Inductive sensing system and method

Abstract

A system and method are provided for extracting component signals related to different physiological phenomena in the body from sensed induced signals. A resonator circuit 10 is oscillated at a specific frequency to generate an alternating electromagnetic field, which is applied to the body being investigated. This magnetic field induces secondary eddy currents in the body, which interact with the primary magnetic field to change at least the frequency and amplitude of the resonator circuit oscillating the current. These changes in current characteristics, particularly frequency and amplitude, are measured to provide first and second input signals. A system or method configured to receive these input signals is then provided by the system. A number of different composite or fusion signals, each formed from a different linear combination ratio of the two input signals, are then generated by the system. These are then evaluated in a signal selection procedure to identify the best candidate signals for providing a measure or indicator of one or more specific physiological phenomena. This is based on predefined selection criteria, for example, related to the signal characteristics of the candidate signals.

Description

Technical Field

[0001] The present invention relates in particular to an inductive sensing system for detecting and separating signal components indicative of different physiological phenomena.

Background Art

[0002] Generally, it is necessary to measure the mechanical operations and dynamic changes of internal body structures such as the heart, lungs, and arteries. For example, it is useful to measure the internal volume or dimensions of periodically changing heart chambers or lung spaces, or the mechanical activity of arteries, such as the change in arterial volume during the cardiac cycle.

[0003] Sensors for measuring mechanical activity are sometimes called kinographic sensors (kinographic biometric sensors in the clinical field). Examples of kinographic biometric sensors include accelerometer-based biosensors, transthoracic impedance biosensors, radar-based biosensors, capacitance sensors, and photoplethysmography (PPG) sensors.

[0004] Magnetic induction sensors may also be used as biometric sensors for sensing mechanical activity. The operating principle of inductive sensing is based on Faraday's law. The oscillating primary magnetic field is generated by a generating loop antenna, which induces eddy currents in the tissue irradiated by the signal according to Faraday's law. The eddy currents generate a secondary magnetic field. In that case, the total magnetic field is the superposition of the primary magnetic field and the secondary magnetic field. Changes induced in the electrical characteristics (antenna current) of the generating antenna can be measured and used to estimate the characteristics of the secondary electric field, and thus the stimulated tissue can be estimated.

[0005] Inductive sensing offers the possibility of simple non-contact measurement of the mechanical activity of the heart and lungs, or of blood vessels such as the radial artery in a human arm.

[0006] A major drawback of common chemographic biosensors (including inductive sensors and other types of sensors) is that it is currently very different to distinguish sensing signals derived from different physiological sources.

[0007] For example, within a single complex return signal, it is very difficult to distinguish between signal elements related to the mechanical activity of the heart (e.g., the heartbeat of the heart) and signal elements related to the mechanical activity of the lungs (e.g., breathing).

[0008] In known systems, this is often done assuming that the heart rate is greater than the respiratory rate. In this way, different signal components are identified and separated based on frequency.

Summary of the Invention

Problems to be Solved by the Invention

[0009] However, the clinically possible ranges of heart rate and pulse rate overlap. For example, the frequencies of particularly high respiratory rates overlap with those of particularly low pulse rates, and vice versa. Therefore, the perceived (high) respiratory rate may be misinterpreted by state-of-the-art systems as a low heart rate, and vice versa, leading to incorrect clinical diagnoses and interventions.

[0010] For example, in neonatal monitoring, apnea is often not detected by impedance-based measurements. This is because the heart contractions characteristic of this condition are misinterpreted as the patient's respiratory rate.

[0011] A further problem associated with known systems is that it is difficult to distinguish movement artifacts from true biophysical signals such as pulse signals or respiratory signals. Distinguishing movement artifacts from true signals is often performed by assuming that the frequency of the artifact is different from the frequency of the biometric signal and / or that the waveform of the artifact is different from the waveform of the biometric signal (e.g., in terms of shape characteristics). However, since both the frequency and waveform of signal artifacts are often similar to the frequency and waveform of biometric signals, these methods are often insufficient.

[0012] For example, if the step frequency of a patient during walking is close to the pulse rate (which is a common case), the measurement of the pulse rate becomes unreliable. Further, even when the frequencies are different, the patient's step frequency may be interpreted as the pulse frequency because, for example, the two waveforms are very similar. As a result, the heart rate is not measured correctly.

[0013] Therefore, there is a need for an improved approach to inductive sensing that can more reliably distinguish different physiological signals from each other and from signal artifacts.

[0014] WO2018 / 127488A1 discloses a magnetic induction sensing device including a loop antenna for inductively combining an electromagnetic (EM) signal emitted from a medium in response to stimulation of the medium by an electromagnetic excitation signal. In an embodiment, two antennas are each connected to associated oscillators for driving the antennas at different frequencies f1 and f2. In response to the generated electromagnetic excitation signal, the electromagnetic signals received at each of the antennas are also at respective frequencies f1 and f2, which are mixed with each other by an applied mixer and a low-pass filter configured to pass the difference frequency. Thereafter, this signal is transferred to a further signal processing element, such as a counter, for example.

Means for Solving the Problem

[0015] The present invention is defined by the claims.

[0016] According to an example according to an aspect of the present invention, there is provided a system for use in inductive sensing for processing a return electromagnetic signal from the body in response to the application of an electromagnetic excitation signal to the body, The system is configured to receive a signal input indicative of the sensed return electromagnetic signal, the return electromagnetic signal corresponding to a signal sensed by a loop antenna of a resonator circuit based on detecting a change in the electrical characteristics of the resonator circuit when the resonator circuit is driven to generate an electromagnetic excitation signal, The system is configured to perform a signal extraction procedure, where the system detects a first input signal based on the frequency of the sensed return electromagnetic signal from the sensed return electromagnetic signal, detects a second input signal based on the sensed amplitude of the sensed return electromagnetic signal from the sensed return electromagnetic signal, applies a signal generation procedure having generating a plurality of candidate signals, each candidate signal being formed from a different linear combination of the first and second input signals, applies a signal selection procedure for selecting one of the candidate signals, the signal selection procedure being based on a predefined criterion related to one or more signal characteristics of the input signals, the criterion being configured to separate signals related to a particular physiological source of the body, and the selected signal being configured to form an output signal.

[0017] The system may be configured to communicate signals, for example, in use with an inductive sensing device including a resonator circuit including a loop antenna, the loop antenna being configured to be driven to generate an electromagnetic excitation signal based on detecting a change in the electrical characteristics of the resonator circuit and including signal sensing means for sensing a return electromagnetic signal from the body.

[0018] The system may include, for example, a processing unit or controller configured to receive a signal input indicative of the sensed return electromagnetic signal and perform a signal extraction procedure.

[0019] Embodiments of the present invention provide an approach for extracting signal components from different physiological sources from measured induction signals. This is based on collecting at least two measured input signals, the first based on a change in the frequency of a resonator circuit and the second based on a change in amplitude. The relative amounts in which a given physiological signal is present in each of the first and second signals typically differ. Thus, by combining (or fusing) the two input signals in particularly different ratios, it is possible to reach an output signal in which undesired physiological signal components are suppressed and desired signal components are enhanced or strengthened.

[0020] However, such a simple signal combination approach depends on knowing the exact ratio for combining the input signals to filter out undesired components, or results in a plurality of possible output signals, in which case it is unclear which is to be used for further analysis.

