In the United Kingdom about 5 fetuses per 1000 die unexpectedly before birth. About 2 per 1000 result from congenital abnormality. Of the remaining losses, approximately half occur in pregnancies classified as high risk, the remainder occurring in pregnancies considered low risk. These statistics give only a partial indication of the adequacy of antenatal care. Fetal morbidity, if it remains undetected, can exert a profound influence on outcome and later development.
The motivation for monitoring the fetus through pregnancy is to recognise pathologic conditions, typically asphyxia, with sufficient warning to enable intervention by the clinician before irreversible changes set in. However, the monitoring techniques in current practice have serious shortcomings. Firstly, randomized trials have revealed that routine screening of high- risk pregnancies has no clinical benefit [Brown et al. 1982, Flynn et al. 1982, Lumley et al. 1983, Kidd et al. 1985]. Secondly, although the use of fetal monitoring has become increasingly widespread over the last 30 years, the incidence of cerebral pals y and mental handicap attributable to fetal asphyxia during pregnancy has not declined [Grant et al. 1989\]. Thirdly, the high cost of fetal monitoring equipment mitigates against the surveillance of low risk pregnancies. This effectively removes the possibility of recognising the incidence of morbidity in the normal pregnancy. To improve the performance of fetal monitoring demands firstly, a more comprehensive understanding of fetal pathology. This would further the design of more sensitive diagnostic tests . Secondly, the more precise recognition of fetal asphyxia is likely to be furthered by monitoring additional fetal activities to the heart rate. Thirdly, inexpensive fetal monitoring systems are needed. This would enable surveillance of normal pregnancie s and promote large population studies of fetal physiological development through pregnancy.
Fetal monitoring using phonography involves the transcription and analysis of vibrations originating in the fetus. This technique has the potential to address the shortcomings of existing fetal monitoring systems. Firstly, fetal phonography is inherently safe. Secondly, it can be used for surveillance of additional modalities to the fetal heart rate and thirdly it is likely that phonography can be implemented in a low cost monitoring system. Results are presented in this thesis to demonstrate how all of this can be realised.
This chapter opens with a description of the reasons for fetal monitoring and the shortcomings of existing techniques. Recommendations are made for improved diagnosis of fetal asphyxia. The principal technologies used to monitor the fetus are discussed and compared with fetal phonography. The chapter closes with a review of research into fetal phonography and a resume of later chapters of this thesis.
Techniques to monitor the fetus through pregnancy have been developed with the aim of providing sufficient information to enable the clinician to diagnose fetal well-being, characterise development and detect abnorma lity. This is antepartum fetal monitoring. Intrapartum monitoring is performed during labour and typically this takes place in a hospital environment where monitoring can be continuous and rapid intervention is feasible.
Antenatal surveillance can be justified on the basis that fetal morbidity can exert a profound influence on fetal outcome and that intervention can be made beneficially on behalf of the fetus many weeks before term. Clinicians are becoming increasingly aw are of the causal links between fet al condition in pregnancy with outcome and even later development [Symonds 1987]. This is a powerful justification for the practice of antepartum fetal monitoring and the development of enhanced techniques to improve its reliability.
The diagnostic tests of fetal well-being can be categorised as to whether they are invasive or non-invasive. Invasive tests impinge on natural (undisturbed) fetal activity. They can be subdivided into tests involving physical intervention i.e. skin ruptur e, for example blood sampling and tests disrupting fetal behaviour e.g. the contraction stress test in which the fetal reactivity to a reference stimulus is assessed.
Non-invasive diagnostic tests aim to assess normal fetal activity. Examples include, antepartum fetal heart rate monitoring and fetal movement counting. Ultrasonography, although in the strictest sense an invasive technique (energy is imparted to the fetu s), is widely considered as being non-invasive.
The greater part of antepartum fetal monitoring is concerned with the recognition of fetal asphyxia.
