducer unit, thereby adjusting the depth region from which the ultrasound signals are acquired.

The transducer unit may comprise a plurality of ultrasound transducers or a single transducer. The received ultrasound data typically represents a plurality of different depths within the target body. From this data, one or more signal channels corresponding to signals for one or more particular depths in the body being scanned may be extracted.

In more detail, in operation, ultrasound pulses may be transmitted by the ultrasound receiving/transmitting unit into the body under investigation (i.e. the uterine region of the subject). Pulses are transmitted at a defined frequency over a defined transmit window or a set of recurring transmit windows. The receiving/transmitting unit comprises one or more ultrasound transducers for generating and sensing ultrasound signals. It is a form of ultrasound transducer unit but does not include signal processing components (in this example, the signal processing components are external to the unit).

The reflected ultrasound signals are then received at an ultrasound receive-transmit unit. Depending on which depth the signal is reflected at, reflections will be received at the receiving-transmitting unit at different points in time. Since the propagation velocity of ultrasound in tissue is known (about 1000 m/s), the time delay between transmission and reception can be mapped to the distance that the ultrasound pulse has traveled. The distance is then proportional to the depth.

In some examples, signals are transmitted in a single direction, and ultrasound signals corresponding to different depths within the resulting single cylindrical beam field are received and gated.

In a further example, the ultrasound transducer unit may comprise an array of individual ultrasound emitters, and wherein the control unit is configured to apply beamforming using the array to control the directionality of the generated ultrasound beams.

In either case, the resulting ultrasound data may be amplified by an amplifier and then divided into a plurality of signal channels corresponding to different depth regions within the subject. Alternatively, if only a single depth region is sought, the channels corresponding to the single depth region may be extracted from the set of channels.

For example, signals corresponding to different depths may be separated by gating incoming signals on different time receive windows, each gating signal then providing a different input signal channel corresponding to a different depth.

In some examples, the durations of the transmit pulse and the receive window, and the timing between the transmit pulse and the receive window, may be adjusted so that signals from a particular desired depth can be obtained, and then the signals are gated on the appropriate time window to provide a different depth signal on each depth channel.

Optionally, the gating pulses for the channels may be generated by a digital logic unit (e.g., microcontroller, FPGA, etc.) included in the receive-transmit (or ultrasound transducer) unit. The strobe signal may then be sampled using an analog-to-digital converter. Relatively low sampling rates (e.g., hundreds to thousands of Hz) may be used.

Those skilled in the art will appreciate the numerous methods for gating signals to extract channels corresponding to different depths within the body being detected.

The one or more strobe depth signals each provide a respective input signal channel (or "depth channel").

A preprocessing step is applied to each of the one or more depth signal channels. These pre-processing steps may be applied after or before the separation of the different input signal channels.

In particular, demodulation and signal integration may be applied to the input signal for each input signal channel. The demodulation generates a signal having a frequency equal to the doppler (frequency) shift of the measured doppler signal compared to the original transmitted signal.

Bandpass filtering may be applied to each depth signal channel. The filtering is configured to select frequency components of the incoming signal that are within a frequency range expected for fetal heartbeat measurement. This ensures that only the relevant frequency components of the data are retained, thereby reducing overall noise.

In some examples (not shown), an envelope demodulator may additionally be applied. For each depth signal channel, this operation provides an envelope signal corresponding to the variation (as a function of time) of the signal strength (e.g., intensity or variance) for the selected (filtered) frequency range.

In some examples, the demodulation function may be incorporated into a digital microprocessor included with the ultrasound transducer unit, in which case the reflections of the transmitted pulses may be sampled at a high sampling rate (multiple times the frequency of the transmitted ultrasound pulses (e.g., several MHz) with an analog-to-digital converter, and a digital logic unit or software in the transducer unit may demodulate the received signals and calculate a signal value for each depth channel.

Providing a digital microprocessor capable of analog-to-digital conversion at the required resolution and speed to provide processing in synchronization with signal acquisition can be challenging with current microprocessor technology. In an alternative example, the division of the depth channels may instead be performed using dedicated analog-to-digital (a/D) converters comprised in the ultrasound transducer unit. The a/D converters should in this case be operated synchronously, which means that the ultrasound frequency matches the a/D conversion frequency (preferably exactly the same as the a/D conversion frequency).

An example embodiment of an ultrasound processing unit will now be briefly outlined.

An ultrasound processing unit as described above is provided and operatively coupled to the ultrasound transducer unit when in use. The ultrasound processing unit in combination with the transducer unit is operable to record doppler ultrasound signals from at least two depth regions within a given object. The depth zone may have adjustable height and depth parameters or may have fixed height and depth parameters.

An ultrasound processing unit records the doppler ultrasound signals from each depth region and calculates a statistical structure metric for each doppler ultrasound signal. Based on the comparison of the value(s) for each signal, the processing unit either adjusts the size of the recorded depth zone and the depth parameters (or adjusts the start depth and end depth, respectively) in order to increase the degree of structuring for the final recording window from which the Fetal Heart Rate (FHR) is to be acquired. Alternatively, the unit may simply select the one of the available depth window channels with the largest statistical structure metric to use as the new recording window for collecting the signal for the FHR calculation.

The ultrasound processor may use signal structure criteria such as normalized kurtosis or negative entropy (or any of the other example criteria outlined above) to compare the different depth signal channels. These measures are not affected by the (average) amplitude of the signal, and can therefore be used to distinguish between channels containing only or predominantly one independent pulse rate signal and channels containing a mixture of pulse rate signals or no pulse rate signal at all.