[0021] Accordingly, embodiments according to the present invention instead propose generating a plurality of candidate signals formed from different combination ratios of the input signals and then applying a further signal selection step for selecting the best candidate signal for a particular physiological phenomenon. This can be based on predetermined signal characteristics. In this way, a combination of signal fusion and subsequent signal selection based on criteria specific to the physiological phenomenon in question enables the extraction of signal components relating to a particular physiological source in the body.

[0022] As described above, the system is configured to select from among the candidate signals based on an analysis of signal characteristics. In some examples, this can have a scoring procedure that scores the candidate signals according to a determined likelihood of representing the physiological source in question, such that the signal with the highest score is selected for further analysis.

[0023] The system is configured to receive an input corresponding to a sensed inductive sensing signal from outside the system, and the system is configured only to perform a signal extraction procedure. The system includes, for example, a processor or a controller unit for this purpose.

[0024] In other embodiments, the system further includes an inductive sensing device for acquiring an inductive sensing signal.

[0025] In particular, according to one or more embodiments, the system further includes a resonator circuit including a loop antenna, signal generating means adapted to excite the loop antenna to generate an electromagnetic excitation signal, and signal sensing means adapted to sense a return signal from the body using the loop antenna based on detecting a change in an electrical characteristic of the resonator circuit. The system includes an inductive sensing device.

[0026] In some examples, the system includes a processor, such as a microprocessor unit, configured to control the resonator circuit, the signal generating means, and the signal sensing means to perform the signal detection, combination, and selection steps outlined above.

[0027] The first and second signals represent fluctuations or changes in frequency or amplitude over time. They represent a deviation from the starting frequency or amplitude of the resonator circuit current. They indicate a deviation from the natural (e.g., resonant) frequency and natural (e.g., resonant) amplitude of the resonator circuit. The amplitude signal is an attenuation signal indicating the decay of the current over time, i.e., the change in the natural amplitude due to the return signal. This indicates the absorption of energy in the applied excitation signal and results in a measurable change in the amplitude of the current in the resonator circuit, and is therefore referred to herein as an absorption signal.

[0028] According to one or more examples, the signal generation procedure is based on the use of the independent component analysis (ICA) method.

[0029] For example, the ICA method is used to determine the combination ratio used to generate a signal. ICA is a well-known signal analysis method and is based on the assumption that the return signal sensed by the resonator circuit antenna is a composite signal formed from a plurality of signal components each corresponding to a different physiological source.

[0030] ICA attempts to determine a weight vector matrix that describes how the underlying physiological signal components are mapped to two detected input signals. This enables the reconstruction of the original physiological signal components from a particular linear combination of the input signals.

[0031] The ICA procedure is rather resource-intensive in terms of processing resources. Thus, alternatively, according to a further set of embodiments, the signal generation procedure is based on the use of a predefined set of signal combination ratios for forming a plurality of candidate signals.

[0032] The signal combination ratios are stored in a list, for example. Alternatively, there is a predefined signal protocol that defines a regimen for combining the input signals at a series of different ratios. For example, a series of combination ratios are defined at set intervals.

[0033] Thus, this can be used in some examples instead of ICA (thereby, for example, improving the processing speed), or in some examples, used in combination with ICA to improve the accuracy of signal selection.

[0034] According to one or more embodiments, the criteria for the signal selection procedure include one or both of the frequency of the candidate signal and the maximum and minimum number of signals in a given time window.

[0035] According to one or more embodiments, the signal extraction procedure further comprises the step of generating an information output indicative of a particular physiological phenomenon based on the selected candidate signals.

[0036] According to one or more embodiments, the signal extraction procedure has an additional step of applying a band-pass filter to the input signal, prior to the signal generation procedure.

[0037] For example, the threshold or parameters of the band-pass filter are set depending on the physiological phenomenon from which the signal is extracted. This is based on, for example, predefined parameters known to be associated with different physiological phenomena. This advantageously pre-suppresses signal frequency components that are known to likely be outside the range associated with the physiological phenomenon in question.

[0038] According to one or more embodiments, the signal extraction procedure has an additional signal processing step applied immediately after detection of the input signal, and the signal processing step is configured to suppress motion artifacts in each of the input signals.

[0039] In a series of advantageous examples, the additional processing step comprises receiving an input indicative of the fundamental frequency of the motion to be suppressed, and applying a notch filter to the input signal, the notch filter having an adaptable frequency setting and being applied to the input signal at one or more multiples of the fundamental frequency.

[0040] This procedure suppresses periodic motion artifacts present in the input signal.

[0041] In some examples, the fundamental frequency is determined based on input from a motion sensor (such as an accelerometer). This may be determined by the system itself or simply received from outside the system.

[0042] In an advantageous example, the signal selection procedure is configured to select a candidate signal determined to indicate the respiration rate of the subject in at least one mode.

[0043] This is based on, for example, predetermined signal characteristics known to be associated with respiration signals.

[0044] In at least one mode, the system is configured to perform two iterations of a signal extraction procedure, and the signal selection procedure is configured to select signals related to different respective first and second physiological phenomena in the first and second executions.

[0045] In a preferred set of examples, in at least the second execution, a band-pass filter is applied to the sensed input signal prior to the signal generation procedure.

[0046] In a preferred set of examples, the signal selection procedure is configured to select a signal related to the respiratory rate of the subject in the first execution and a signal related to the heart rate of the subject in the second execution.

[0047] Examples according to a further aspect of the invention provide a method for use in inductive sensing for processing a return electromagnetic signal from the body in response to the application of an electromagnetic excitation signal to the body, the method comprising: receiving a signal input indicative of the sensed return electromagnetic signal, the return electromagnetic signal corresponding to a signal sensed by a loop antenna of a resonator circuit based on a step of detecting a change in an electrical characteristic of the resonator circuit when the resonator circuit is driven to generate an electromagnetic excitation signal; performing a signal extraction procedure, the performing of the signal extraction procedure comprising: detecting a first input signal based on the frequency of the sensed return electromagnetic signal from the sensed return electromagnetic signal; detecting a second input signal based on the sensed amplitude of the sensed return electromagnetic signal from the sensed return electromagnetic signal; applying a signal generation procedure having a step of generating a plurality of candidate signals, each candidate signal being formed from a different linear combination of the first and second input signals; Applying a signal selection procedure for selecting one of the candidate signals, the signal selection procedure being based on a predefined criterion related to one or more signal characteristics of the input signals, the criterion being configured to separate signals related to a specific physiological source of the body, and the selected signal forming an output signal.

[0048] The method is only for performing signal processing, and outside the scope of the claimed method, the generation and sensing of physical induction signals are performed independently.

[0049] However, in a further set of embodiments, the method comprises the following steps, namely, Applying an electromagnetic excitation signal to the body using a resonator circuit including a loop antenna; Sensing the return electromagnetic signal from the body using the loop antenna based on detecting a change in an electrical characteristic of the resonator circuit.

[0050] Thus, in this further set of embodiments, the method further comprises steps for performing the steps of physical induction sensing itself.

[0051] An example according to a further aspect of the invention provides a computer program product including code means configured to cause a processor to execute, when executed on the processor, a method according to any example or embodiment outlined above, or described below, or according to any claim of this application.

[0052] These and other aspects of the invention will become apparent and elucidated with reference to the embodiments described hereinafter.

[0053] For a better understanding of the present invention, and to more clearly show how the present invention is implemented, reference is made, by way of example only, to the accompanying drawings.