Severe fetal asphyxia can result in death, cerebral palsy and lesser degrees of neurological damage. Less than 10% of cases of cerebral palsy can be attributed to asphyxia during labour Fetal asphyxia can result from :
Insufficiency of uterine blood flow
Decrease in maternal arterial oxygen content
Other causes, including fetal anaemia or increased fetal demand for oxygen
In devising diagnostic tests for asphyxia two principal issues must be addressed. Firstly, what are the physiological activities that most precisely indicate the condition? Secondly, what are the normal values? Surveying the efficacy of current fetal moni toring practice suggests that there are no clear answers (yet) to either of these qu estions [Brown et al. 1982, Flynn et al. 1982, Lumley et al. 1983, Kidd et al. 1985]. They are still the subject of research and debate. The accelerating trend to seek litigation in the event of abnormal pregnancy outcome lends further urgency to investig ations aiming to improve fetal monitoring and diagnosis [Symonds 1993].
The fetal responses to asphyxia have been shown to follow a progression [Visser et al. 1990]. A number of cardio-respiratory system adjustments are made as asphyxia progresses in the f etus. This is to preserve the oxygenation of vital organs such as the brain and heart. The depth of the asphyxia can be gauged from recognising which stage the asphyxia response has reached. The correlation between asphyxia and changed fetal heart pattern s has led to fetal distress being defined in terms of these patterns [Parer & Livingston 1990] and justifies antepartum monitoring of the fetal heart rate. Fetal heart rate monitoring, typically by use of the cardiotocograph, is an accepted part of antena tal screening. However, the precise relationship between asphyxia and fetal heart rate is not well understood. This situation is likely to explain some of the inaccuracy inherent in diagnoses of asphyxia based on fetal heart rate observations.
An ensemble of measurements of fetal activity can be used to form a biophysical profile of the fetus [Manning et al. 1980, Manning et al. 1986]. It has been shown that fetal compromise is diagnosed more accurately using such a biophysical profile [Manning et al. 1986, Thacker & Berkelman 1986].
The catastrophic effect on the fetus of prolonged asphyxia has led to the widespread use of abnormal fetal heart rate patterns as an indication for intervention by caesarean section. This operation can be performed from the beginning of the third trimeste r, (this is the earliest time that the fetus can be sustained outside of the womb). The grave consequences of asphyxia combined with the difficulty of reliably diagnosing the condition provide some explanation for the increased r ate of caesarean sections amongst monitored fetuses. To control this trend requires improvements in fetal monitoring techniques and diagnostic practice.
Design of fetal tests must consider the normal patterns of fetal activity. For example many, such as breathing or movement are episodic [e.g. Dalton et al. 1977, Patrick & Richardson 1985]. Improved fetal assessment is possible by testing over longer periods [Brown and Patrick 1981]. Normal activity cycle times can be up to 120 minutes and, as a result, diagnosis of fetal distress from shorter monitoring periods must allow for this.
Computerised analysis of fetal activities offers many benefits associated with improved objectivity in diagnosis, standardisation of results and reduced operator demands.
i) In general a manually collected dataset is liable to errors if subjective observations are required or more than one observer is involved. These errors can be eliminated by replacing manual data acquisition with an automated method.
ii) The widespread adoption of consistent data collection standards permits data from different sites to be compared and a large database to be established. This is especially important in applications, such as fetal surveillance, that demand the screenin g of large populations.
iii) Data analysis can be improved by automation. In fetal monitoring many aspects of analysis rely on visual interpretation of results, for example recognition of heart rate patterns on a fetal heart rate plot . These techniques are poorly defined and as a result the inter-observer variation in interpretation is high [Nielsen et al.1987]. Inherent in automated analysis is the rigorous definition of how analyses are to be performed. iv) Automation supports the rapid availability of data analyses. The important benefits of rapid recognition of abnormality and reduced data storage requirements accrue from this.
v)Automation in monitoring technology can reduce the demands made on the operators. This can enable in experienced operators to collect valid results, a side effect of this is a likely reduction in monitoring costs.
