System and method for detecting placental volume changes

Abstract

The system (100) comprises a flexible support element (10) that is wearable on the abdomen of a pregnant subject and one or more strain sensors (20) supported by the flexible support element (10) and arranged to detect strain in the skin of the abdomen of the subject indicative of a placental contraction. The one or more strain sensors (20) may comprise a fibre Bragg grating. The system (100) may comprise at least one electrode (51) arranged to be held in contact with the skin by the flexible support element (10). The system (100) may comprise a processor (70, 90) which may be configured to identify a placental contraction by determining a strain envelope by finding troughs in a strain signal. The processor (70, 90) may be configured to implement a machine learning algorithm that has been trained to classify signal data as indicating, or not indicating, a placental contraction.

Description

[0001] SYSTEM AND METHOD FOR DETECTING PLACENTAL VOLUME

[0002] CHANGES

[0003] FIELD OF INVENTION

[0004] The present invention relates to a system and method for detecting placental volume changes, which may also be termed placental contractions.

[0005] BACKGROUND OF THE INVENTION

[0006] Complications can arise in fetal development. The potential success of medical intervention is likely to be increased by early detection of any abnormalities in fetal development, or in the development of the placenta, which is critical for fetal health and development. Systems and method for monitoring physiological parameters that are indicative of placental and fetal health are therefore desirable. It is further desirable that such monitoring systems and methods are cost effective and straightforward to deploy.

[0007] Medical devices are known that can be used to detect a fetal electrocardiogram (fECG) without making physical contact with the fetus. Such devices use electrodes that are placed on the mother's skin to detect electrophysiological signals. Maternal uterine contractions, often referred to as uterine activity (UA) can be determined by electrohysterography (changes in electrical potential due to uterine contractions).

[0008] SUMMARY OF THE INVENTION

[0009] The present invention provides systems and methods for detecting placental contractions.

[0010] In recent research, orchestrated contractions of the placenta and the underlying uterine wall have been observed, leading to transient reductions in placental volume, with subsequent relaxation. This phenomenon is referred to as a “uteroplacental pump” and contractions associated with this have been named placental contractions. In contrast to uterine contractions which involve the whole uterine wall, placental contractions involve contraction of the placental bed and a reduction in placenta volume. This is a contraction of the placenta and the part of the uterine wall to which it is attached. During a placental contraction, the placenta and underlying uterine wall contract independently of the rest of the uterus, expelling maternal blood from the intervillous space. Such contractions differ from other uterine contractions, such as Braxton Hicks contractions in which the entire uterus contracts in practice for labour.

[0011] It is believed that the placental contractions reduce blood stagnation in parts of the placenta. Compromised fetal health is often due to insufficient supply of nutrients through the placenta. The significance of the placental contractions in abnormal pregnancy has yet to be studied in detail, but it is believed there is a link between placental contractions and the health and development of the fetus . One hypothesis is that the placenta contracts to pump fresh oxygenated blood from the mother to the fetus. This would in turn remove the deoxygenated blood along with any impurities the fetus is attempting to remove. Fewer placental contractions may mean the fetus is not getting enough oxygenated blood and is not able to remove impurities fast enough, which could cause adverse health implications for the fetus.

[0012] In a clinical context, placental contractions have been observed using MRI. The contractions involve a reduction in the area and thickening of the myometrium underlying (covered by) the placenta, a stretching or expansion of the rest of the uterine wall, and a simultaneous reduction in placental volume by up to 40%. The contractions vary in strength and duration and cause the placenta to become thicker with less flat edges and to appear darker and more heterogeneous on T2 -weighted images, with dark bands in T2 corresponding to bright bands on the susceptibility maps. Braxton Hicks contractions, on the other hand, are characterised by contraction of the entire uterine wall without alteration in placental volume.

[0013] It is desirable to be able to monitor the placental contractions over a longer period than would be practical with MRI. Moreover, it is important to monitor placental contractions in the mother’s normal environment, including during sleep. The system and method disclosed herein is suitable for monitoring placental contractions and may be configured to distinguish them from uterine contractions and fetal movements.

[0014] According to a first aspect, there is provided a system for detecting placental contractions, comprising a flexible support element that is wearable on the abdomen of a pregnant subject; and one or more strain sensors supported by the flexible support element, arranged to detect strain in the skin of the abdomen of the subject indicative of a placental contraction. Such a system can provide information that is indicative of placental and fetal health and development, which in turn may support diagnostic processes and clinical decision-making.

[0015] By providing indications derived from the utero -placental pump, the system disclosed herein may be beneficial for regular (e.g. daily and / or nocturnal) monitoring of fetal activities to reassure the mother of their fetus’ well-being. The system may contribute to strengthening the bond between mother and fetus. The system may further provide support for clinical trials to establish the benefits of fetal movement monitoring, such as early detection of fetal distress or pathologies of the placenta / fetus. The system as disclosed herein may additionally be used to refine guidelines on definitions of PVCiU (Premature Ventricular Contractions in-Utero), reduced fetal activities and recommendations for daily PVCiU / fetal movement monitoring. The system may be used by consumers and in clinical practice (hospitals, clinics, Children’s Centres etc).

[0016] The one or more strain sensors may be arranged to detect strain in the skin of the abdomen of the subject indicative of maternal heart rate and / or fetal heart rate. The one or more strain sensors may comprise one or more strain sensors arranged at a suitable position on the abdomen of the subject to determine maternal heart rate and / or one or more strain sensors arranged at a suitable position on the abdomen of the subject to determine fetal heart rate. Maternal and fetal heart beats will produce periodic changes in strain in the skin of the abdomen of the subject and the one or more strain sensors may be arranged to detect such periodic changes in strain. Maternal heart rate and / or fetal heart rate can be inferred form the periodic changes in strain based on a known relationship between maternal heart rate and / or fetal heart rate and strain , for example.

[0017] The one or more strain sensors may comprise an optical fibre strain sensor. One or more optical fibre strain sensors may be embedded in biocompatible patches and arranged on the abdomen in various formations. The optical fibre strain sensors may be configured to record abdominal movements. The abdominal movement detected by the optical fibre strain sensors may arise as a consequence of placental contractions, but may also result from other abdominal movement such as fetal kicks, fetal rolls, uterine and Braxton Hicks contractions. The strain sensor may be used in combination with reference devices to isolate strain signals from other sources of cross sensitivity, such as temperature , and from the mother’s physiological parameters such as breathing and heart beat, and mother’s activities such as coughing and talking . The optical fibre strain sensor may comprise a fibre Bragg grating (FBG) strain sensor. FBGs are sensitive to strain - the Bragg wavelength that is reflected by the grating depends on the spacing of the grating, and longitudinal strain on the fibre will change the grating spacing, resulting in a shift in the Bragg wavelength of the grating. Where the FBG is embedded in a patch or sheet of elastic material, pressure on the patch / sheet will lead to a transverse extension of the patch (the Poisson effect). An optical fibre strain sensor embedded in a patch may therefore also be responsive to pressure exerted on the patch. The optical fibre may be thin, small and flexible and therefore conform easily to the curvature of the abdomen. Fibre optic technology is non-invasive, passive, chemically inert, non-toxic and immune to electromagnetic interference. It is safe to use and poses negligible risk to both mother and fetus.

[0018] Alternatively, or in addition to at least one FBG strain sensor, the optical strain sensors may comprise:

[0019] • Intensity-based sensors, where bends in the fibre due to movement change the transmitted intensity.