Brief Description of the Drawings

[0054]

Fig. 1

Fig. 2

Fig. 3

Fig. 4

Fig. 5

Fig. 6

Fig. 7

Fig. 8

Fig. 9

Fig. 10

Fig. 11

Fig. 12

DETAILED DESCRIPTION OF THE INVENTION

[0055] The present invention will be described with reference to the drawings.

[0056] The detailed description and specific examples illustrate exemplary embodiments of apparatuses, systems, and methods and are for illustrative purposes only and are not intended to limit the scope of the invention. These and other features, aspects, and advantages of the apparatuses, systems, and methods of the present invention will be better understood from the following description, the appended claims, and the accompanying drawings. It is to be understood that the figures are merely schematic and are not drawn to scale. It is also to be understood that the same reference numbers are used throughout the figures to indicate the same or similar parts.

[0057] The present invention provides a system and method for extracting component signals related to different physiological phenomena in a body from a sensed induction signal. A resonator circuit is oscillated at a specific frequency to generate an alternating electromagnetic field, and this alternating electromagnetic field is applied to the body under investigation. This magnetic field induces secondary eddy currents in the body, which interact with the primary magnetic field to change at least the frequency and amplitude of the oscillating current of the resonator circuit. These changes in the current characteristics, particularly in frequency and amplitude, are measured and provide first and second input signals. A system or method configured to receive these input signals is provided by embodiments of the present invention. Thereafter, a number of different composite or fusion signals are generated by the system, each formed from different linear combination ratios of the two input signals. These are then evaluated by a signal selection procedure to identify the best candidate signal for providing a measure or indicator of one or more specific physiological phenomena. This is based on, for example, predefined selection criteria related to the signal characteristics of the candidate signals.

[0058] Embodiments of the present invention provide a system and method for processing sensed signals acquired by an induction sensing device. In some cases, the system and method perform steps for acquiring the induction sensing signal. Thus, the embodiments generally are based on the principle of magnetic induction. First, the basic principle of magnetic induction is briefly outlined.

[0059] Inductive sensing is based on the principle of inductive combination, whereby a coil or wire has a potential difference induced across its ends when exposed to a time-varying magnetic field. Embodiments of the present invention use this principle to measure the intensity of electromagnetic signals generated within a body region by sensing changes in the inductance of a coil or loop antenna placed near the body, and these changes are detected based on changes in the resonance characteristics of the antenna or resonator circuit.

[0060] Certain embodiments of the present invention utilize a resonator including an antenna (in a preferred embodiment, including only a single-turn loop) to stimulate or excite the body with electromagnetic signals (waves) and sense signals emitted from the body in response to those excitation signals.

[0061] The coil is driven with an alternating current to generate an excitation signal for application to the body. These are, in some cases, propagating electromagnetic signals propagated through a medium, or the signals are non-propagating electromagnetic fields applied to the medium, i.e., by bringing a loop antenna source in proximity to the target medium. The alternating current creates an electric field of alternating electric field strength.

[0062] When the coil is in proximity to the body, the inductance L acquires an additional reflected inductance component L r due to eddy currents induced in the stimulated body as a result of the application of the excitation signal.

[0063] This is schematically shown in FIG. 1, which shows an example of driving a loop antenna 12 with an alternating current near the chest 16 of a subject to apply an electromagnetic signal 22 to the chest.

[0064] As a result, eddy currents 18 are induced within the chest.

[0065] These eddy currents effectively contribute to the inductance of the loop antenna 12. This is because the eddy currents themselves result in the generation of a time-varying magnetic flux 24 at a frequency equivalent to that generated by the primary antenna 12. These eddy current fluxes combine with the primary flux of the antenna, resulting in a change in the induced back EMF within the antenna and an increase in the measurable effective inductance.

[0066] The additional component of inductance arising from the eddy currents is the "reflected inductance" L r which is so called. The total inductance L t of the coil antenna 12 is expressed as follows. L t =L0 + L r Here, L0 is the self-inductance of the coil antenna 12 in free space, and L r is the reflected inductance caused by the presence of the nearby body.

[0067] Generally, the reflected inductance L r is a complex number and

Number

Number

Number

[0068] The addition of the reflected component of the inductance L r leads to a detuning of the electrical characteristics of the resonator circuit. In particular, both the natural radial frequency of the resonator circuit and the attenuation coefficient of the resonator circuit change. By measuring this detuning of the electrical characteristics, the real and imaginary parts of the reflected inductance L r can be detected.

[0069] In particular, the real part of the additional inductance component L r appears at the frequency of the resonator circuit or antenna. The imaginary part of the additional inductance component appears at the amplitude of the resonator circuit. Thus, by measuring the changes in the frequency and amplitude (current) of the resonator circuit and deriving the first and second input signals respectively, a signal indicating the underlying anatomical movement and phenomenon is detected.

[0070] For the sake of simplicity and ease of explanation, embodiments of the present invention will be described below. These embodiments have components or method steps for both generating and sensing induction signals and for performing the processing of the sensed signals (in particular, performing signal extraction procedures). However, it should be understood that embodiments of the present invention include only the components for performing signal processing, and that an input induction signal is received as a signal input in the system. For example, a series of embodiments include only a processor or control unit configured to perform signal extraction procedures. Thus, the descriptions and options described below should be understood as equally applicable to embodiments in which the system includes only the means for performing signal extraction procedures.

[0071] FIG. 2 shows a block diagram of the components of an exemplary induction sensing system 8 according to one or more embodiments.

[0072] The induction sensing system 8 is for sensing a return electromagnetic signal from the body in response to the application of an electromagnetic excitation signal to the body.

[0073] The system includes a resonator circuit 10 that includes a loop antenna 12 and a capacitor 13 electrically combined therewith. The capacitance of the capacitor 13 at least partially defines the natural resonance frequency of the resonator circuit (in the absence of forcing or damping). When the antenna 12 is excited, it tends to resonate naturally at the defined resonance frequency and generates an electromagnetic signal at the same frequency. Thus, by selecting the capacitance of the capacitor, it becomes possible to at least partially adjust the frequency of the generated electromagnetic signal.

[0074] The system 8 further includes signal generating means 14 adapted to excite the loop antenna to generate an electromagnetic excitation signal. The signal generating means includes driving means for driving the antenna, for example, at a radial frequency ω, that is, for driving the antenna with an alternating current at frequency ω. The driving means is, for example, an oscillator or includes an oscillator.

[0075] The signal generating means drives the antenna and the resonator circuit with a current at a radial frequency ω for which an electromagnetic excitation signal at the radial frequency ω is required.

[0076] By exciting the resonator, a resonant current is induced and flows back and forth through the loop antenna and into the capacitor. Driving an alternating current through the antenna stimulates the generation of an oscillating electromagnetic signal (wave).

[0077] The same antenna used to generate the electromagnetic excitation signal is used and, in response thereto, to sense the electromagnetic signal received from the body.

[0078] To avoid misunderstanding, the "electromagnetic excitation signal" simply means an electromagnetic signal for application to the body for the purpose of exciting or stimulating the generation of eddy currents in the body to stimulate the emission of electromagnetic signals from the body that are perceivable by the sensing system.

[0079] The "electromagnetic signal" generally means electromagnetic radiation emission or electromagnetic near-field vibration or electromagnetic vibration and / or electromagnetic waves.

[0080] System 8 further includes signal sensing means ( "signal sensing") 20 adapted to sense the return electromagnetic signal from the body using loop antenna 12 based on detecting a change in the electrical characteristics of resonator circuit 10. The signal sensing means includes signal processing or analysis means for detecting or monitoring the electrical characteristics of the current within resonator circuit 10.

