GB2220487A - Apparatus for monitoring fetal heart rate - Google Patents

Abstract

A problem which occurs in monitoring the fetal heart using Doppler shift in ultrasound is to distinguish between maternal blood flow and blood flow in the fetal heart. The present invention is directed to overcoming this problem in two ways: firstly, by using a multi-range receiver (12), in which output signals are multiplied together, to receive and process the incoming Doppler signals so that varying positions of the fetus can be accommodated together with mothers of widely differing sizes; and secondly, by using a waveform comparison (14) to compare the waveform received with a typical fetal heart waveform. The waveform comparison is carried out by means of two registers containing samples of the waveforms and a multi-bit exclusive NOR gate comparing the samples on a bit by bit basis. The comparison includes summing the bits in the exclusive NOR gate without taking account of their significance.

Description

APPARATUS FOR MONITORING FETAL HEART RATE The present invention relates to apparatus for use in monitoring the fetal heart by observing modulation due to the Doppler effect, in ultrasound signals reflected from the fetal heart. The invention also relates to apparatus for comparing two-dimensional shapes, such as waveforms which may be used in the monitoring apparatus.

In recent years Doppler ultrasound has become a widely accepted screening technique for the assessment of the fetus in routine, pre-term clinical visits. Doppler ultrasound gives a non-invasive monitoring of fetal heart rate and its associated variations.

Whilst the use of fetal heart rate assessment during labour is often not now considered to be useful, clinical evidence and understanding has grown with regard to the use of heart rate measurement to characterise the response and health tone of the fetus prior to delivery. This has led to the development of domicilary fetal monitors and ambulatory fetal monitors.

Domicilary fetal monitors provide for the assessment with Doppler ultrasound of fetal heart rate in the home environment with the heart rate measurements being transferred to a clinical centre by a modem link for subsequent clinical evaluation by the obstetrician.

Ambulatory monitors have also been developed with a view to monitoring mothers at risk of premature delivery.

However these and other ultrasound systems which are commercially available suffer from serious technical and clinical limitations. The techniques and principals used for obtaining information from Doppler ultrasound are well established and were laid down some years ago. As a consequence, transducer design, the technique of range gating (to define a depth of sampling) the manner of signal processing (with autocorrelation) have become a 'de facto' standard for this type of clinical instrumentation which is adhered to even in the most recent products.

Some of the technical and clinical limitations of the current commercial ultrasound instrumentation utilised in fetal assessment are now given.

Most commercial ultrasound monitors operate in a pulsed mode in which a quartz crystal is gated to commence detection at a finite instant after the transmission of an ultrasound pulse has occurred from the same crystal. The gating interval defines a depth of sampling, with an optimum sensitivity also being associated with the defined sampling depth. This sampling depth is fixed for any given instrument and is a function of the gating interval, frequency of operation of the transducer crystals and attenuation characteristics of fetal and maternal tissue at the frequency of operation. The inflexibilities of sampling depth pose problems in situations of maternal obesity where it often proves impossible to detect a signal characteristic of the cardiac cycle.

The physiological phenomena that generates the Doppler frequency shift has never been clearly established in any of the commercial instrumentation that is available. Some manufacturers claim that fetal heart rate values are obtained from the ultrasound detecting a movement of the fetal myocardium wall. Others claim that the blood flow through the fetal heart provides the Doppler shift. In the former situation, assessment of the fetal cardiac cycle relies upon following the movement of the myocardium wall geographically at the same pooint throughout each cardiac cycle.

Unfortunately both the fetus itself and the fetal myocardium can have rapid and transient movements in three dimensions. The movement is not tracked where, as at present, Doppler shifted ultrasound measurements are performed in one dimension. As a consequence, ultrasound assessment of cardiac action is carried out at a number of different cardiac sites. The variation in site of detection produces a low frequency, high amplitude noise component in the analogue waveform generated from the intermediate mixing of the Doppler shifted components of the reflected ultrasound. This causes error in the subsequent measurement of fetal heart rate and its associated variations.