[0020] • Fabry-Perot interferometric based sensors, where strain over the sensing region leads to a change of a cavity length of a Fabry-Perot cavity and hence the detected optical signal (a wavelength of interrogation may be selected to match a maximum sensitivity of the etalon response, so that changes in strain modulate the intensity of light reflected by the Fabry -Perot etalon).

[0021] • Michelson Interferometric and Mach-Zehnder based sensors, where strain in one arm of the interferometer changes the optical phase and hence the detected optical signal.

[0022] The system may advantageously comprise an optical interrogator supported by the flexible support element and configured to detect signal data indicating strain from the one or more strain sensors. Thereby, the system remains wearable and portable even when it comprises the optical interrogator. The optical interrogator may be a wireless, battery powered, monitoring unit that records the acquired signals. The optical interrogator is advantageously of a miniaturised size and with a light weight. By way of example, the interrogator may be approximately 75 x 50 x 15 mm, and weigh approximately 60 grams. Thereby, the interrogator is easily supported by the flexible support element. The interrogator may be attached to the support element using any suitable restraints or fastening means, such as straps, cables, hook-and-loop type fasteners, and / or magnetic arrangements. In some examples, the optical interrogator is arranged in a pouch or pocket of the flexible support element.

[0023] The optical interrogator may be arranged with more than one channels. Thereby it is possible to have two optical fibres comprising strain sensors (FBG sensors, for example) placed at different locations of the abdomen to cover the full region of interest. In other embodiments a single optical fibre can be disposed to cover the full region. Another embodiment is to provide optical couplers to link together at least 2 arrays of FBGs (all with different centre wavelengths) to connect to a single channel interrogator.

[0024] The one or more strain sensor may be embedded in at least one flexible substrate. The flexible substrate may be biocompatible and may comprise any suitable material, such as a biocompatible silicone (e.g. platinum based RTV2 silicone. Each flexible substrate may comprise an adhesive (or tacky) layer (e.g. so that the substrate is self-adhesive) for improving coupling of the flexible substrate to the skin. The adhesive layer may assist in coupling strain in the skin to the flexible substrate. The adhesive layer may provide a temporary yet secure and effective attachment to the skin.

[0025] Advantageously, each of the at least one flexible substrate may comprise a patch. The at least one flexible substrate may comprise at least one elongate patch. The (or each) elongate patch may be longer than it is wide. In some examples the elongate patch may be three times longer than it is wide. In other examples the elongate patch may be five, ten or fifteen times longer than it is wide. In some examples, the elongate patch is between 2 cm and 5 cm wide and between 15 cm and 25 cm long. In other examples, the elongate patch is between 1 cm and 3 cm wide and between 10 cm and 20 cm long. The elongate patch is beneficially thin, with a thickness of less than 5 mm, and preferably less than 3 mm.

[0026] In some examples, the one or more strain sensor comprises an embedded optical fibre. One or more optical fibres, each comprising multiple FBGs, may be embedded in one or more elongate patches covering the full region of interest. Such a configuration may provide a good fit for mothers of different BMI and gestational age. For detecting placental contractions, the strain sensors may be embedded into a soft silicone patch. The patches with embedded strain sensors may be suitably biocompatible and reusable (for example after suitable cleaning) . The patch may be configured to be washable (using cleaning wipes, for example) and subsequently re-usable.

[0027] Each flexible substrate may be supported in contact with the skin in a predetermined layout by the flexible support element. Such a predetermined layout may be based on a region of interest of a particular case, which may be important for obtaining signal data from which placental contractions can be identified. Using multiple flexible substrates may improve the versatility of the system and may contribute towards providing a good fit for mothers of different BMI and gestational age.

[0028] The flexible support element may comprise a textile or fabric material. The one or more strain sensor may be woven or knitted or threaded through with the textile or fabric material. An optical fibre, for example, can be threaded through a fabric or textile in such a way that strain in the textile results in a corresponding strain in the optical fibre. The flexible substrate may thereby comprise a portion of the flexible support element. The surface of the fabric or textile facing the skin of the abdomen may be provided with a self-adhesive(or tacky) polymeric coating to enhance the coupling of the fabric or textile with the skin.

[0029] The flexible support element may comprise a pregnancy support belt or belly band (or any other garment), arranged to wrap around the abdomen. The flexible substrate, for example in the form of a patch, may be incorporated into the support belt or belly band. The optical interrogator may also be arranged or attached to the support belt or belly band, or within a pocket arranged in or on the support belt or belly band. The interrogator may be arranged or disposed in a position that is suitable for sleeping while wearing the support belt or belly band (for example, on the front of the abdomen, accommodating typical sleeping position for a pregnant woman). Such an arrangement may improve the possibility for continuous monitoring of placental contractions, as the presence of the sensor and the interrogator does not inconvenience the wearer. Such an arrangement may provide a system for detecting placental contractions that can be worn comfortably by the mother during sleep. The support belt or belly band may include additional padding for increased comfort. Being able to provide a wearable, and even comfortable, system for detecting placental contractions may be also beneficial in a clinical trial setting, as it may decrease the risk of test subjects choosing to withdraw from the trial. The system may also be used during non-clinical trials, for example in a normal home setting, and / or in clinical evaluation.

[0030] In some examples, the sensors may be embedded in a single patch placed on the abdomen. In other examples, other configurations of sensors may advantageously be selected for different purposes. In some examples, the at least one strain sensor may comprise sensors that are oriented in different directions. The sensors may, for example, be arranged in substantially perpendicular directions. In some examples, at leas t one first strain sensor may be arranged substantially over the placenta and at least one second strain sensor may be arranged perpendicular to the first strain sensor to detect any physical movement propagating perpendicular to the placenta.

[0031] The at least one strain sensor may comprise at least one strain sensor that is oriented in an approximately horizontal direction and at least one strain sensor that is oriented in an approximately vertical direction. The flexible support element may be configured to orient the strain sensors when the system is worn by the subject. The orientation of the strain sensors may be configured to conform with the abdomen and may thus vary somewhat based on the physiology of the subject wearing the strain sensor. The horizontal direction may also be described as extending substantially along a frontal axis of the subject, essentially in a direction extending across the abdomen from the right-hand side to the left-hand side at a substantially constant height from the feet of the subject. The vertical direction may also be described as stretching substantially along the craniocaudal direction, essentially in a direction from the feet to the head of the subject. In other words, the vertical direction may be expressed as an up-down direction of the abdomen, and the horizontal direction may be expressed a left -right direction across the abdomen. The strain sensors may lie between the uterus fundus and symphis pubis.

[0032] In examples where the at least one strain sensor is embedded in a patch, two patches may be arranged approximately perpendicular to each other. Alternatively, a single patch can replace the perpendicularly arranged horizontal and vertical patches .

[0033] It is not essential that the strain sensor is an optical strain sensor (for example, in which readout is via an optical fibre). In some examples one or more strain sensors may be implemented using another technology, for example piezoresistive semiconductor strain sensors, foil strain gauge sensors or any other suitable type of strain sensor. Smart textile-based strain sensors may be used, for example, comprising conductive yarn. In some embodiments it may be possible to detect strain using electrical sensors - for example, when a fetus moves or kicks the impedance between two electrodes also changes. An electrical impedance can therefore be used as a proxy measurement for strain.