[0081] For example, signal sensing means 20 is adapted to monitor at least the frequency of the resonator circuit current and the amplitude of the resonator circuit current. These characteristics of the current vary in accordance with the intensity of the reflected electromagnetic signal that returns from the body and is detected by the antenna.

[0082] The sensing of these signal characteristics is performed at the same time as (i.e., simultaneously with) the excitation of the antenna for generating the electromagnetic excitation signal. Thus, signal transmission and sensing are performed simultaneously.

[0083] The system is configured to perform a signal extraction procedure. For example, a processor or controller unit 56 is provided for executing the signal extraction procedure. The steps of an exemplary signal extraction procedure 30 according to one or more embodiments are outlined in block diagram form in FIG. 3.

[0084] The system is configured to detect (32) a first input signal from the sensed return electromagnetic signal, the first input signal being based on the frequency of the sensed return electromagnetic signal.

[0085] The system is further configured to detect (34) a second input signal from the sensed return electromagnetic signal, the second input signal being based on the sensed amplitude of the return electromagnetic signal sensed at the antenna.

[0086] The system is further configured to apply (36) a signal generation procedure having a step of generating a plurality of candidate signals, each candidate signal being formed from a different linear combination of the first and second input signals.

[0087] The system is further configured to apply a signal selection procedure (38) for selecting one of the candidate signals. The signal selection procedure is based on a predefined criterion related to one or more signal characteristics of the input signals, which is configured to separate signals related to a specific physiological source in the body. The selected signal forms the output signal. For example, according to embodiments of one or more applications, the selection criterion is configured to select a signal related to the respiration rate signal and / or the heart rate signal of the subject.

[0088] System 8 is adapted to perform a series of steps outlined using the above-described components. The system includes a controller or microprocessor ("MPU") 56 adapted to perform or facilitate these steps. An exemplary microprocessor 56 is shown in the exemplary system of FIG. 2 for purposes of illustration. However, a dedicated controller or microprocessor is not essential. In other examples, one or more of the other components of the system, such as the signal sensing means 20 and / or the signal generating means 14, are adapted to perform the steps.

[0089] In FIG. 2, the signal sensing means 20 is shown connected to the resonator circuit 10 via the signal generating means 14. However, this is not essential, and the signal sensing means and the signal processing means are connected independently to the resonator.

[0090] The signal extraction procedure is configured to extract a respiration signal, for example, like the respiration rate, in some examples. In some examples, it is configured to extract the heart rate. These are only two advantageous examples shown.

[0091] As described above, according to one or more embodiments, the resonator circuit 10, the signal generation means 14, and the signal sensing means 20 are omitted from the system 8. The system includes only a processor unit 56 configured to receive a signal input indicative of a return electromagnetic signal sensed from the body, and the return electromagnetic signal is, for example, when the resonator circuit is driven to generate an electromagnetic excitation signal, based on detecting a variation in the electrical characteristics of the resonator circuit, corresponding to the signal sensed by the loop antenna of the resonator circuit.

[0092] In a preferred series of examples, the system is configured to perform at least two repetitions or executions of a signal extraction method that, on the first pass, extracts a signal related to the breathing rate of the subject and, on the second pass, extracts a signal indicative of the heart rate (or vice versa).

[0093] The signal generation procedure includes the step of fusing or linearly combining the input signals in different ratios. This means adding or superimposing the input signals with different linear combination coefficients. For example, a series of exemplary candidate signals C generated from the input signals s1 and s2 by this process includes the following. C1 = 1.0s1 + 1.0s2 C2 = 1.5s1 + 1.0s2 C3 = 1.0s1 + 1.5s2 ··· C n = αs1 + βs2 Here, α and β are positive or negative.

[0094] Furthermore, the inductive sensing device according to an embodiment of the present invention measures the volume change of the human body. However, the input signal sensed by the antenna usually includes components of a plurality of different physiological phenomena. Furthermore, some phenomena are more strongly present in the induced signal than others. For example, since the respiratory signal component is stronger than the cardiac signal component, in the case of an antenna arranged on the chest, the respiratory signal component dominates the input signal, making it difficult to detect the heart rate signal.

[0095] Similarly, in the measurement of respiratory rate, the presence of cardiac components becomes an issue. Even when the respiratory rate increases up to 60 breaths per minute, the heart rate may decrease up to 30 beats per minute. Therefore, the frequency range of 30 - 60 BPM is shared by both phenomena. This means that purely frequency-based signal separation is impossible to achieve.

[0096] Signal fusion is a method of combining signals from various sources to cancel out unwanted components. The contributions of volume changes due to respiration and heartbeat differ in terms of frequency and absorption. Due to this difference, when signals are linearly combined (added or subtracted from each other with specific coefficients), the resulting signal is, in most cases, more suitable for extracting one of these physiological signals (heart rate and respiratory rate) than the other signals, and in most cases, more suitable for extracting the respiratory rate than the heart rate. By combining the signal with an appropriate linear multiple or ratio, the signal component related to the desired physiological signal is emphasized while suppressing the signal components from other physiological phenomena.

[0097] This is schematically illustrated in Figure 4. For reference, signal (a) represents the true respiratory signal measured, for example, by an additional auxiliary sensor. Signal (b) represents an exemplary first input signal (indicating the frequency of the received induction signal sensed by antenna 12). Signal (c) represents an exemplary second input signal (indicating the energy absorption by the body, for example, the amplitude of the induction signal sensed by antenna 12). Signal (d) represents an exemplary candidate signal formed from the linear combination of input signal (a) and input signal (b), particularly formed from the difference between the frequency signal (signal (b)) and the absorption signal (signal (c)).

[0098] The signal generation procedure 36 is performed in different ways. In particular, one broad approach is to determine or calculate a specific combination ratio or set of ratios of input signals to maximize the desired physiological signal components, and then select the best signal from this selected group. Another broad approach is simply to generate a large number of candidate signals formed from different mixing ratios of the input signals, where the ratios are either random or follow a standard sequence of combinatorial coefficients that change incrementally, and then determine the best signal from this large group. Examples of both of these approaches are outlined next.

[0099] There are various ways to search for a specific mixing matrix of multi-channel signals that accentuates specific features contained in the signal. In a particular example, each of the input signals can be regarded as a multi-channel signal (i.e., containing signal components from multiple physiological sources). The mixing matrix means a series of linear combination coefficients for combining the multi-channel input signals so as to accentuate or not accentuate specific signal components.

[0100] According to one or more examples, an independent component analysis (ICA) procedure is applied to determine the combination ratios or linear combination coefficients of different input signals for forming a plurality of candidate signals.

[0101] ICA is a well-known signal analysis method based on the assumption that the return electromagnetic signal sensed by a resonator circuit antenna is a composite signal formed from a plurality of signal components each corresponding to a different physiological source.

[0102] ICA attempts to determine a weight vector matrix that describes how the underlying physiological signal components are mapped to two detected input signals. This makes it possible to reconstruct the original physiological signal components from a specific linear combination of the input signals.

[0103] Using ICA (Independent Component Analysis), the multi-channel signal (i.e., the input signal) is decomposed into independent non-Gaussian signals. According to this approach, the induction frequency and absorption can be regarded as the two axes of a 2D vector array.

[0104] Generally speaking, ICA can be understood as rotating the signal vector so that each axis looks as non-Gaussian as possible.