As one example, by seeking to maximize the approximate negative entropy of the depth channel signal, the selection algorithm will have the effect of adjusting the window size and depth position until only one independent pulse rate signal is recorded.

Examples according to further aspects of the present invention provide ultrasound apparatus comprising: an ultrasound processing unit according to any of the examples or embodiments outlined above or described below or according to any claim of the present application; and one or more ultrasound transducers 76 operatively coupled to the ultrasound processing unit for providing doppler ultrasound data to the ultrasound processing unit, and optionally having adjustable acquisition settings for fixing a depth window of the acquired ultrasound data.

The apparatus may for example comprise an ultrasound probe unit comprising an ultrasound processing unit and one or more ultrasound transducers. For example, the probe may include a housing containing one or more ultrasound transducers and an ultrasound processing unit.

Examples according to further aspects of the invention provide a patient monitoring system. Fig. 3 illustrates an example patient monitoring system 70 in accordance with one or more embodiments.

The patient monitoring system 70 includes an ultrasound processing unit according to any of the examples or embodiments outlined above or described below or according to any claim of the present application. In the example shown, the ultrasound processing unit is incorporated inside the base station unit 72. However, in other examples, the ultrasound processing unit may be incorporated locally within the ultrasound transducer unit 76, with the base unit (being able) to connect with the ultrasound transducer unit 76.

The patient monitoring system 70 further comprises a connection interface in the form of an input connector port 74 for connecting, in use, to an ultrasound transducer unit 76, said ultrasound transducer unit 76 being for receiving input doppler ultrasound data or data derived therefrom. Figure 3 shows an example of an ultrasound transducer unit 76 for connecting in use to a base station. The transducer unit comprises an output connector 78, said output connector 78 being shaped to engage with the input connector 74 of the base station.

Where an ultrasound processing unit is included in the base station 72, the input connector 74 may be coupled to the ultrasound processing unit to communicate the received ultrasound data.

In fig. 3, the connector 74 is shown as a wired connector port. In other examples, the connector may include a wireless connection interface for connecting to a wireless ultrasound probe.

The patient monitoring system 70 may further comprise an ultrasound transducer unit 76 coupled to the input connector. For example, the transducer elements may each be an ultrasound probe.

The patient monitoring system in this example further comprises a display 80, said display 80 being operatively coupled to the ultrasound processing unit of the base station 72 for displaying the results of the performed analysis procedure, e.g. displaying a visual representation of the one or more second output signals.

The patient monitoring system 70 may further comprise a controller adapted to control the acquisition of ultrasound data by the connected transducer units when in use.

The controller may control the transmit and receive circuitry of the ultrasound transducer unit to acquire ultrasound signals representing different depths. The controller may control the duration of the transmit pulse and the receive window and the timing between the transmit pulse and the receive window. The controller may control the gating of the input doppler signal data over a defined time window to separate or extract different input signal channels corresponding to one or more particular depths within the tissue of the subject.

In some examples, the controller may be included locally within the ultrasound transducer unit, or the control steps performed by the controller may be performed locally at the ultrasound transducer unit.

As mentioned above, the ultrasound transducer unit may comprise an ultrasound processing unit. It may be an ultrasound probe unit, e.g. comprising one or more ultrasound transducers and an ultrasound processing unit operatively coupled to the processing unit. The ultrasound transducer unit may locally perform at least a subset of the ultrasound data pre-processing steps and/or control steps described above.

The patient monitoring system may take a different form to that described above. For example, the patient monitoring system may include a monitoring station (e.g., a cart-type monitoring station) that includes a display and is connectable with the ultrasound transducer unit.

In any example, the patient monitoring system can be connected with any number of other sensors or data sources for monitoring the same patient or different patients.

Examples according to further aspects of the invention provide a method of ultrasound processing for isolating individual heart rate sources within received doppler ultrasound data in fetal monitoring, the method comprising:

receiving 32 input doppler ultrasound data corresponding to at least one trial depth region within a uterine region of a subject, the region having a defined height and depth;

extracting 34 an ultrasound signal from the ultrasound data corresponding to the at least one trial depth region and determining 36 a defined statistical structural metric of the signal;

applying 38 a selection algorithm to determine a depth and a height of a new recording region within the subject from which ultrasound signals are collected for measuring a fetal heart rate, the selection algorithm being configured to select the new recording region based on the determined statistical structure metric for the trial region and based on maximizing the statistical structure metric of signals acquired from the new recording region.

Variations to the disclosed embodiments can be understood and effected by those skilled in the art in practicing the claimed invention, from a study of the drawings, the disclosure, and the appended claims. In the claims, the word "comprising" does not exclude other elements or steps, and the word "a" or "an" does not exclude a plurality. A single processor or other unit may fulfill the functions of several items recited in the claims. Although some measures are recited in mutually different dependent claims, this does not indicate that a combination of these measures cannot be used to advantage. If a computer program is discussed above, it may be stored/distributed on a suitable medium, such as an optical storage medium or a solid-state medium supplied together with or as part of other hardware, but may also be distributed in other forms, such as via the internet or other wired or wireless telecommunication systems. If the term "adapted" is used in the claims or the description, it should be noted that the term "adapted" is intended to be equivalent to the term "configured to". Any reference signs in the claims shall not be construed as limiting the scope.