[0034] In some examples, the system may comprise at least one electrode (e.g. two) arranged to be held in contact with the skin by the flexible support element. The strain sensor may in such examples be used in combination with electrical measurements made via the at least one electrode placed in contact with the skin, preferably on the abdomen of the subject. The electrical activity may be sensitive to, for example: uterine contractions, Braxton-Hicks contractions, maternal and fetal heart rate (e.g. baseline and variability), and fetal and maternal movements. By providing electrical data the system may be able to determine fetal and maternal heart rate and variability changes (for example, LTV (Long Term Variation), STV (Short Term Variation), RMSSD (Root Mean Square Successive Differences) , pNNx (probability of successive cardiac interbeat intervals greater than x ms)) before, during and after a placental contraction. Such information allows fetal and maternal sympathetic and parasympathetic nervous feedback mechanisms to be assessed.

[0035] The at least one electrode may comprise a plurality of electrodes spaced apart by at least 3 cm. The at least one electrode may comprise an electrode configured to be positioned substantially on the midline of the subject on the abdomen, spaced apart from and below the navel by at least 3 cm. The system may comprise an electrical readout circuit supported by the flexible support element and configured to detect electrical signals from the at least one electrode.

[0036] The combination of electrical and strain measurements, such as with FBGs, may allow separation of minor fetus movements and full body movements, and may assist with other identification of false positives and / or false negatives placental contractions.

[0037] Furthermore, a combination of strain sensing and electrical sensing improves the ability of the system to discriminate placental contractions from other maternal movements, fetal kicks, rolls, uterine and Braxton-Hicks contractions. For example, the system may be configured to perform electrohysterography (EHG) to identify uterine activity. A component of strain may be identified as resulting from uterine activity and subsequently discounted from indicating a placental contraction. Similarly, EHG may be able to identify periods of strain resulting from fetal activity or from Braxton-Hicks contraction, and these may subsequently be discounted as indicative of a placental contraction. Other methods may be used for discounting strain resulting from other sources than placental contraction. For example, machine learning may be used to jointly process electrical signal data (e.g. EHG) and strain data to classify periods as placental contractions or not placental contractions. The machine learning model may be trained so that it can use the electrical signals and strain signals to discount strain from another source that can be identified from the electrical signals. In addition, a combination of placental contractions and other monitoring may be useful to improve the understanding of the health of the fetus. For example, in addition to contraction monitoring, the sensors may be used to monitor the physiology of the placenta, such as oxygen saturation during contractions. In combination with fetal and maternal heart rate monitoring, the effect of placental contractions on the fetus may be better understood using data from an embodiment.

[0038] In examples in which the one or more strain sensors are arranged to detect strain in the skin of the abdomen of the subject indicative of maternal heart rate and / or fetal heart rate, the at least one electrode may be omitted. Advantageously, the one or more strain sensors may be arranged to detect strain in the skin of the abdomen of the subject indicative of both a placental contraction and maternal heart rate and / or fetal heart rate, thereby removing the need for at least one electrode to monitor maternal heart rate and / or fetal heart rate. This provides a simplified system.

[0039] In some examples, the system may comprise a processor configured to receive signal data comprising an indication of the strain detected by the one or more strain sensors. The processor may be configured to identify a placental contraction in the subject from the signal data. The processor may be configured to determine maternal heart rate and / or fetal heart rate from the signal data. The processor may be supported by the flexible support element. Alternatively, the processor may be remote (e.g. not worn with the flexible support element). A combination of a local processor and a remote processor may be used. The remote processor may be configured to receive signal data wirelessly (or by a wired connection). Signal processing algorithms may be used, which classify different types of abdominal activity based on the acquired signals. The types of abdominal activity may include placental contractions, maternal heart rate, fetal heart rate, uterine contractions, Braxton Hicks contractions, and fetal movement, including fetal rolls.

[0040] The at least one strain sensor may comprise a plurality of strain sensors, and the processor may be configured to combine strain detected at each strain sensor to determine an average strain signal.

[0041] Identifying a placental contraction may comprise determining a strain envelope by finding troughs in a strain signal. Determining maternal heart rate and / or fetal heart rate may comprise identifying individual heart beats by finding peaks in the strain signal. The strain signal may, in some examples, be the average strain signal determined by the processor (or a smoothed average strain signal) . The processor may additionally be configured to smooth or low pass filter the strain envelope.

[0042] The processor may be configured to implement a machine learning algorithm that has been trained to classify signal data as indicating, or not indicating, a placental contraction. The machine learning algorithm may comprise an artificial neural network that has been trained using supervised learning from a training dataset derived from MRI data in which placental contractions have been measured.

[0043] In any example, the signal data may comprise the electrical signals. The machine learning algorithm may be trained using a labelled dataset in which events causing strain in the skin of the abdomen which are not placental contractions are identified, for example derived from MRI data. The machine learning algorithm may be configured to classify signal data that indicates one or more of: fetal movement, breathing, and / or uterine contractions (including Braxton Hicks contractions).

[0044] In some examples, the system may comprise a wireless transmitter configured to transmit (i) signal data comprising an indication of the strain detected by the one or more strain sensors to a remote device; and / or (ii) an indication of the presence or absence of placental contractions; and / or (iii) electrical data either raw or processed to indicate fetal and maternal wellbeing. In some embodiments of the system disclosed herein, the strain sensor may be used in combination with Near Infrared Spectroscopy (NIRS) where the oxygen saturation of the placenta is measured. The relative changes in placental contractions and oxygenation of the tissue are potentially useful for prediction of fetal and / or placental development issues.

[0045] In other embodiments of the system disclosed herein, the strain sensor may be used in combination with capillary refill time monitoring of the superficial tissue. Changes in deep tissue blood flow e.g. muscle, manifests itself as changes in superficial blood flow. Relative changes in placental contraction and changes in superficial blood flow potentially contains useful information for prediction of fetal and / or placental pathologies.

[0046] According to a second aspect, there is provided a method of detecting placental contractions. The method comprises arranging at least one strain sensor on the abdomen of a pregnant human subject; obtaining signal data from the strain sensor; and processing the signal data to determine whether the signal data indicates a placental contraction.

[0047] The method of the second aspect may be performed using the system for detecting placental contractions disclosed herein (e.g. according to the first aspect, including any optional features thereof, or according to any embodiment described herein) .

[0048] According to a third aspect, there is provided a wearable device for detecting placental contractions. The device comprises a flexible support element and strain sensors supported by the support element. The strain sensors may be embedded in, or attached to, the flexible support element. The device may further comprise an optical interrogator supported by the support element. The flexible support element may comprise a textile or fabric material. The flexible support element may be a component of a pregnancy support belt or belly band. In some examples, the flexible support element itself is a pregnancy support belt or belly band. The flexible support element may comprise a pocket, suitable for housing the optical interrogator.

[0049] The features of each aspect may be combined with those of any other aspect or embodiment. Any of the methods may be performed using a system according to another aspect. Any of the system aspects or embodiments may be configured to perform any of the methods.

[0050] BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Embodiments of the invention shall now be described in detail by way of example and with reference to the accompanying drawings in which:

[0052] Figure 1 shows a schematic view of a system for detecting placental contractions according to some embodiments / examples.

[0053] Figure 2 shows a schematic view of a system for detecting placental contractions according to some embodiments / examples.

[0054] Figure 3 shows a schematic cross-sectional view of a flexible substrate according to some embodiments / examples.