[0105] According to one or more examples, the first step is to whiten (or "spherify") the data (i.e., the input signal). This makes the mean zero (centering) and normalizes the variance in all directions (whitening), thereby containing the vector substantially within a sphere around zero.

[0106] One common method of ICA suitable for use in the signal generation procedure according to one or more examples is known as FastICA. The details of this algorithm are described in detail, for example, in the paper [by Hyvarinen, A. and Oja, E. (2000), Independent component analysis: algorithms and applications. Neural Networks (Independent Component Analysis: Algorithms and Applications. Neural Networks), Vol. 13, pp. 411-430]. This has an effective iterative method for optimizing the weight vector matrix (or its inverse matrix) in order to use an efficient approximation of the entropy of the underlying source signal and minimize the entropy (maximize the non-Gaussianity).

[0107] According to one or more examples, instead of using ICA to generate candidate signals, the signals are combined in random combinations, or in standard pre-defined sequences, or a set of combinations (combination ratios or coefficients).

[0108] For example, candidate signals are generated according to a predefined signal generation protocol that defines sequentially increasing and / or decreasing linear combination coefficients of two or more input signals. The signal generation protocol effectively steps through a sequence of incrementing or decrementing the linear combination coefficients for each of the input signals, for example, with a regular increment, thereby generating a series of candidate signals according to a standard sequence of incrementing the linear combination coefficients.

[0109] For example, let r represent the combination ratio with a uniform step size of 0.1. r = [0.0, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1.0]

[0110] Inductive frequency signal (first input signal) I f and inductive absorption I a By adding and subtracting signals (second input signal), combinations of two series of signals can be generated. s1 = r·I f + (I - r)·I a s2 = r·I f - (I - r)·I a

[0111] It should also be noted that the final sign (i.e., positive or negative) of the combined (candidate) signal and the final scale of the signal (i.e., the amplitude of the signal) are not important. What is important is the relative combination ratio of the input signals included in the candidate signal.

[0112] One possible improvement to this procedure is to use an adaptive step size (i.e., a non-uniform step size), for example, based on gradient descent search.

[0113] When there are only a few input signals, such as only two input signals as in this example, the computational load of generating 100 different candidate signals according to a pre - defined signal generation protocol as described above is (in most cases) faster than a single ICA calculation. Thus, according to one or more examples, instead of using the ICA procedure, the computational demand or usage can be advantageously reduced by following a random signal generation process, or by following a fixed protocol or a linear combination coefficient of an example.

[0114] Once the candidate signals are generated, it is necessary to determine which one (or more) of the candidate signals to use in order to extract a measurement or sample of the physiological parameter or signal of interest (e.g., respiratory rate or heart rate according to two examples). Thus, a signal selection procedure is applied.

[0115] According to one or more examples, the signal selection procedure includes a scoring procedure or algorithm where each candidate signal is "scored" according to its determined likelihood of best reflecting the physiological parameter of interest, e.g., according to how "similar to the respiratory rate" or how "similar to the heart rate" the signal is.

[0116] Scoring is based on one or more pre - defined signal characteristics known to be associated with signals that reflect a particular one or more physiological phenomena. In other words, scoring is done by looking for characteristics typical of the physiological phenomenon of interest.

[0117] Even when using ICA to generate initial candidate signals, it is advantageous to use a signal selection procedure. This is because although ICA may succeed in separating one or more specific physiological signals from other distortions or other physiological signals (e.g., heart signals or movement signals), it is not known which of the resulting separated signal components to use for a particular physiological phenomenon. In other words, each source cannot be attributed to a different separated signal component, and it is not known which physiological phenomenon or source of noise each separated signal component is related to.

[0118] Therefore, additional steps are required to identify the generated candidate signals for which the output signal relates to a particular physiological phenomenon of interest. For example, additional steps are required to "score" all output signals based on how well they match signal characteristics known to be associated with the physiological signal source in question.

[0119] For example, one possible approach to identifying the best candidate for representing the respiratory rate of a subject is to analyze how "low frequency" the signal is. The typical frequency of the respiratory rate is lower than both the typical frequency range of the heart rate and the distortion of periodic movement due to walking.

[0120] In this example, scoring is performed, for example, by counting the maximum and minimum of the signal. The signal with the minimum count obtains the highest score and is selected as the output signal.

[0121] This approach is a simple and fast way to perform signal scoring. For example, it is simpler than using a band - pass filter. With just a band - pass filter, it is not possible to determine whether a signal is low - frequency or not. To make this determination, it is necessary to apply multiple band - pass filters and identify which of the filtered signals has the highest energy, or to perform a Fourier transform and analyze the spectrum. These are computationally expensive operations in terms of processing resources. Counting the zero - crossings of peaks / valleys is a simpler and faster alternative.

[0122] This is schematically shown in FIG. 5, which shows four exemplary series of candidate signals and sets the count numbers of peaks and valleys, i.e., scores, respectively. The "best" signal is selected as signal - 1, which has the fewest number of peaks and valleys.

[0123] When using ICA, as described above, scoring can be applied only to the two output signals of the ICA procedure, or it can also be applied to these two outputs (I f and I a ) in addition to the two original input signals to the ICA procedure. In the latter case, this means that in the event that the ICA procedure fails to generate a signal such as better respiration, the original input signals can instead be selected as candidate signals without affecting the final result.

[0124] As a further example, to identify the best candidate for indicating a heart - rate signal, the scoring of candidate fusion signals for heart - rate measurement is performed based on two signal characteristics according to one or more examples. 1. For example, the frequency of the signal, indicated by the number of zero - crossings (Z). 2. The flatness (F) of the energy envelope. This is defined as the value obtained by multiplying the ratio of the (uncorrected) standard deviation (S) to the mean absolute deviation (M) by a coefficient 200 (determined empirically, for example).

Number

Number

Number

[0125] Different candidate signals are scored based on their respective Z - values and F - values, and higher scores result from higher Z - values and F - values.

[0126] This approach is based on the recognition that signal components not related to the heart typically exhibit lower frequencies than the heart rate signal (respiratory signal), and the signal envelope of the heart pulse (within the time window) is generally relatively constant.

[0127] The values of the two features can be summed to calculate the final score, score = Z + F.

[0128] This exemplary approach is shown schematically in FIG. 6, showing four exemplary sequences of candidate signals and their respective Z - values and F - values. The "best" candidate signal (i.e., the signal with the highest Z - value and F - value) is shown. In this example, the selected signal is Signal - 1.

[0129] According to one or more embodiments, the signal extraction procedure performed by the system has additional signal processing steps applied immediately after the detection 32, 34 of the input signal, and the signal processing steps are configured to suppress motion artifacts in each of the input signals.

[0130] Motion suppression is performed in various ways. According to one or more examples, it includes the simple application of a filter such as a band - pass filter. The filter parameters are configured to select frequency components within a range known to be typical of the physiological phenomenon or parameter in question, or, at least, to exclude one or more frequency components known to be outside the typical range of the phenomenon in question.

[0131] One advantageous method for suppressing motion artifacts in an input signal is briefly outlined here. This approach is particularly suitable for filtering or removing artifacts of periodic motion, i.e., those exhibiting a periodic frequency. Such artifacts are caused, for example, by a patient's walking or running (when the sensor is a portable wearable sensor), or other regular motions affecting the subject.

[0132] In this example, the further motion suppression processing step is based on detecting or otherwise identifying the fundamental frequency of the periodic motion to be suppressed, and then applying a filter to remove the frequency components around the range of this fundamental frequency.