If alternatively the Doppler shift arises from fetal blood flow through the fetal heart or major arteries, the situation becomes even more complex. The fetus itself is also surrounded by a complex of maternal arteries. These include the inferior epigastric artery. the internal/external iliac arteries, the uterine arteries and the maternal aorta. The fetus itself is supported by placental blood flow. There is also forward and reverse blood flow occurring in the fetal aorta which relates to systole and diastole changes in the blood pressure. The systole and diastole changes in blood flow also cause dramatic changes in the Doppler shifted frequencies of the ultrasound.

The confusion of varying cardiac sites and differIng blood flows is often compounded by the manner of application of the ultrasound transducer by clinical staff. The transducer is usually moved over the mother's abdomen in order to search for a fetal heart rate signal. Once a beat-like signal has been detected, usually via an acoustic output, the signal is assumed to originate from the fetus. No discrimination is made with regard to the origin of the signal. It is for such reasons that reports have begun to appear of cases in which the fetus has died or has subsequently been delivered stillborn when an antenatal assessment with ultrasound has generated patterns of heart rate variation characteristic of a healthy fetus. The heart rate variations have been generated by the system presumably locking into the maternal blood flow.

In all commercial units the Doppler shifted waveform is mixed with a fixed frequency to generate an analogue waveform which is supposedly characteristic of the fetal cardiac cycle. This analogue waveform is recovered from noise by an auto-correlation procedure in which algorithms are implemented on computer boards.

The correlation process may be in a subtractive or multiplicative form but such procedures are expensive to implement and suffer from the problem of halving or doubling of the frequency of the heart rate. This in turn generates spurious heart rate dips and variations. The frequency doubling or halving of heart rate is a consequence of the time window which is set for the limits of the correlation procedure. The window also creates limits of measurement accuracy for both low and high values of heart rate. It is for such reasons that fetal heart rates of less than 80 beats/minute do not register accurately on a number of commercial ultrasound machines employed for fetal assessment.

According to a first aspect of the present invention there is provided apparatus for use in monitoring the fetal heart comprising a probe containing an ultrasound transducer, transmission means for exciting the transducer at intervals to transmit ultrasound, reception means for receiving signals from the transducer when it is not excited by the transmission means and for demodulating the received signals to derive a signal representative of the value of any Doppler shift at the frequencies of blood flow in a fetus, and comparison means for comparing the derived signal with a typical waveform derived from the reception means and due to operation of the fetal heart, and means coupled to the comparison means for indicating whether the received signals from the transducer are likely to originate from a fetal heart.

The comparison means may comprise a matched filter matched to the said typical waveform. As an alternative or in addition the comparison means may comprise means for comparing the signal derived by the receiver means, or the output of the matched filter, with the said typical waveform which is then held in memory.

An important advantage of the first aspect of the invention is that it allows signals originating from the fetal heart to be located much more easily, thus considerably reducing the problems mentioned above in relation to movement of the fetus, the presence of maternal arteries and selection of the correct signal by clinical staff.

According to a second aspect of the present invention there is provided apparatus for use in monitoring the fetal heart comprising a probe containing an ultrasound transducer, transmission means for exciting the transducer to transmit intervals of ultrasound, a plurality of reception means arranged to be enabled at different times to receive signals from the transducer after each interval of ultrasound transmission, means for demodulating the received signals to derive signals representative of the value of any Doppler shift imparted by blood flow related movements in the fetal heart, and combining means for combining signals from the receiver means before or after demodulation to provide an output signal which has an enhanced signal to noise ratio.

The combining means may be coupled to the demodulating means to receive an even number of individual signals representative of any Doppler shift and the combining means may then comprise means for multiplying the signals from the demodulating means together in pairs and then multiplying the resultant signals in pairs until a single output signal is obtained.

The problems mentioned above and relating to sampling depth can be greatly reduced by apparatus according to the second aspect of the invention since enabling the reception means at different times defines different depths of sampling. By combining, for example, four signals from the demodulation means each representative of the value of any Doppler shift in a certain range of sampling depth, a signal due to the fetal heart can be located and enhanced, thus making it easier for clinical staff to locate the correct signal.