[0055] Figure 4 shows a schematic view of a sensor arrangement according to some embodiments / examples.

[0056] Figure 5 shows a schematic view of a flexible support element according to some embodiments / examples.

[0057] Figure 6 shows a schematic view of components of a system for detecting placental contractions according to some embodiments / examples.

[0058] Figure 7 shows a schematic view of a method for detecting placental contractions according to some embodiments / examples.

[0059] Figure 8 shows wavelength shift signals indicative of strain, obtained from FBG strain sensors.

[0060] Figure 9 shows an average wavelength shift (or average strain signal) determined from the signals of Figure 8.

[0061] Figure 10 shows a smoothed average strain signal determined from the data shown in Figure 9 along with the average strain signal.

[0062] Figure 11 shows the positions of troughs in the average strain signal, the troughs determined from the smoothed average strain signal .

[0063] Figure 12 shows positions of troughs in the average strain signal over a shorter time period than is shown in Figure 11.

[0064] Figure 13 shows the average strain signal of Figure 9 with an envelope determined by smoothing the positions of the troughs shown in Figure 12.

[0065] Figure 14 shows the envelope over a shorter time period than is shown in Figure 13. Figure 15 shows a comparison between the envelope shown in Figures 13 and 14 and a percentage change in placental volume determined by MRI.

[0066] Figure 16 shows wavelength shift over time and an electrical signal with placental volume changes indicated.

[0067] Figure 17 shows wavelength shift over time, as measured with first and second fibre Bragg gratings, and heart rate variability with a placental volume change indicated.

[0068] Figure 18 shows wavelength shift over time and acceleration over time along x -, y-, and z-axes.

[0069] Figure 19 shows first and second data sets of duration of placental volume change in relation to pregnancy outcome.

[0070] Figure 20 shows mean wavelength shift as measured by first and second fibre Bragg gratings with first and second groups of three heartbeats indicated.

[0071] DETAILED DESCRIPTION

[0072] Figure 1 schematically illustrates a system 100 for detecting placental contractions, the system 100 comprising: flexible support element 10 and strain sensor 20.

[0073] The flexible support element 10 is wearable on the abdomen of a pregnant subject. The system 100 also comprises one or more strain sensors 20, supported by the flexible support element. The one or more strain sensors 20 are arranged to detect strain in the skin of the abdomen of the subject indicative of a placental contraction. The one or more strain sensors 20 may also be arranged to detect strain in the skin of the abdomen of the subject indicative of maternal heart rate and / or fetal heart rate.

[0074] Figure 2 shows a schematic view of a further example system 100 for detecting placental contractions, the system 100 comprising: flexible support element 10, strain sensors 21 , optical fibre 22, optical interrogator 30, electrode arrangement 50, electrode 51 , electrical readout circuit 60, processor 70, wireless transmitter 80, and remote processor 90. The strain sensors 21 may also be arranged to detect strain in the skin of the abdomen of the subject indicative of maternal heart rate and / or fetal heart rate.

[0075] The strain sensors 21 in this example are optical fibre strain sensors 21 , integrated with optical fibre 22. The depiction of the optical fibre strain sensors 21 is indicative only and not to scale - it will be understood that the strain sensors may be very small and would be difficult to see in a diagram if they were drawn to scale. The optical fibre strain sensors 21 are configured to record abdominal movements. The abdominal movement detected by the optical fibre strain sensors 21 may arise as a consequence of placental contractions but may also result from other abdominal movement such as fetal kicks, rolls, uterine and Braxton Hicks contractions. The optical fibre strain sensors 21 may also be configured to record periodic changes in strain in the skin of the abdomen of the subject which can then be interpreted to obtain measurements of maternal heart rate and / or fetal heart rate.

[0076] The (or each) optical fibre strain sensor 21 may comprise a fibre Bragg grating (FBG) strain sensor. A FBG is a microstructure in the core of an optical fibre defining a spatially periodic variation in refractive index. The microstructure can be written directly, for example using a UV laser beam or femtosecond laser beam. The periodic structure has a maximum reflectivity corresponding with a Bragg wavelength, defined by the period of the periodic structure. As the spatial period changes with strain, the Bragg wavelength of the FBG changes with strain. This type of strain sensor can provide high fidelity measurements of strain. In addition optical fibres are thin, (e.g. with a diameter of 250pm or less), is flexible and conforms to the curvature of the abdomen.

[0077] In addition to being sensitive to strain, the Bragg wavelength may also vary with temperature. A change in temperature may change a refractive index of at least part of the microstructure and may also change the spatial period (due to thermal expansion / contraction). A reference FBG may be provided for compensation of temperature related effects. The reference FBG may be encapsulated in a rigid sleeve (e.g. bonded to a metal sleeve using adhesive) to prevent strain being coupled to the FBG so that any changes in the Bragg wavelength are temperature dependent. The reference FBG may be in close proximity to the strain sensors (i.e. in a representative thermal environment - in the same sort of thermal contact with the subject).

[0078] The fibre optic technology is non-invasive, passive, chemically inert, non-toxic and immune to electromagnetic interference. It is safe to use and poses negligible risk to both mother and fetus.

[0079] Alternatively, or in addition to one or more FBG strain sensor, at least some strain sensors 21 may be configured as: • Intensity-based, where bends in the optical fibre due to movement modulate a transmitted intensity due to bending of the fibre. This type of strain sensor is responsive to bending, rather than to longitudinal strain in the fibre .

[0080] • Interferometric based sensors. Such sensors comprise a reference optical path that does not change length in response to strain in the skin on the abdomen (e.g. a coiled length of fibre housed adjacent to the interrogator), and an active optical path comprising an optical fibre coupled to the skin of the abdomen . Strain in the active optical path of the interferometer changes a path length of the active optical path. The reflected optical signal from the reference optical path and the active optical path are combined at a detector, where the difference in optical phase result in intensity modulation with changes in relative path length between the reference and active paths. In the case of a Michelson interferometer, the reference and active optical path may be provided with a distal reflector. In the case of a Mach Zehnder interferometer such a reflector may be dispensed with (and the light re-combined without two passes through the reference and active optical path).

[0081] • Fabry-Perot interferometric based sensors, where strain over the sensing region leads to a change of the cavity length and hence a modulation of the detected optical signal. Air cavity based Fabry-Perot cavities may be formed in an optical fibre by splicing together two single mode fibres following pre- treatment / etching of the cores of the optical fibres.

[0082] The optical interrogator 30 is supported by the flexible support element 10 and configured to detect signal data indicating strain from the one or more strain sensors 21. The optical interrogator comprises a light source and detector and is capable of detecting a spectrum of reflectance peaks from each of a plurality of FBG strain sensors on an optical fibre connected to the optical interrogator 30. An example of a portable FBG interrogator is the FiSpec FBG X100, which is capable of reading out up to 30 FBG sensors, over a wavelength range of 8O8 -88Onm. Other similar FBG interrogators are available, and the general trend is toward smaller devices with better performance. It is therefore feasible for the system to remain wearable and portable even when it comprises the optical interrogator 30. The optical interrogator 30 may be a wireless, battery operated, monitoring unit that records the acquired signals. The optical interrogator 30 is advantageously of a miniaturised size and with a light weight. By way of example, the interrogator 30 may be approximately 75 x 50 x 15 mm and weigh approximately 60 grams. Thereby, the interrogator 30 is easily supported by the flexible support element 10.