[0133] For example, the system is configured to receive an input indicating the fundamental frequency of the motion to be suppressed and apply a filter (e.g., a notch filter) to the input signal. The notch filter has an adjustable frequency setting and is applied to the input signal at one or more multiples of the fundamental frequency.

[0134] As an example, motion artifacts related to walking or running are typically limited within a specific frequency band of the induction sensor signal. Therefore, these periodic distortions can be suppressed, for example, by an adaptive notch filter, at multiples of the fundamental frequency (f0) of the periodic motion.

[0135] The fundamental frequency f0 of the periodic motion can be determined or detected by the system itself or received as an input to the system and already determined or specified externally. The fundamental frequency f0 is detected, for example, by an external motion sensor element and transmitted to the system of the present invention.

[0136] According to one or more examples, the motion frequency is determined by the system based on, for example, an input motion signal from a motion sensor carried by the subject.

[0137] For example, the movement frequency can be measured using the input from an accelerometer, such as a 3D accelerometer combined with a subject.

[0138] The 3D accelerometer measures the acceleration in three directions: X, Y, and Z (in the reference frame of the accelerometer). These are called body axes.

[0139] The 3D accelerometer measures both the acceleration due to gravity and the acceleration due to movement. Gravity introduces a constant bias (g) that spreads across various axes. To extract only the acceleration due to movement (A m ), it is necessary to subtract the gravity component. This is done in two steps.

[0140] First, the norm of the acceleration vector is calculated.

Number

[0141] Second, by subtracting a constant gravity, the gravity bias is removed and the signal is adjusted. A m =|A n -g| where

Number

[0142] The movement suppression process has an initial step of detecting whether the input signal contains movement artifacts that need to be suppressed. The remaining steps of the movement suppression process are only applied if movement artifacts are detected.

[0143] As an example, a movement threshold T m can be applied to the signal A m , and a movement signal A m that exceeds the threshold T mIt is treated as indicating that the signal has motion artifacts that need to be suppressed (e.g., after window processing with a 10 - second hanning window).

[0144] The threshold is predefined and stored locally in the system or determined empirically, for example. As an example, in one exemplary application, a volunteer study in which an indication - sensing device according to an embodiment is combined with an accelerometer and worn by each volunteer was conducted with 15 volunteers. The measurements were repeated at a frequency of once per second. As a result of this study, a threshold T 2 of 1 m / s m was identified. In some embodiments, a threshold T 2 of 1 m / s m is used.

[0145] As described above, the movement frequency can be measured using signals from a 3D accelerometer. In particular, the frequency can be determined from the norm of the 3D accelerometer signal. The distortions introduced by the left and right feet are somewhat different. In some examples, the fundamental frequency f0 measured from the acceleration is divided by 2 to suppress them.

[0146] By examining the measurement spectra of the acceleration frequency and the sensed induction signal frequency respectively, it was found that the main contribution of the movement distortion is on the order of approximately 50 rpm (revolutions per minute), which is about half of the movement fundamental frequency f0.

[0147] Numerous methods for extracting the fundamental frequency f0 from the input signal are known in the art, and those skilled in the art will be aware of such exemplary methods. As an example, the detection of the fundamental frequency is usually performed based on the step of finding peaks in the amplitude spectrum of the signal or in the time - domain correlation function (such as autocorrelation) of the signal.

[0148] In a preferred embodiment, the fundamental frequency f0 is determined from the input acceleration signal using a particularly advantageous algorithm known as combined spectral pitch detection. The full details of this algorithm are described in detail in document WO2012 / 063185 and are referred to by the reader for further implementation details.

[0149] Briefly, in a typical approach to detecting the fundamental frequency based on detecting peaks in the amplitude or time-domain correlation function, false detections occur at multiples of the fundamental frequency. In the case of spectral analysis, this is, for example, harmonics. In the case of time-domain correlation, this is a multiple of the repetitive pulse signal.

[0150] (Another pitch detection method outlined in WO2012 / 063185) is based on combining the frequency-domain signal and the time-domain signal such that the resulting signal has only the f0 component.

[0151] Figure 7 briefly outlines the steps of the combined spectral pitch detection algorithm in block diagram form.

[0152] In a first step 62, the input signal S is windowed by applying a window function. Further details can be found in WO2012 / 063185, page 8, lines 14 to 30.

[0153] The resulting windowed signal S w is transformed from the time domain to the frequency domain in step 64 based on the application of a discrete Fourier transform (DFT) that provides the spectrum of the signal. For reasons of efficiency, preferably a fast Fourier transform (FFT) (e.g., a radix-2 FFT) is used. Further details can be found in WO2012 / 063185, page 8, line 31 to page 9, line 9.

[0154] The amplitude |S| of the signal is extracted. Further details can be found in WO2012 / 063185, page 9, lines 10 to 25.

[0155] The window-compressed amplitude spectrum is then converted to the time domain in step 66 using an inverse Fourier transform (IFT). For example, an inverse fast Fourier transform (IFFT) is used. This conversion to the time domain is used to obtain a correlation signal c that includes peaks at multiples of the fundamental frequency. Further details can be found in WO2012 / 063185, page 10, line 21 to page 11, line 9.

[0156] In step 68, the combined spectrum b is formed by multiplying the amplitude spectrum S and the correlation signal c. This combined spectrum b has a distinct peak at the fundamental frequency. By multiplying these spectra, the harmonics of the frequency spectrum are attenuated and the fundamental frequency remains as the dominant peak. Further details can be found in WO2012 / 063185, page 11, line 18 to page 12, line 25.

[0157] Finally, in step 70, a peak position detection step is performed that includes a step of searching for the maximum value of the combined spectrum b. Thereby, an output frequency value p in Hz indicating the fundamental frequency f0 is obtained. Further details can be found in WO2012 / 063185, page 13, line 1 to page 13, line 13.

[0158] As described above, when the fundamental frequency f0 of the motion artifact is detected or received by the system, the motion artifact is suppressed in the input signal based on the application of a filter having frequency parameters set according to the fundamental frequency. For example, the filter is a notch filter such as an adaptable notch filter that can adjust the frequency parameters and filter multiple frequency components.

[0159] As an example, when the physiological phenomenon being measured is the respiratory rate, the first and second input signals (frequency signal and amplitude or absorption signal) are each filtered by a notch filter at f0 and f0 / 2 Hz, where f0 is the fundamental wave of the periodic motion artifact.

[0160] As an example, FIG. 8 shows the function of a notch filter applied at a frequency of 40 rpm with various different quality factors Q from Q = 2 to Q = 6. Lines 102 to 110 each show the filter function for Q - factors 2, 3, 4, 5, and 6.

[0161] As an example, when the physiological phenomenon to be extracted is the respiratory rate, Q is preferably set to Q = 5. When the physiological phenomenon to be extracted is the heart rate, Q is preferably set to Q = 6 for heart rate measurement.

[0162] In the calculation of the heart rate, it is desirable that only f0 / 2 Hz is suppressed. The reason for this is that usually the fundamental frequency f0 of the associated motion artifact is much higher than the respiratory frequency, whereas in the case of heart rate measurement, the motion f0 is usually much closer to the heart rate frequency. Thus, applying a filter at f0 in the case of heart rate detection would completely remove the fundamental wave of the heart rate signal itself. Therefore, to avoid this, it is desirable to set the notch filter to simply f0 / 2 Hz.

[0163] When the best candidate signal is selected in the signal selection procedure (step 38 of the method of FIG. 3 above), further steps need to be performed to actually extract the measured value of the physiological phenomenon in question from the selected candidate signal.