According to a third aspect of the present invention there is provided a method of comparing two-dimensional s#hapes comprising the steps of storing a plurality of digital samples of a first shape, sampling a second shape to generate a succession of digital samples, each sample of each said shape being expressed as the same number of binary bits, the said number being at least four, repeatedly comparing corresponding bits of the digital samples of the first shape with at least one of the digital samples of the second shape to derive respective resultant binary bits dependent on the comparison, and summing the resultant bits after each comparison, without taking account of arithmetic significance, to derive a comparison coefficient.

The two-dimensional shapes to be compared are often predetermined portions of repetitive waveforms and from now on are therefore referred to as waveforms, although this aspect of the invention is expected to prove useful in pattern recognition.

The step of comparing corresponding bits may comprise an Exclusive-NOR (XNOR) operation but other operations may be suitable.

In summing the resultant bits, each comparison coefficient may be derived by summing the resultant bits from a comparison in which the bits of every byte representing respective samples of the first waveform were compared with the bits of one byte representing a single sample of the second waveform.

Alternatively, or in addition, each comparison coefficient may be derived by summing the resultant bits from a comparison in which the bits of every byte representing respective samples of the first waveform were compared with the bits of different bytes representing successive samples of the second waveform.

According to a fourth aspect of the present invention there is provided apparatus for carrying out the method of the third aspect.

The apparatus may comprise a first register for storing bits representing the bytes of a plurality of digital samples of a first two-dimensional shape, a second register for receiving bits representing the bytes of a succession of digital samples of a second two-dimensional shape, means for repeatedly comparing corresponding bits of the digital samples of the first shape with at least one of the digital samples of the second shape to derive respective resultant binary bits dependent on the comparison, and means for summing the resultant bits after each comparison without taking account of arithmetic significance, to derive a comparison repetitive coefficient.

The comparison means of the first aspect of the invention preferably comprises the apparatus of the fourth aspect.

There are important advantages in the third and fourth aspects of the invention. The need for autocorrelation in the processing of Doppler analogue waveforms is eliminated. Autocorrelation whether in a multiplicative or subtractive form requires considerable numerical processing power. Window effects which are generated as a consequence of having to impose a moving time window upon the waveforms involved in the autocorrelation procedure are also eliminated. The usual numerical processing facility is replaced by an algorithmic procedure (the third aspect of the invention) which can be implemented into silicon at the semi-custom level - this being cheaper to implement than a full custom design in silicon.

The algorithmic procedure also provides for a pattern recognition process that discriminates against waveforms not having a defined morphological shape as represented by their variation of amplitude with time, and a signal recovery procedure that discriminates against the low frequency noise components to ~bye found in the analogue envelope generated by Doppler shifted ultrasound.

The present invention also relates to methods corresponding to the apparatus of the first and second aspects of the invention.

Certain embodiments of the invention will now be described by way of example with reference to the accompanying drawings, in which: Figure 1 is a block diagram of an embodiment of the invention, Figure 2 is a block diagram of the transmitter and multi-range receiver of Figure 1, Figure 3 is a block diagram of the waveform comparison circuit of Figure 1, Figure 4 is a graph of a comparison coefficient versus time for a first mode of operation of the circuit of Figure 3, Figure 5 shows graphs of the comparison coefficient versus time for a single real time signal and several reference waveforms, Figure 6 is a block diagram of the discriminator of Figure 1, Figure 7 shows waveform sampling with time in a second mode of operation of Figure 3 and a resultant graph of comparison coefficient versus time, and Figure 8 is a graph of comparison coefficient versus time for the second mode derived with different bit weightings.

In Figure 1 a transmitter 10 generates bursts of 2 MHz oscillations at intervals of 256 microseconds which are applied to an ultrasound transducer 11. In order to assess the heart rate of a fetus, the transducer is held against the mother's abdomen and the transducer receives Doppler shifted ultrasound signals reflected from the heart of the fetus. These signals are gated into a multi-range receiver 12. As will be explained in more detail later, the receiver has four paths which are gated at different times after the end of each transmitter pulse to select reflected signals at different ranges. The ultrasound signals are demodulated in the receiver and then combined before being passed to a matched filter 13.