[0083] In some embodiments the optical interrogator 30 may be arranged with more than one channel (e.g. two channels), so that more than one fibre can be interrogated. Thereby it is possible to have two optical fibre strain sensor arrangements (comprising FBG sensors 21 , for example) placed at different locations of the abdomen to cover the full region of interest. In other embodiments, a single optical fibre may be integrated into more than one patch. For example, there may be a region of optical fibre comprising at least 3 strain sensors embedded in a horizontal patch and a region of optical fibre comprising at least 3 strain sensors embedded in a vertical patch.

[0084] The processor 70 may be supported by the flexible support element 10 (for example in a pocket of the flexible support element). The processor 70 is configured to receive signal data indicative of strain from the optical interrogator 30 and to receive electrical signal data from the electrode readout circuit 60, and may be configured to pre-process the data and format it for wireless transmission.

[0085] The processor 70 may be configured to identify a placental contraction in the subject from the signal data. The processor 70 may also be configured to determine maternal heart rate and / or fetal heart rate from the signal data. Alternatively, a remote processor 90 may (e.g. in a local computer, or an off-site server, or in a cloud based computing resource) be configured to receive signal data wirelessly (via the wireless transmitter 80).

[0086] Signal processing algorithms may be used (in either the local processor 70 or remote processor 90), which classify different types of abdominal activity based on the acquired signals. The types of abdominal activity may include placental contractions, uterine contractions, Braxton Hicks contractions, and fetal movement, including fetal rolls.

[0087] The wireless transmitter 80 is configured to transmit (i) signal data comprising an indication of the strain detected by the one or more strain sensors to a remote processor 90; and / or (ii) an indication of the presence or absence of placental contractions ; and / or (iii) electrical data, either raw or processed, to indicate fetal and maternal wellbeing. The electrode arrangement 50 comprises one or more electrodes 51. The electrical readout circuit 60 may comprise instrumentation amplifiers for reading out any electrodes of the electrode arrangement 50 and converting the signals to a digital format for processing (by the processor 70 / 90).

[0088] Figure 3 shows a schematic cross-sectional view of a flexible substrate 40 which may be used in system 100. Figure 3 illustrates: flexible substrate 40, adhesive layer 41 , optical fibre 43, and optical fibre sensors 21 (e.g. intrinsic FBG sensors written to the fibre 43).

[0089] The strain sensors 21 (e.g. FBG sensors) are embedded in the flexible substrate 40, and preferably may be embedded in the adhesive layer 41 to enhance coupling with the skin and reliably secure the strain sensors 21 . The flexible substrate 40 may comprise any suitable material, such as silicone, and is preferably relatively flexible, so that it does not impose a significant constraint on the expansion and contraction of the skin to which it is adhered. The at least one flexible substrate consists of biocompatible materials so that it is suitable for sustained contact with the abdomen of a subject. An adhesive or tacky layer 41 may be provided to improve the coupling of the flexible substrate 40 with the skin (e.g. using a suitable biocompatible skin adhesive) . The adhesive layer 41 may provide a temporary yet secure and effective attachment to the skin.

[0090] The patch may further comprise a fabric layer not shown, through which the optical fibre may be threaded or stitched. The fabric layer may help to secure the optical fibre and make sure the fibre is horizontal during patch making.

[0091] The flexible substrate 40 may be an elongate patch. The elongate patch is longer than it is wide. In some examples the elongate patch 40 may be at least three times longer than it is wide. In other examples the elongate patch 40 may be at least five, or ten times longer than it is wide. In some examples, the elongate patch 40 is between 2 and 3 cm wide and between 15 cm and 30 cm long. The elongate patch 40 is preferably thin, with a thickness of less than 5 mm, preferably less than 3 mm.

[0092] In some examples, there may be a plurality of elongate patches, each comprising an embedded optical fibre with multiple FBG sensors disposed over the elongate patch. A single optical fibre may be embedded in more than one elongate patch. A configuration with multiple discrete patches may provide for a good fit for mothers of different BMI and gestational age.

[0093] In some embodiments, a patch may be adhered to a fabric wearable garment. The wearable fabric garment may be configured to encircle the abdomen of the subject. For example, one or more patches may be provided on an interior surface of a belly -band. Optical fibres may be threaded through the wearable garment, interconnecting the one or more patches and the interrogator.

[0094] More generally, each flexible substrate 40 is supported in contact with the skin in a predetermined layout by the flexible support element 10 (e.g. wearable fabric garment). Such a predetermined layout may be based on the region of interest of a particular case, to optimise the detection capacity of the strain sensors.

[0095] Figure 4 shows a schematic view of a sensor arrangement of the system 100. Figure 4 illustrates: elongate patch 42, optical fibre 22, FBG strain sensors 21a-e, and interrogator 30. The FBG strain sensor 21a is a reference FBG, for temperature compensation (as described above). Again, the position of the FBG strain sensors 21a-e are schematically illustrated by boxes - it will be understood that the FBG is a periodic variation in refractive in refractive index of the fibre core which will not necessarily be obvious from visual inspection of the optical fibre 22.

[0096] Figure 5 shows a schematic view of a flexible support element 10 according to some examples, the flexible support element 10 comprising: pocket 11, and textile or fabric material 15. The flexible support element in this example is configured as wearable garment such as a pregnancy support belt / belly band. The wearable garment may comprise a band that encircles the abdomen of the subject . The optical interrogator 30 may be disposed within the pocket 11 , along with a power source (e.g. battery) and processor 70.

[0097] The flexible support element 10 may comprise a textile or fabric material 15. The flexible substrate 10, for example in the form of a patch 42, may be incorporated into the wearable garment 16. In some examples the patch 42 may be permanently bonded to the wearable garment 16. In other arrangements the patch may be temporarily adhered to the wearable garment 16, using an adhesive on the side facing away from the skin. In this way a wearable garment may be re-used with multiple different patches. The integration of patches 42 with the wearable garment 16 may be arranged to facilitate putting on and taking off the wearable garment without requiring new patches.

[0098] The optical interrogator 30 may also be arranged either attached to the wearable garment 16, or within a pocket 11 arranged in or on the wearable garment 16. Such an arrangement may improve the possibility for continuous monitoring of placental contractions, as the presence of the strain sensor 20 and the interrogator 30 does not inconvenience the wearer. Such an arrangement provides a system 100 for detecting placental contractions that can be worn comfortably by the mother during sleep. The wearable garment 16 may include additional padding for increased comfort. Being able to provide a wearable and comfortable system 100 for detecting placental contractions is also beneficial in a clinical trial setting, as it decreases the risk of test subjects choosing to withdraw from the trial.

[0099] The interrogator 30 may be attached to the flexible support element 10 using any suitable restraints or fastening means, such as straps, cables, hook-and-loop type fasteners, and / or magnetic arrangements. In some examples, the optical interrogator 30 is arranged in a pouch or pocket 11 of the flexible support element 10.

[0100] Figure 6 shows a schematic view of components of the system 100 for detecting placental contractions. Figure 6 illustrates: horizontal elongate patch 42a, vertical elongate patch 42b, optical fibre 43, electrodes 51 , vertical axis V, and horizontal axis H.

[0101] The patches 42a, 42b and electrodes 51 may each be attached on an inner surface of a wearable garment 16.