[0164] In some examples, this is very simple and is done, for example, by simply detecting the frequency of the candidate signal (e.g., to obtain the respiratory rate or heart rate).

[0165] In further examples, additional processing steps are applied to extract a more stringent or accurate measure of the physiological phenomenon.

[0166] For example, when the physiological phenomenon or parameter to be obtained is a number, such as the respiratory rate or heart rate, the extraction of the measure of the physiological parameter from the candidate signal has the step of detecting the fundamental frequency f0 from the selected candidate signal. This helps to ensure that the derived measurement of the parameter or phenomenon is not overly affected by the noise-related frequency components within the signal.

[0167] As one example, the extraction of the respiratory rate measurement or signal from the selected candidate signal will be described.

[0168] Figure 9 outlines the steps of an exemplary algorithm for extracting or deriving the respiratory rate from the selected candidate signal s.

[0169] This procedure is based on the step of detecting the fundamental frequency of the selected candidate signal s. The procedure for doing this is shown in Figure 7 and is substantially the same as the exemplary algorithm described above. Accordingly, the similar steps will not be described in detail here and the reader is referred instead to the above description of the relevant steps associated with Figure 7.

[0170] Similar to the procedure of Figure 7, this procedure begins with a windowing step 82, where the input candidate signal s is windowed in time to produce the windowed signal s w is generated.

[0171] The resulting windowed signal S w is then transformed from the time domain to the frequency domain in step 84 based on the application of the discrete Fourier transform (DFT) to provide the spectrum of the signal. For reasons of efficiency, preferably the fast Fourier transform (FFT) (e.g., a radix-2 FFT) is used.

[0172] The amplitude |S| of the signal is extracted. In contrast to the method of FIG. 6, the extracted amplitude is raised to the 1.7th power. This raised index has the effect of emphasizing the spectral peak and reducing the contribution of noise. The power value (1.7) is a value that the inventor has empirically identified as being particularly advantageous. In the case of the respiration signal, this has the effect of further emphasizing the fundamental frequency component (f0). In motion detection and heart rate extraction, this emphasis is not performed because the fundamental frequency in these signals may be weaker than the harmonics.

[0173] The window-compressed amplitude spectrum is converted to the time domain in step 86 using an inverse Fourier transform (IFT). For example, an inverse fast Fourier transform (IFFT) is used. This conversion to the time domain is used to obtain a correlation signal c containing peaks at multiples of the fundamental frequency.

[0174] In step 88, a combined spectrum b is formed by multiplying the amplitude spectrum S and the correlation signal c. This combined spectrum b has distinct peaks at the fundamental frequency. By multiplying these spectra, the harmonics in the frequency spectrum are attenuated and the fundamental frequency remains the dominant peak. Further details can be found in WO2012 / 063185, page 11, line 18 to page 12, line 25.

[0175] In step 90, a peak position detection step is performed that includes a step of searching for the maximum value of the combined spectrum b. This results in an output frequency value f0 indicating the fundamental frequency of the respiration rate in Hz. Further details can be found in WO2012 / 063185, page 13, line 1 to page 13, line 13.

[0176] The calculated f0 can be used as the final respiration rate measurement value. In some examples, the algorithm parameters are configured to provide measurement values every second over a 25-second window (thus having 24 seconds of overlap between windows). The measurements every second in a 25-second window are called frames.

[0177] The method optionally has an additional averaging step 94, where the output over multiple frames is averaged over a longer period to provide a more stringent prediction. For example, as currently required by the current World Health Organization (WHO) guidelines, the derived frequency rate value is averaged over a 1-minute window.

[0178] An exemplary procedure for extracting the heart rate from the selected candidate signal s is outlined in block diagram form in FIG. 10. This procedure is based on identifying the fundamental frequency f0 of the selected candidate signal s.

[0179] An exemplary algorithm for determining the fundamental frequency (f0) of the heart rate was described above in connection with motion detection and is substantially the same as the exemplary algorithm outlined in FIG. 7. Accordingly, similar steps are not described in detail here, and the reader is instead referred to the above description of the relevant steps associated with FIG. 7.

[0180] The calculated f0 can be used as the final heart rate measurement value. In some examples, the algorithm parameters are configured to provide measurements every second over a 25-second window (thus having 24 seconds of overlap between windows). The measurements every second in the 25-second window are called frames. In this example, the corresponding (computation) delay is 12.5 seconds.

[0181] The method optionally has an additional averaging step, where the output over multiple frames is averaged over a longer period to provide a more stringent prediction.

[0182] As described above, according to one or more embodiments, the signal extraction procedure has an additional step of applying a band - pass filter to the input signal before the signal generation procedure. The band - pass filter makes it possible to filter out frequency components known to be outside the typical range of the physiological phenomenon in question, and ultimately makes the extracted measurements more rigorous and reliable.

[0183] The frequency band of the band - pass filter is set depending on the physiological signal to be extracted.

[0184] As an example, when the physiological parameter to be extracted is the respiratory rate, the band - pass filter is set based on the typical range of the respiratory rate of the subject population.

[0185] The typical respiratory rate (for adults) ranges from 4 BPM (breaths per minute) to 60 BPM.

[0186] Therefore, a band - pass filter is set to remove frequency components outside this range.

[0187] As an example, the filter includes a cascade of two Butterworth IIR filters with a second - order high - pass filter having, for example, an attenuation of - 6 dB at the cut - off frequency and a third - order low - pass filter.

[0188] As an example, the duration of the analysis window is set to 25 seconds. The filter is applied independently to each analysis window in the forward and reverse (zero - phase) directions without introducing filter delay.

[0189] As a further example, when the physiological parameter to be extracted is the heart rate, the band - pass filter is set based on the typical range of the heart rate or pulse rate of the subject population.

[0190] The typical heart rate (for adults) ranges from 30 BPM (beats per minute) to 220 BPM. Therefore, a band - pass filter is set to remove frequency components outside this range.

[0191] In some examples, when the lower cut-off of 30 BPM is lower than twice the respiratory rate, 30 BPM is used.

[0192] The filter applied is the same as that used for measuring the respiratory rate. For example, the filter applied includes a cascade of two Butterworth IIR filters with a second-order high-pass filter having an attenuation of, for example, -6 dB at the cut-off frequency and a third-order low-pass filter.

[0193] The duration of the analysis window is 25 seconds. The filter is applied independently to each analysis window in the forward and reverse directions (zero phase) without introducing filter delay.

[0194] According to an advantageous series of embodiments, the system is configured to perform at least two executions of a signal extraction procedure in at least one mode, and the signal selection procedure is configured to select signals related to different respective first and second physiological phenomena in the first and second executions.

[0195] In a preferred example, in at least the second execution, a band-pass filter is applied to the sensed input signal prior to the signal generation procedure.

[0196] The signal selection procedure is preferably configured to select a signal related to the respiratory rate of the subject in the first execution and a signal related to the heart rate of the subject in the second execution.

[0197] An exemplary signal extraction method according to this approach is outlined in block diagram form in FIG. 11.

[0198] As described above, the method has a step of applying an electromagnetic excitation signal to the body of the subject using a resonator circuit including a loop antenna.

[0199] The method further comprises a step of sensing the return electromagnetic signal from the body using a loop antenna, based on a step of detecting a change in the electrical characteristics of the resonator circuit.