At the input to the matched filter the ultrasound envelope has a periodicity characteristic of the fetal heart rate but also has a significant noise content. In addition the frequency content of the envelope varies dramatically according to the origin of the Doppler shift and it is the function of the matched filter to distinguish between Doppler shifts from the mother or the placenta, for example, and from the fetal heart. Thus the filter 13 has characteristics which are matched to the ultrasound response identified with movement of the fetal myocardium. In practice a matched filter has been found to be capable of rejecting Doppler shifted responses originating from maternal blood flow.

In order to make further noise reductions the output of the matched filter is applied to a waveform comparison circuit 14 which is also described in more detail later and which has as its output a pulse signal of repetition frequency equal to that of the fetal heart. The output from the circuit 14 is applied to a discriminator circuit 41 with an adaptive threshold and the output of the circuit 41 is passed to further circuits (not shown) which, in one embodiment, provide an indication of fetal heart rate in beats per minute.

Figure 2 shows that part of Figure 1 enclosed by the dashed line 15 and gives more detail of the transmitter 10 and the multi-range receiver 12. An 8 MHz oscillator 16 provides pulses for a counter 17 which gates a 2 MHz sinusoidal oscillator 18 for 100 microsecs once every 256 microsecs. Thus the transducer 11 provides 100 microsec bursts of ultrasound at a frequency of 2 MHz, with each burst appearing at intervals of 256 microsecs. The counter 17 is also coupled to a range pulse generator 20 coupled to gates 21 to 24 receiving signals from the transducer 11. The range pulse generator 20 applies signals opening the gates 21 to 24 at different instants; the gate 21 is gated into operation 10 microsecs after the end of the transmit pulse and the gates 22, 23 and 24 are gated into operation at further successive intervals of 10 microsecs, respectively.The gates remain open for 116 microsecs but the periods when they are open overlap. In this way Doppler shifted ultrasound received by the transducer 11 in each 256 microsec period represents reflection from a different range. The ultrasound is applied to four channels and induces an optimum response in the channel having the gating, timing and depth sampling characteristics which correspond with the depth of the signal origin. The other channels also generate a response although the signal-to-noise ratio will be less in them because they are not so closely matched to the depth of the origin of the Doppler shifted signal. Thus the reflection from the fetal heart can be received with reasonable sensitivity at different distances.

The outputs of the gates 21 to 24 are passed to demodulators 25 to 28 which demodulate the 2 MHz bursts to provide envelope signals corresponding to the Doppler shifts in the ultrasound. The demodulator outputs contain a waveform which is ideally periodic at the fetal heart rate but in practice is usually contaminated with noise, including noise caused by maternal blood flow.

The outputs of the demodulators 25 and 26 are passed by way of respective band pass filters (not shown) to remove the worst excesses of noise, and sample and hold circuits (not shown) to an analogue multiplier 30 while those of the demodulators 27 and 28 are passed (again by way of band pass filters and sample and hold circuits) to an analogue multiplier 31. A further stage of multiplication is carried out by an analogue multiplier 32 which receives as inputs the outputs from the multipliers 30 and 31.If the signals from the demodulators 25 to 28 are denoted by X, T, Z and W, respectively, then the output of the multiplier 32 is given by (X.Y). (Z.W) and if Xs, X's, X" s and X " 's are the signal components from the four modulators, respectively, while the noise components are xn, yn, Zn and wn, then the resultant signal from the multiplier 30 has the following components (X5 + xn)(X's + Yn) = XsXgs + X,y, + xnX's + x,y, Since the first term is the only one which does not contain noise components, the required signal is enhanced.A similar process takes place in the multiplier 31 and the required signal is further enhanced when the outputs of the multipliers 30 and 31 are multiplied in the multiplier 32.

The output of the multiplier 32 passes by way of an amplifier 33 and a band-pass filter 34 to a rectifier 35 which detects the multiplier signal providing a unidirectional envelope signal. Another band-pass filter 36 and amplifier 37 is interposed between the rectifier output and the matched filter 13 of Figure 1.