Automated analysis has been applied to the interpretation of fetal heart rate traces. For example, the Sonicaid 8000 (Oxford Medical) range of cardiotocographs incorporate heart rate analysis software developed by Dawes and Redman [Dawes et al. 1990]. How ever, such systems are not universally available. The acceptance of protocols for automatic analysis of observed fetal modalities must follow c linical consensus. Consensus is not yet apparent in the diagnosis of fetal asphyxia. This is a further argument for an improved understanding of fetal physiology during pregnancy.
The difficulties facing antepartum fetal monitoring and the diagnosis of fetal asphyxia can now be summarised:
The fetal mortality rate is so low that to reduce the likelihood of false diagnosis, tests of fetal well-being must have a very high sensitivity (at least 95%) [Thacker & Berkelman 1986]. Attention must also focus on recognition of fetal morbidity. To design and use such diagnostic tests requires a precise model of 'normal' activity.
In the absence of continuous monitoring, fetal condition should be predictable from the results of short term tests. This calls for a model of long term fetal activity and development.
Fetal activities are episodic [e.g. Campbell et al. 1980, Dalton et al. 1977, Patrick & Ric hardson 1985] and normal cycle times can be up to 120 minutes. This sets a minimum period for the observation of these activities and necessitates long term monitoring.
Results should be available in real time enabling rapid diagnosis and reducing data storage requirements.
Fetal well-being is more sensitively tested by the simultaneous analysis of several measures of fetal activity. Fetal monitors should ideally record fetal activities i n addition to the fetal heart rate.
Objective analysis of fetal activities offers benefits associated with improved objectivity in diagnosis and standardisation of results.
Available antepartum monitoring technologies are now reviewed with an assessment of how well they meet these requirements.
Of the technologies available for non-invasive antepartum fetal monitoring systems, two may be directly compared with fetal phonography. These are ultrasonography and antepartum fetal electrocardiography (ECG).
Since the mid 1970s fetal surveillance has relied heavily on ultrasound imaging. Ultrasonic images can be formed by the selective reflection of acoustic energy at inhomogeneities in soft biological tissue. In medical imaging ultrasound in the frequency ra nge 2-20Mhz is coupled into the body by means of a piezo-electric transducer. The available systems for fetal monitoring can be divided into those providing 1 and 2 dimensional image data.
The 1 dimensional systems use a narrow ultrasound beam which is used to 'illuminate' specific fetal structures. The resulting ultrasound reflections are detected and by using the Doppler principle the motion of the reflecting structure can be quantified. Doppler ultrasound is routinely used to 'visualise' the motion of fetal structures such as heart valves. This can be used as the basis for estimating heart rate. Doppler ultrasound can also be used to quantify blood flow in major arterie s. This can be useful to assess cardiohemic responses.
Portable (bedside) Doppler ultrasound systems to monitor the fetal heart rate, e.g. the cardiotocograph, are in common use. However there are objections to their routine use in long term fetal monitoring. Firstly, there is some concern amongst clinicians on the safety to the fetus of prolonged exposure to ultrasound radiation. Secondly, the naturally occurring fetal movements must be tracked by the ultrasound beam. This implies close operator supervision since gross fetal movements are often sporadic.
The 2 dimensional systems generate an array of ultrasound beams. The beam array can be swept or scanned across the fetus and a 2 dimensional image formed using the ultrasound reflections detected. This is referred to as B-mode (brightness mode) ultrasonography. The 'images' produced are very noisy but a skilled operator can routinely visualise fetal heart action and fetal movements including fetal breathing.
The systems providing 2 dimensions of image data enable the routine characterisation of many physical fetal activities. However, their high capital and running costs limits their use to short term monitoring in a clinical environment.