[0102] The horizontal elongate patch 42a has a plurality of embedded strain sensors arranged to detect a horizontal component of strain in the skin on the abdomen. The vertical elongate patch 42b has a plurality of embedded strain sensors arranged to detect a vertical component of strain in the skin on the abdomen. The horizontal patch 42a is arranged above the navel, in a position that may correspond approximately with a location of the placenta. The vertical patch extends from near to (e.g. within 5cm) to the horizontal patch to a position close to (e.g. within 5 cm) the pubis. The vertical patch may detect fetal movements because it may overlap with the position of the fetus’ limbs. The configuration of the patches has been found to be suitable for detection of placental contractions. There may be at least three strain sensors embedded in each of the patches 42a, 42b (for example, four or five or six or eight or ten).

[0103] The directions of placement of the patches 42a, 42b and the strain sensors 20 conform with the abdomen and may thus vary somewhat based on the physiology of the subject wearing the strain sensor. As shown in Figure 6, the horizontal direction H may also be described as stretching substantially along a frontal axis of the subject, essentially in a direction stretching across the abdomen from the right-hand side to the left-hand side at a substantially constant height from the feet of the subject. The vertical direction V may also be described as stretching substantially along the craniocaudal direction, essentially in a direction from the feet to the head of the subject. In other words the vertical direction V may be expressed as an up -down direction of the abdomen, and the horizontal direction H may be expressed a left -right direction across the abdomen.

[0104] A single optical fibre 43 may be embedded in both of the patches 42a, 42b. In some embodiments a different optical fibre may be embedded in each of the patches 42a, 42b. In alternative embodiments, a single patch 42 can replace the perpendicularly arranged horizontal and vertical patches 42.

[0105] In some examples, the system 100 may comprise at least one electrode arrangement 50 comprising at least one electrode 51 arranged to be held in contact with the skin by the flexible support element 10. Electrical measurements obtained from the electrode arrangement 50 may be indicative of, for example, uterine contractions, Braxton-Hicks contractions, maternal and fetal heart rate, and fetal and maternal movements.

[0106] The at least one electrode arrangement 50 may comprise a plurality of electrodes 51 spaced apart by at least 3cm. The at least one electrode arrangement 50 may comprise a first electrode 51a positioned substantially on the midline (e.g. within 3cm of the midline) of the subject on the abdomen, spaced apart from and below the navel by at least 3 cm. One or more further electrodes 51b, 51c may be spaced apart from the first electrode 51a, in a line on an angle of between 45 and 60 degrees from the vertical . Additional channels / electrodes can be added to detect electrophysiological signals on other parts of the abdomen in order to increase FECG detection for example in early gestation or different fetal presentations . The system 100 may comprise an electrical readout circuit 60 supported by the flexible support element 10 and configured to detect electrical signals from the at least one electrode arrangement 50. The combination of electrical and optical measurements, such as with FBGs, may allow separation of minor movements and full body movements, and assist with other identification of both false positives and false negatives placental contraction. Also, by providing electrical data the system can determine fetal and maternal heart rate and variability changes (LTV, STV, RMSSD, PNN etc) before, during and after a placental contraction. Such information allows fetal and maternal sympathetic and parasympathetic nervous feedback mechanisms to be assessed.

[0107] Furthermore, a combination of optical fibre sensing, electrical and other optical measurements improve the ability of the system to discriminate placental contractions from other maternal movements, such as fetal kicks, fetal rolls, uterine and Braxton- Hicks contractions. In addition, a combination of placental contractions and other monitoring is useful to improve the understanding of the health of the fetus. For example, in addition to contraction monitoring, the sensors may be used to monitor the physiology of the placenta, such as oxygen saturation during contractions. In combination with fetal and maternal heart rate beat to beat monitoring, the effect of placental contractions on the fetus can be better understood.

[0108] Figure 7 shows a schematic view of a method for detecting placental contractions according to some embodiments / examples. Steps of the method may be taken in any suitable order. The method shown in Figure 7 may be performed using the system 100 as disclosed herein.

[0109] As illustrated in Figure 7, the method comprises arranging slOlO at least one strain sensor on the abdomen of a pregnant subject. The strain sensor may be the strain sensor 20, 21 of the system 100. Thereafter, the method comprises obtaining sl020 signal data from the strain sensor arranged on the abdomen of a pregnant subject. When the signal data has been obtained, the method comprises processing sl030 the signal data to determine whether the signal data indicates a placental contraction. The signal processing is performed to derive signals indicating placental contraction from the strain sensor signals obtained from the at least one strain sensor 20. Examples of suitable signal processing methods will be described below.

[0110] In some examples, the signal processing may comprise obtaining a normalised respiration signal and removing the normalised respiration signal from the obtained sensor signal, using any suitable method. Thereby, indications of a placental contraction may be observed from the processed signal.

[0111] The processor 70 or 90 described in the example embodiments is configured to receive signal data comprising an indication of the strain detected by the one or more strain sensors 20 and to identify a placental contraction in the subject from the signal data. An example of some steps comprising in processing the signal data to identify placental contractions are explained below with reference to Figures 8 to 17. The example steps described below may be performed at a processor on the flexible support element, or a remote processor on a local computer (in the same room), a server (not in the same room) or a cloud based computing resource or virtual computer hosted on any suitable computing medium.

[0112] The signal data may be processed by importing raw FBG wavelength data and performing any pre-processing steps to format the data for further processing. Preprocessing in this example comprises converting the signal data from raw wavelength to ‘wavelength shift’ . The signal data in this example comprises a matrix in which each row comprises Bragg wavelengths for a specific FBG. A wavelength shift value can be determined for each FBG by subtracting the first value in each row of the matrix from the subsequent values in that row.

[0113] Figure 8 shows Bragg wavelength shifts for four active strain sensors (FBG 1.1 to 1.4) and one Bragg wavelength FBG signal from a reference FBG (FBG 1.0) that is not subject to strain (but which could be used as a temperature reference, for example). Having determined wavelength shift values, an average wavelength shift from the active FBG sensors can be used in the following data processing steps. Regardless of the method used, it is the change in the wavelength that relates to strain, not the raw wavelength as such. Figure 9 shows the average wavelength shift values obtained by finding the mean value of the wavelength shift data for FBGs 1.1 to 1.4 shown in Figure 8 - representing a mean strain measured by the active strain sensors. In some approaches a mean strain value may not be used - instead it may be appropriate to use a weighted sum or some other combination of the strain signals (e.g. obtained from a neural network) or simply a single strain reading (e.g. from a single FBG).

[0114] The average strain signal that is shown in Figure 9 includes changes in strain that are the result of the subject breathing (at a frequency typically between 10 and 30 breaths per minute, depending on whether the subject is at rest or working). To identify placental contraction and maternal heart rate and / or fetal heart rate, it is desirable to remove the modulation of the strain signal that results from breathing (which may be referred to as a breathing signal). There are a number of ways to do this. In some embodiments, the breathing signal may be identified by spectral analysis of the average strain signal (e.g. by performing a fast Fourier transfer, FFT). A maximum power spectral density in the FFT (optionally, over a range of frequencies associated with breathing) may be identified as corresponding with the breathing signal, and a filter may be used to remove the component identified by the FFT. A low pass filter with a predetermined cutoff may be used to remove the breathing signal, and / or a moving average with a predetermined window size, or any other suitable method (e.g. Hilbert Transform etc).