The waveform comparison circuit 14 of Figure 1 is shown in more detail in Figure 3 and comprises a 64 bit reference register 45 considered as divided into eight bytes and a "real time" register 46 also comprising 64 bits and considered as divided in the same way as the register 45. The registers 45 and 46 are connected to a group 47 of 64 Exclusive-NOR (XNOR) gates, each gate receiving two inputs one from one of the bits of the register 45 and the other from the corresponding bit of the register 46, there being one of the XNOR gates for, and connected to, each of the bits of the register 45. A sixty-four bit comparison register 48 is connected to the outputs of the XNOR gates and a digital- summing circuit 49 is connected to sum the contents of the comparison register 48 without allowing for- the significance of the bits in the register 48.

The arrangement of Figure 3 can be operated in either of two modes and as far as Figure 1 is concerned the waveform comparison circuit 14 can therefore be switched between these modes, as required.

In mode 1, eight 8 bit samples of a reference waveform characteristic of a waveform to be recognised are shifted into the register 45, for example from a memory store (not shown). In the present example the reference waveform is a characteristic fetal waveform formed by averaging envelope signals corresponding to Doppler shifts for a number of typical healthy fetuses. The characteristic fetal waveform spans a period of about 100 ms which is about half a complete cycle from the multi-range receiver 12 and the samples are taken at 13 ms intervals. The reference-waveform samples remain in the register 45 without change while samples of a real time waveform obtained from a matched filter 13 by way of an analogue to digital converter 40 are shifted in the register 46.

The ultrasound envelope is digitised by the A/D converter 40 at a sampling rate of 3.2 ps and is serially shifted through the register 46 at a data transfer rate of 2.5 MHz. At this data transfer rate all the bytes in the register 46 contain the same sample of the real time waveform after 8 shifts have occurred and at this time control circuits (not shown) cause the XNOR gates 47 to operate. As a result each bit in the register 45 is compared with the corresponding bit in the register 46 and where the two bits are the same a "1" appears In the comparison register 48.

After the contents of the comparison register are stable the digital summing circuit 49 sums these bits without taking account of their significance so that the output of the summing circuit 49 is in the range 0 to 64.

Further samples of the real time waveform are shifted into the register 46 and when each sample occupies all the bytes of the register 46 further exclusive NOR operations and summing operations are carried out with the result that a series of comparison coefficients C(N) are derived.

If the reference waveform rises from near zero steadily to a peak and then falls steadily to near zero and the real time waveform matches it fairly closely then the comparison coefficients for one real time waveform period take the form shown in Figure 4 where a minimum is generated at the maximum of the real time waveform. Since the process described in connection with Figure 3 is continuous the output of the summing circuit 49 is periodic with minima occurring in the output of the summing circuit. The reason for the minima can be understood by the realisation that the sample values associated with the maximum values of the real time signal are compared with a smaller number of reference signal bytes having comparable values than bytes in the lower amplitude section of the real time waveform.Bytes in this lower amplitude section have similar amplitudes and therefore generate more Sls' s"in the comparison register 48 because both the leading and falling edges of the reference waveform have similar amplitudes.

The periodicity occurring in the output of the digital summing circuit 49 is equal to the fetal heart rate and therefore provides an Indication of the heart rate. However as is described below more processing provides a more useful output.

The above example refers to a particular type of waveform but the procedure has also been found to operate with more complex waveforms and examples are given in Figure 5. Assuming a real time signal 51 and different reference waveforms 52, 53 and 54 then the comparison coefficient varies as shown at 55, 56 and 57, respectively. Figure 5 illustrates how the procedure described discriminates against waveforms which do not match the reference waveform. For example the comparison coefficient series 56 has several minima the smallest of which is not at such a low level as the minimum of the series 55. Although the waveform 54 is a fairly good match for the waveform 51 and the resulting comparison coefficient series 57 has a clearly defined minimum this minimum is not at such a low level as the minimum of the series 55.This discrimination against waveforms having different shapes (or in medical applications morphologies) and amplitudes from that of the reference waveform demonstrates the capacity of the arrangement of Figure 3 to carry out a pattern recognition procedure which is also capable of generating (in the case of obstetric monitoring) a reliable fetal heart rate strobe.