Fetal electrocardiography provides a method for monitoring the fetal cardiac cycle. Two systems are in use; direct, in which the fetal ECG is recorded using an electrode attached to the scalp, and indirect or abdominal fetal ECG, in which the fetal ECG is sensed at the maternal abdominal wall. The former technique is invasive and can only be performed during labour. As for the adult, by reflecting cardiac and metabolic activity, the fetal ECG is potentially a sensitive indicator of fetal state.
A number of difficulties are associated with recording the abdominal fetal ECG, some of which remain unsolved. They principally concern isolating the abdominal fetal ECG signal from the other electrical activity detected at the maternal abdomen. For examp le, the magnitude of the fetal ECG signal at the ma ternal abdomen is of the order of several microvolts. This is a fraction of the maternal ECG amplitude recorded at the maternal abdomen. Other major sources of interference are maternal muscle activity and signals generated by electrical equipment. These difficulties are compounded by the sensitivity of the abdominal fetal ECG signal amplitude to fetal position and to the normal reduction in fetal ECG signal amplitude with gestational age. Techniques to improve fetal ECG signal acquisition remain the subj ect of ongoing research [Callearts et al 1989, Oostendorp et al. 1986].
Fetal phonography is the transcription of fetal sounds. Phonocardiography specifically involves the transcription of fetal heart sounds. In both cases this is achieved by sensing fetal vibrations incident on the maternal abdomen.
Fetal phonocardiography has a history dating from 1818 and is reviewed in Goodlin . It is important to note that to date the clinical results obtained using phono cardiography for fetal monitoring have been poor in comparison those obtained using Doppler ultrasound systems [Solum 1980]. Ultrasonography currently dominates fetal heart monitoring practice. The lack of success of phonographic monitoring systems is attributed to inappropriate transducer design; typically the transducers used for fetal heart monitoring were variants of those used for adults. The resulting signal had a poor signal to noise rati o. This required heavy filtering which in turn led to attenua tion of potentially valuable signal information. The design of compliance matched phonography transducers [Talbert et al. 1986, Cohen 1989, Goovaerts 1989] represent a considerable improvement over earlier designs. Firstly, the phonography signals acquire d potentially have a signal to noise ratio sufficient that heavy filtering is unnecessary. Secondly, these phonography sensors can characteristically operate over a wide range of fetal vibrations. Taken together, these factors have given rise to the conce pt of wide bandwidth fetal phonography. Wide bandwidth fetal phonography enables recording of low frequency vibrations arising from activities such as fetal breathing and movements, in addition to the higher frequency vibrations arising from heart sounds.
Although phonography transducer development is at an advanced stage, robust signal processing techniques to allow routine estimation of fetal activities are not yet available. As a result, interpretation of wide bandwidth phonography and phonocardiography signals obtained using the latest transducers remains at an experimental stage.
The important features that distinguish phonography from ultrasonography and fetal electrocardiography are :
The latest phonography transducers are able to sense fetal vibrations over a wide frequency range and therefore can record a range of fetal activities [Cohen 1989, Colley et al.1986].
Phonography is non-invasive, imparts no energy to the fetus and therefore is inherently safe for long term monitoring.
The latest phonography transducers are sufficiently sensitive that fetal activity can still be recorded after the position of the fetus, relative to the transducer, has changed. This is in direct contrast to Doppler ultrasound systems for which fetal movements must be tracked. In this respect, modern phonographic systems may have a reduced requirement for skilled operators. The latest phonographic techniques have the capa bility for long term simultaneous monitoring of a range of fetal activities. This ensures that potentially, phonography has an important role in antepartum fetal health care.
The requirements of a precise test of fetal distress were summarised above. . These are potentially capable of being satisfied using a monitoring technique based on phonography:
i)Low risks and reduced requirement for operator attendance permit long term monitoring using phonography. Episodic fet al activities might thereby be more adequately characterised.
ii)The estimation of several fetal activities using phonography may be used to create a phonographic biophysical profile. This is likely to promote more accurate diagnoses of fetal state.
iii) The low operating and likely equipment costs dispose the technique to a large population survey. Such a survey would enable normal activity patterns to be characterised.