[0115] In an example embodiment, removing the breathing signal comprises finding troughs in the average strain signal. The position of the abdomen tends to return to a more consistent position after breathing out and the depth of breathing in tends to vary more, so fitting a curve to the troughs of the breathing signal may be a particularly effective way to remove the breathing signal. The fit to the troughs of the average strain signal may be termed the envelope. Alternatively, the peaks of the breathing can be deployed in the same manner.

[0116] In order to find the envelope, the average strain signal shown in Figure 9 may be smoothed. This is shown in Figure 10, in which smoothed average strain and average strain are shown (over a relatively short timescale of 16 seconds). The troughs can be found in the smoothed average strain by taking the numerical derivative of the smoothed average strain, and locating where the derivative crosses from a negative value to a positive value. This denotes a turning point in the data associated with a trough in the smoothed average strain. After locating the troughs, the locations of the troughs may be stored. Figure 11 plots the average strain as a line with crosses marking the positions (in time) of the troughs of the smoothed average strain signal. Figure 12 shows the locations of the identified troughs (as crosses) obtained from the smoothed average strain on the average strain signal.

[0117] An envelope may be created from the stored troughs. Due to possible issues with changes in strain due to maternal movement (such as wriggling), it may be advantageous to smooth the stored troughs to determine the envelope. Any appropriate filter or smoothing function may be used to do this. In this example smoothed median filter is used to determine the envelope from the stored troughs. A median filter of three points may be used (i.e. data points 1 , 2, 3 then 2, 3, 4 then 3, 4, 5 etc). The median will thereby be as close to the troughs as possible whilst still mitigating motion artefacts.

[0118] Figures 13 and 14 show the envelope with the average strain signal at different time scales.

[0119] The above steps represent an example of how an envelope signal can be obtained, from which placental contractions may be identified. Figure 15 shows a comparison between the envelope 501 determined according to the steps set out above and a percentage change in placental volume 502 determined from MRI. An MRI image 503 is shown at the time indicated by the vertical line on the graphs. There is a striking correlation between the envelope signal and the placental volume determined by MRI.

[0120] Any suitable technique may be employed to determine maternal heart rate and / or fetal heart rate from the signal data discussed above. Examples of techniques which may be employed to determine maternal heart rate and / or fetal heart rate include peak detection, frequency analysis, such as by performing a FFT, autocorrelation, and adaptive filtering.

[0121] Figure 16 shows a wavelength shift 161 over time as measured with a fibre Bragg grating arranged on the abdomen of a pregnant subject and an electrical signal 162 over time as measured with electrodes arranged on the abdomen of the pregnant subject. Figure 16 further shows volume changes 163a-e of the placenta. Figure 16 illustrates changes in wavelength shift and changes in electrical signal which correspond to volume changes of the placenta of the pregnant subject.

[0122] Figure 17 shows a first wavelength shift 171 over time as measured with a first fibre Bragg grating arranged on the abdomen of a pregnant subject and orientated in an approximately horizontal direction as described above , a second wavelength shift 172 over time as measured with a second fibre Bragg grating arranged on the abdomen of the pregnant subject and orientated in an approximately vertical direction as described above, and a heart rate variability 173 of the pregnant subject over time as an RMSSD obtained from an electrical signal as measured with electrodes arranged on the abdomen of the pregnant subject. Figure 17 further shows a volume change 174 of the placenta of the pregnant subject. Figure 17 illustrates a large change in the RMSSD during the volume change 174 of the placenta.

[0123] Figure 18 shows a wavelength shift 181 over time as measured with a fibre Bragg grating arranged on the abdomen of a pregnant subject, acceleration 182 over time along an x- axis as measured by a first accelerometer arranged on the abdomen of the pregnant subject, acceleration 183 over time along a y-axis as measured by a second accelerometer arranged on the abdomen of the pregnant subject, and acceleration 184 over time along a z-axis as measured by a third accelerometer arranged on the abdomen of the pregnant subject. In this examples, the x-axis extends in a horizontal direction as described above, i.e., left-right across the abdomen, the y-axis extends in a vertical direction as described above, i.e., up-down over the abdomen, and the z-axis is perpendicular to both the x-axis and the y-axis.

[0124] Figure 18 shows how changes in acceleration 182, 183, 184, representative of movements of the pregnant subject, affect the wavelength shift 181. This information may be used in methods and systems described herein to compensate for motion artefacts in strain data or to reject stain data influenced by motion artefacts. Systems described herein may comprise at least one accelerometer as required.

[0125] Figure 19 shows a first data set 191 of duration of placental volume change (placental contraction) and a second data set 192 of placental volume change (placental contraction). Each point in the first data set 191 is representative of a ‘not near miss’ pregnancy outcome and each point in the second data set 192 is representative of a ‘near miss’ pregnancy outcome. A ‘near miss’ pregnancy outcome is defined at least one of: (i) severe fetal growth restriction; (ii) hypoxia at birth; (iii) symptoms of pre-eclampsia; (iv) placental abruption; (v) abnormal fetal or umbilical doppler results ; and (vi) serious findings on placental histology. Figure 19 shows a statistically significant difference (P = 0.011), as obtained using a Mann Whitney-U test, between the distribution of the first data set 191 (‘not near miss’) and the second data set 192 (‘near miss’ ). Figure 19 supports a relationship between duration of placental volume change (placental contraction) and abnormalities in fetal development or in the development of the placenta.

[0126] Figure 20 shows a first mean wavelength shift 201 over time as measured with a first fibre Bragg grating arranged on the abdomen of a pregnant subject in an approximately horizontal direction as described above and a second mean wavelength shift 202 over time as measured with a second fibre Bragg grating arranged on the abdomen of the pregnant subject in an approximately vertical direction as described above . Figure 20 also shows a first group of three maternal heartbeats 203a-c and a second group of three maternal heartbeats 204a-c. The peaks before the first group of three heartbeats 203a- c, between the first group of three heartbeats 203a-c and the second group of three heartbeats 204a-c, and after the second group of three heartbeats 204a -c represent a respiration signal.

[0127] Figure 20 demonstrates how individual maternal heartbeats can be recognised from FBG data and therefore how maternal heart rate can be determined from FBG data, i.e., by counting the number of individual heartbeats occurring per minute. It will be appreciated that the same approach can be applied to determine fetal heart rate. Furthermore, an RMSSD can be obtained from data as illustrated in Figure 20, for example to determined maternal or fetal heart rate variability. Systems and methods described herein may use the technique represented by Figure 20 to determine maternal and / or fetal heart rate and / or heart rate variability.

[0128] It is believed that monitoring changes in placental volume and / or identifying placental contractions will provide information that is useful for the identification and treatment of placental and fetal pathologies. For example, a deviation from an expected range of frequency of placental contractions may indicate a pathology. Similarly, a deviation in strength of placental contractions from an expected range may indicate a pathology. It is not essential that data obtained from abdominal strain sensors is classified (in a binary sense) as corresponding with a placental contraction. In some embodiments an envelope (or similar signal) could be processed (e.g. using an FFT or similar) to identify whether a normal pattern of movement is present that can be expected to be associated with typical placental contractions. For example, a wavelet decomposition may be performed based on a wavelet that is representative of a typical placental contraction to identify a placental contraction signal. Alternatively, a Fourier (or wavelet) decomposition may be used to identify frequency components of a signal that are likely to be associated with placental contractions. A normal or healthy range for a power spectral density of contraction signals within a particular frequency range may indicate “acceptable” placental / fetal health, with deviations from this pattern indicative of a pathology that may require medical intervention or further monitoring.