To accommodate varying amplitude real time signals, and to discriminate against signals not conforming to the reference waveform, an adaptive threshold is generated by the discriminator 41 of Figure 1 which is shown in more detail in Figure 6. The output from the digital summing circuit 49 is taken to a digital-toanalogue converter 58 whose output is coupled by way of a low pass filter 59 to an analogue peak detector 60. Thus the 7 bits of the time series from the summing circuit are continuously converted into analogue form where the time constant of the filter 59 carries out a smoothing function which is followed by peak detection.The output of the peak detector 60 is applied to a potential divider 62 which provides a suitable adaptive threshold and this is converted by an analogue-to-digital converter 63 to a form in which it can be applied to a threshold register 64 which also receives the 7 bits from the summing circuit 49. Each time the value represented by the signal from the circuit 49 exceeds the adaptive threshold, the threshold register 64 provides an output. Thus each time an output from this register is obtained in a complete period of the real time waveform a heart rate strobe signal is generated, which is readily converted by further circuits (not shown) into a fetal heart rate expressed in beats per minute.

In the second mode of operation of Figure 3, the real time signal is sampled at the same rate at which the reference waveform was originally acquired (that is at 13 ms intervals). Samples of the real time signal from the analogue-to-digital converter 40 are serially loaded into the register 46 with the result that 8 different samples of the reference waveform are compared with 8 different samples of the real time waveform, the comparison being carried out in the same way as described for the first mode. As each new real time waveform sample is obtained it is shifted into the left-hand end of the register 46 and the other samples are moved up by one stage so that the oldest sample at the right-hand end is lost.As this process continues comparison is carried out in the way illustrated in Figure 7 where the upper waveform 66 represents the reference waveform and the remaining waveforms represent the real time waveform shifted through the register 46 as represented by the vertical lines 67. The resulting series of comparison coefficients is shown at 68, plotted as becoming more positive to the right along the horizontal axis. Control circuits (not shown) ensure that a comparison coefficient is generated at each time shift, that is at 1.3 ms intervals. With the 8 byte registers 45 and 46 a new portion of the real time waveform appears after 8 shifts, that is each 13 ms. Maxima, such as the maximum 69, occur periodically in the comparison coefficient series when the reference and real time waveforms coincide (unlike mode 1 where minima indicate coincidence).The period of the fetal heart beat is, of course, the time between the peaks.

The comparison coefficient series is sensitive to departure of the real time waveform from the reference waveform and where the morphologies of these two waveforms have minor differences, the degree of mismatch is reflected in the reduced range of excursion of the generated values in the comparison coefficient series.

Operation in mode 2 is found to be capable of better discrimination of different waveform morphologies than mode 1. The same method of providing an adaptive threshold as is described in connection with Figure 6 is usually applied to mode 2 operation.

The sensitivity of the first and second modes can be changed by partitioning the byte representing the comparison coefficient and weighting the portions obtained. For example the two most significant bits may be shifted up by one bit position, the three least significant bits may be shifted down by two positions and the two remaining bits may retain their original significance. The result can then be compressed into 8 bits where the third most significant bit and the third least significant bits are inserted as zero bits and the original least significant bit is lost. In this way extra weighting is given to the two most significant bits and reduced weighting given to the two least significant bits.

Figure 8 shows the improvement in sensitivity obtained, since the line 71 shows the series of comparison coefficients obtained without weighting and the line 72 shows the series obtained with the weightings suggested in the example.

In another example the four most significant bits are shifted up by one bit, the three least significant bits are shifted down by one bit, intermediate bits in the result being represented by zeros. As before the least significant bit is lost in the shift process and the result is an 8 bit number representing a comparison coefficient. With this scheme the sensitivity is shown by the line 73 in Figure 8.

The weighting technique is also useful for improving the signal contribution to the processing procedure and simultaneously discriminating against noise. These criteria are the basic essentials of any signal recovery procedure. The low and mid band frequency noise encountered in ultrasound makes a significant contribution to the sample value at small amplitude of the real time signal. (By definition there is no noise component in the reference waveform.) The random nature of noise makes itself apparent in the smaller amplitude values of the comparison coefficients since, when a comparison coefficient in the series is large, the random contribution to its magnitude is swamped by the matching of the larger amplitude components of the samples present in the registers 45 and 46. These large amplitude components have a smaller noise contribution.When a comparison coefficient tends towards a small value the large amplitude components are not matched in time and the random nature of the noise component in the real time signal makes a significant contribution to the reduced magnitude of the comparison coefficient. The reduced weighting given to the least significant bits and the enhanced weighting given to the most significant bits is a way of improving the signal to noise configuration of the processing hardware.

The circuit of Figure 3 may conveniently be constructed as an A.S.I.C. and a suitable integrated circuit of this type can be obtained from T.R.W. As one of many alternatives a PLA may be used.

The invention also has applications for detecting blood flow in general in the human body, since the technique of using a matched filter and/or the comparison technique employing a reference waveform can be used to identify blood flow in other parts of the body.

As mentioned above the comparison technique is also expected to have applications in pattern matching.

Claims (23)

1. Apparatus for use in monitoring the fetal heart comprising a probe containing an ultrasound transducer, transmission means for exciting the transducer at intervals to transmit ultrasound, reception means for receiving signals from the transducer when it is not excited by the transmission means and for demodulating the received signals to derive a signal representative of the value of any Doppler shift at the frequencies of blood flow in a fetus, and comparison means for comparing the derived signal with a typical waveform derived from the reception means and due to operation of the fetal heart, and means coupled to the comparison means for indicating whether the received signals from the transducer are likely to originate from a fetal heart.

2. Apparatus according to Claim 1 wherein the comparison means comprises a first register for storing bits representing the bytes of a plurality of digital samples of a first two-dimensional shape formed by the said typical waveform, a second register for receiving bits representing the bytes of a succession of digital samples of a second two-dimensional shape formed by the derived waveform, means for repeatedly comparing corresponding bits of the digital samples of the first shape with at least one of the digital samples of the second shape to derive respective resultant binary bits dependent on the comparison, and means for summing the resultant bits after each comparison without taking account of arithmetic significance, to derive a comparison coefficient.

3. Apparatus according to Claim 1 or 2 wherein the comparison means is arranged to derive the comparison coefficient repeatedly by summing the resultant bits from repeated comparisons in which the bits of every byte representing respective samples of the first shape are compared with the bits of one byte representing a single sample of the second shape.

4. Apparatus according to Claim 1 or 2 wherein the comparison means is arranged to derive the comparison coefficient repeatedly by summing the resultant bits from repeated comparisons in which the bits of every byte representing respective samples of the first shape are compared with the bits of different bytes representing successive samples of the second shape.

5. Apparatus according to any preceding claim wherein the reception means comprises a plurality of receiver means arranged to be enabled at different times to receive signals from the transducer after each interval of ultrasound transmission, means for demodulating the received signals to provide demodulated signals representative of the value of any Doppler shift imparted by blood flow related movements in the fetal heart, and combining means for combining the demodulated signals by multiplying at least some of them together to provide the said derived signal with an enhanced signal to noise ratio.

6. Apparatus according to Claim 5 wherein the combining means is coupled to the demodulating means to receive an even number of individual signals representative of any Doppler shift and comprises means for carrying out the multiplication of pairs of signals in multiple stages arranged to multiply the signals from the demodulating means together in pairs and multiplying the resulting products in pairs to derive further products and so on for as many stages as necessary to produce a single output signal.

7. A method for use in monitoring the fetal heart comprising exciting a probe containing an ultrasound transducer at intervals to transmit ultrasound, receiving signals from the transducer when it is not excited and demodulating the received signals to derive a signal representative of the value of any Doppler shift at the frequencies of blood flow in a fetus, and comparing the derived signal with a typical waveform derived from the operation of the fetal heart, and indicating on the basis of the comparison whether the received signals from the transducer are likely to originate from a fetal heart.