[0129] In some examples, a machine learning algorithm may be used to obtain information for classifying the subject as having a potential placental / fetal pathology. In some embodiments a machine learning algorithm may have been trained to classify signal data as indicating, or not indicating, a placental contraction. The machine learning algorithm may comprise an artificial neural network that has been trained using supervised learning from a training dataset derived from MRI data in which placental contractions have been identified.

[0130] The machine learning algorithm may be trained using a labelled dataset in which events causing strain in the skin of the abdomen which are not placental contractions are identified, for example derived from MRI data. The machine learning algorithm may be configured to classify signal data that indicates one or more of: fetal movement, breathing, and / or uterine contractions (including Braxton Hicks contractions). The machine learning model may be implemented essentially in real time (for example with a latency and update rate of 10 seconds or less, or 5 seconds or less , or 2 seconds or less). The processor may thereby rapidly indicate when a placental contraction occurs, such that the indication can be conveyed to a monitoring unit. The monitoring unit may monitor a frequency and / or strength of placental contractions over an extended period of time (e.g. hours or days) so that any potential pathologies can be detected. In other examples, the machine learning algorithm may be trained to determine a strength and frequency of placental contractions from the data (potentially without an intermediate step of classifying regions of the signal data explicitly as corresponding with placental contractions).

[0131] In any example, the signal data may comprise signals obtained from strain sensors disposed on the abdomen of the subject.

[0132] In some examples, the system may comprise a wireless transmitter 80 configured to transmit (i) signal data comprising an indication of the strain detected by the one or more strain sensors 20 to a remote device 90; and / or (ii) an indication of the presence or absence of placental contractions ; and / or (iii) electrical data, either raw or processed, to indicate fetal and maternal wellbeing.

[0133] From reading the present disclosure, other variations and modifications will be apparent to the skilled person. Such variations and modifications may involve equivalent and other features which are already known in the art and which may be used instead of, or in addition to, features already described herein.

[0134] Although the appended claims are directed to particular combinations of features, it should be understood that the scope of the disclosure of the present invention also includes any novel feature or any novel combination of features disclosed herein either explicitly or implicitly or any generalisation thereof, whether or not it relates to the same invention as presently claimed in any claim and whether or not it mitigates any or all of the same technical problems as does the present invention.

[0135] Reference may be made herein to the spatial relationships between various components and to the spatial orientation of various aspects of components as the devices are depicted in the attached drawings. However, as will be recognized by those skilled in the art after a complete reading of the present disclosure, the devices, members, apparatuses, etc. described herein may be positioned in any desired orientation. Thus, the use of terms such as “above,” “below,” “upper,” “lower,” or other like terms to describe a spatial relationship between various components or to describe the spatial orientation of aspects of such components should be understood to describe a relative relationship between the components or a spatial orientation of aspects of such components, respectively, as the device described herein may be oriented in any desired direction.

[0136] Features which are described in the context of separate embodiments may also be provided in combination in a single embodiment. Conversely, various features which are, for brevity, described in the context of a single embodiment, may also be provided separately or in any suitable sub -combination. The applicant hereby gives notice that new claims may be formulated to such features and / or combinations of such features during the prosecution of the present application or of any further application derived therefrom. For the sake of completeness, it is also stated that the term "comprising" does not exclude other elements or steps, the term "a" or "an" does not exclude a plurality, a single processor or other unit may fulfil the functions of several means recited in the claims and reference signs in the claims shall not be construed as limiting the scope of the claims.

Claims

CLAIMS1. A system for detecting placental contractions, comprising: a flexible support element that is wearable on the abdomen of a pregnant subject; one or more strain sensors supported by the flexible support element, arranged to detect strain in the skin of the abdomen of the subject indicative of a placental contraction.

2. The system of claim 1 , wherein the one or more strain sensors comprise an optical fibre strain sensor.

3. The system of claim 2, wherein the optical fibre strain sensor comprises a fibre Bragg grating strain sensor.

4. The system of claim 2 or 3, further comprising an optical interrogator supported by the flexible support element and configured to detect signal data indicating strain from the one or more strain sensors.

5. The system of any preceding claim, wherein the one or more strain sensor is embedded in at least one flexible substrate, and each flexible substrate comprising an adhesive layer for attaching the flexible substrate to the skin.

6. The system of claim 5, wherein each of the at least one flexible substrate comprises an embedded optical fibre.

7. The system of claim 5 or 6, wherein each flexible substrate is supported in contact with the skin in a predetermined layout by the flexible support element.

8. The system of claim 7, wherein the flexible support element comprises a textile or fabric material.

9. The system of claim 8, wherein the flexible support element comprises a pregnancy support belt or belly band, arranged to wrap around the abdomen.

10. The system of any preceding claim, wherein the at least one strain sensor comprises sensors that are oriented in different directions.

11. The system of claim 10, wherein the at least one strain sensor comprises a strain sensor that is oriented approximately along the craniocaudal direction and a strain sensor that is oriented approximately along the frontal axis.

12. The system of any preceding claim, further comprising at least one electrode arranged to be held in contact with the skin by the flexible support element.

13. The system of claim 12, wherein the at least one electrode comprises a plurality of electrodes spaced apart by at least 3 cm.

14. The system of claim 14, wherein the at least one electrode comprises : an electrode positioned within 3 cm of a midline of the subject on the abdomen, spaced apart from and below the navel by at least 3 cm ; and optionally one or more further electrodes spaced apart from the first electrode in a line on an angle of between 45 and 60 degrees from the vertical .

15. The system of any of claims 12 to 14, further comprising an electrical readout circuit supported by the flexible support element and configured to detect electrical signals from the at least one electrode.

16. The system of any preceding claim, further comprising a processor configured to receive signal data comprising an indication of the strain detected by the one or more strain sensors, the processor configured to identify a placental contraction in the subject from the signal data.

17. The system of claim 16, wherein the at least one strain sensor comprises a plurality of strain sensors, and the processor is configured to combine the strain detected at each strain sensor to determine an average strain signal.

18. The system of claim 16 or 17 wherein identifying a placental contraction comprises determining a strain envelope by finding troughs in a strain signal.

19. The system of claim 18, wherein the processor is configured to smooth or low pass filter the strain envelope.

20. The system of any of claims 17 to 19, wherein the processor is configured to implement a machine learning algorithm that has been trained to classify signal data as indicating, or not indicating, a placental contraction.

21. The system of claim 20, wherein the machine learning algorithm comprises an artificial neural network that has been trained using supervised learning from a training dataset derived from MRI data in which placental contractions have been measured.

22. The system of any of claims 12 to 15 including the features of any of claims 16 to 21 , wherein the signal data comprises the electrical signals .

23. The system of claim 22, wherein the processor is configured to use the electrical signals to remove false positive placental contractions and / or to determine physiological parameters of the subject and / or of the fetus that are not placental contractions .

24. The system of any preceding claim, further comprising a wireless transmitter configured to transmit: i) signal data comprising an indication of the strain detected by the one or more strain sensors to a remote device; and / or ii) an indication of the presence or absence of placental contractions ; and / or (iii) electrical data either raw or processed to indicate fetal and maternal wellbeing.

25. A method of detecting placental contractions, comprising: arranging at least one strain sensor on the abdomen of a pregnant human subject; obtaining signal data from the strain sensor; processing the signal data to determine whether the signal data indicates a placental contraction.

Images