8. A method according to Claim 7 wherein comparing the derived signal with the typical waveform comprises storing a plurality of digital samples of a first shape formed by the said typical waveform, sampling a second shape, formed by the waveform of the said derived signal, to generate a succession of digital samples, each sample of each said shape being expressed as the same number of binary bits, the said number being at least four, repeatedly comparing corresponding bits of the digital samples of the first shape with at least one of the digital samples of the second shape to derive respective resultant binary bits dependent on the comparison, and summing the resultant bits after each comparison, without taking account of arithmetic significance, to derive a comparison coefficient.

9. A method according to Claim 8 wherein the step of comparing corresponding bits comprises an Exclusive-NOR operation.

10. A method according to Claim 8 or 9 wherein each comparison coefficient is derived by summing the resultant bits from a comparison in which the bits of every byte representing respective samples of the first shape were compared with the bits of one byte representing a single sample of the second shape.

11. A method according to Claim 8 or 9 wherein each comparison coefficient is derived by summing the resultant bits from a comparison in which the bits of every byte representing respective samples of the first shape were compared with the bits of different bytes representing successive samples of the second shape.

12. A method of comparing two-dimensional shapes comprising the steps of storing a plurality of digital samples of a first shape, sampling a second shape to generate a succession of digital samples, each sample of each said shape being expressed as the same number of binary bits, the said number being at least four, repeatedly comparing corresponding bits of the digital samples of the first shape with at least one of the digital samples of the second shape to derive respective resultant binary bits dependent on the comparison, and summing the resultant bits after each comparison, without taking account of arithmetic significance, to derive a comparison coefficient.

13. A method according to Claim 12 wherein the two-dimensional shapes are predetermined portions of repetitive waveforms.

14. A method according to Claim 12 or 13 wherein the step of comparing corresponding bits comprises an Exclusive-NOR operation.

15. A method according to Claim 12, 13 or 14 wherein each comparison coefficient is derived by summing the resultant bits from a comparison in which the bits of every byte representing respective samples of the first shape were compared with the bits of one byte representing a single sample of the second shape.

16. A method according to Claim 12, 13 or 14 wherein each comparison coefficient is derived by summing the resultant bits from a comparison in which the bits of every - byte representing respective samples of the first shape were compared with the bits of different bytes representing successive samples of the second shape.

17. Apparatus for comparing two-dimensional shapes comprising a first register for storing bits representing the bytes of a plurality of digital samples of a first two-dimensional shape, a second register for receiving bits representing the bytes of a succession of digital samples of a second two-dimensional shape, means for repeatedly comparing corresponding bits of the digital samples of the first shape with at least one of the digital samples of the second shape to derive respective resultant binary bits dependent on the comparison, and means for summing the resultant bits after each comparison without taking account of arithmetic significance, to derive a comparison coefficient.

18. Apparatus for use in monitoring the fetal heart comprising a probe containing an ultrasound transducer, transmission means for exciting the transducer to transmit intervals of ultrasound, a plurality of reception means arranged to be enabled at different times to receive signals from the transducer after each interval of ultrasound transmission, means for demodulating the received signals to derive demodulated signals representative of the value of any Doppler shift imparted by blood flow related movements in the fetal heart, and combining means for combining the demodulated signals by multiplying at least some of them together to provide an output signal which has an enhanced signal to noise ratio.

19. Apparatus according to Claim 18 including comparison means for comparing the output signal of the combining means with a typical waveform derived from the reception means and due to operation of the fetal heart, and means coupled to the comparison means for indicating whether the received signals from the transducer are likely to originate from a fetal heart.

20. Apparatus according to Claim 18 or 19 wherein the combining means is coupled to the demodulating means to receive an even number of individual signals representative of any Doppler shift and comprises means for carrying out the multiplication of pairs of signals in multiple stages, arranged to multiply the signals from the demodulating means together In pairs and multiply the resulting products in pairs to derive further products and so on for as many stages as necessary to produce a single output signal.

21. Apparatus for monitoring fetal heart rate as hereinbefore described with reference to Figures 1, 2, 3 and 6 of the accompanying drawings.

22. Apparatus for comparing two-dimensional shapes as hereinbefore described with reference to Figures 3 and 6 of the accompanying drawings.

23. A method of comparing two-dimensional shapes as hereinbefore described.