[0200] From the sensed return electromagnetic signal, at least a first input signal 32 based on the frequency of the sensed return electromagnetic signal and a second input signal 34 based on the sensed amplitude of the sensed return electromagnetic signal are detected. The amplitude signal indicates the absorption of the initially applied electromagnetic excitation signal by the body.

[0201] Preferably, simultaneously with the detection of the input signal, the method further comprises a step of detecting the movement (e.g., walking motion) of the subject in step 52, using a motion sensor that is, for example, worn by the subject or attached to a qualified person and carried by the subject. The motion sensor includes, for example, an accelerometer such as a 3D accelerometer.

[0202] When movement is detected, the system is configured to apply a movement suppression procedure in step 54 to suppress movement artifacts in the detected input signal. Examples of the movement suppression procedure are described in detail above with reference to FIG. 7, and the reader is referred to this description for details of the movement suppression procedure in step 54.

[0203] Thereafter, the method proceeds to a first execution or iteration of steps for extracting physiological signals or measurements from the input signal.

[0204] In the first execution, the method is configured to extract a measurement of the respiratory rate of the subject.

[0205] In the first execution, the method preferably has a step (55) of applying a band-pass filter to each of the input signals. The frequency parameters of the band-pass filter are set based on, for example, the typical frequency range of the respiratory rate.

[0206] Thereafter, the method has a step (36) of applying a signal generation procedure having a step of generating a plurality of candidate signals, each candidate signal being formed from a different linear combination of the first and second input signals. In other words, the induced frequency signal and the absorption signal (i.e., the first and second input signals) are combined into each of a series of candidate signals. This results in, for example, an effect of eliminating high-frequency distortion in the resulting signal.

[0207] Thereafter, the respiratory rate is calculated.

[0208] This has a step of applying a signal selection procedure 38 for selecting one of the generated candidate signals, the signal selection procedure being based on a predefined criterion related to one or more signal characteristics of the input signals, the criterion being configured to separate signals related to respiratory activity in the body. The selected signal forms the output signal. The criterion is predetermined or predefined and is based, for example, on empirical observations of typical signal characteristics that tend to be reliably associated with signals resulting from respiratory action in the body.

[0209] The respiratory rate is calculated (40) using the selected candidate signal. The procedure for extracting the respiratory rate from the selected candidate signal is described in detail in the first half of this disclosure.

[0210] Thereafter, the method proceeds to a second execution of the signal extraction procedure, which is configured to extract a measurement or signal indicative of the subject's heart rate at this time.

[0211] After band-pass filtering 55, a signal generation procedure 36 having a step of generating a plurality of candidate signals, each formed from a different linear combination or combination of the two input signals, is applied again.

[0212] Thereafter, the heart rate is calculated, which is based on the step of applying the signal selection procedure 38 again, and the selection criterion for the procedure in this case is configured to select a signal related to the heart rate. Again, these criteria are predetermined or predefined, for example, based on empirical investigations or observations.

[0213] When a candidate signal is selected for the calculation of the heart rate, the heart rate is extracted from the selected signal (40). An exemplary procedure for extracting a heart rate measurement value from the selected candidate signal has been described in detail above.

[0214] The final result of the method is a series of output measurements 42 indicating the respiratory rate and the heart rate, respectively.

[0215] Another exemplary signal extraction method according to this approach is outlined in block diagram form in FIG. 12, which is similar to the exemplary signal extraction method shown in FIG. 11, but provides additional feedback from the number calculations 38, 40 to the band-pass filtering 55.

[0216] The effective heart rate (for adults) is in the range of about 30 BPM to 220 BPM. When measuring the heart rate using a biometric signal, it is advantageous to remove frequency components outside the effective range from the signal or at least limit the search space between the minimum and maximum numbers. A band-pass filter can be used to remove such unwanted frequency components. Any means of further restricting this effective range improves the measurement results.

[0217] According to the embodiment shown in FIG. 12, if the respiratory rate is known, the respiratory rate can be used to further limit the effective heart rate range. It is even more advantageous to limit the heart rate range when the biometric signal used to derive the heart rate also includes frequency components related to the respiratory rate.

[0218] For example, the breathing range of an adult usually ranges from 4 to 60 times per minute. Therefore, the fundamental frequency (f0) measured at 30 to 60 beats per minute belongs to both the heart rate and the respiratory rate. Using the finding that the heart rate is at least the first coefficient F1 (e.g., 2 times) of the respiratory rate helps to suppress the first harmonic (e.g., the first two harmonics) of the respiratory rate in the signal and improve the measurement accuracy of the heart rate measurement.

[0219] Similarly, using the finding that even when the heart rate is at its maximum, the second coefficient F2 of the respiration rate (e.g., 10 times) is also used to limit the upper limit of the effective heart rate frequency, which also improves the measurement results.

[0220] Therefore, in the embodiment shown in FIG. 12, the output measurement value 42 of the respiration rate is fed back to the band-pass filtering 55 that uses this finding to remove the frequency components in the detected input signal that cannot be components of the heart rate signal for determining the heart rate signal. For example, when the output measurement value 42 of the respiration rate is R1, multiplying the value of R1 by the first coefficient F1 gives the lower limit of the frequency considered in the input signal detected to determine the heart rate. Optionally, further multiplying the value of R1 by the second coefficient F2 gives the upper limit of the frequency considered in the input signal detected to determine the heart rate.

[0221] The coefficient F1 and the coefficient F2 can be set to fixed values (e.g., for adults, F1 = 2 and F2 = 10), or can be set in advance or individually by the user according to patient characteristics (e.g., age, gender, health status, etc.). The coefficients 2 and / or 10 applied to adults do not need to be exact. Values close to 2 and / or 10, or other individual values, are used as well.

[0222] This embodiment is preferably applied when respiration measurement and heart rate measurement are performed simultaneously. The measurement of the respiration rate usually precedes the measurement of the heart rate. In an exemplary implementation, the output of the respiration rate measurement is doubled, and if this value is higher than the minimum heart rate (e.g., 30 BPM) expected to be measured by the system, twice the respiration rate is used. As a further improvement step, the upper limit of the heart rate range is also limited to a maximum of 10 times the respiration rate.

[0223] Each implementation option and detail of the above steps is understood and interpreted according to the description and illustration previously provided in this disclosure regarding the aspects of the apparatus of the present invention (i.e., the aspects of the system).

[0224] Any of the features or details of the examples, options, or embodiments described above with respect to the apparatus (or system) of the present invention, with the necessary changes, are applied or combined or incorporated into the current method aspects of the present invention.

[0225] The above examples specifically mention the extraction of respiratory rate and heart rate, but these are only two possible examples. Embodiments of the present invention are applicable to the extraction of any physiological or anatomical signal from the body, particularly signals related to or caused by the movement of one or more features within the body or the body. Inductive sensing is particularly suitable for detecting the movement of the body containing water.

[0226] An example according to a further aspect of the present invention provides an inductive sensing method based on the step of sensing a return electromagnetic signal from the body in response to the application of an electromagnetic excitation signal to the body.

[0227] In a series of embodiments, the method has the step of receiving a signal input indicative of the sensed return electromagnetic signal, and the return electromagnetic signal corresponds to the signal sensed by the loop antenna of the resonator circuit based on the step of detecting a change in the electrical characteristics of the resonator circuit when the resonator circuit is driven to generate an electromagnetic excitation signal.

[0228] The method further has the step of performing a signal extraction procedure 30.

[0229] The signal extraction procedure has the step 32 of detecting a first input signal from the sensed return electromagnetic signal, and the first input signal is based on the frequency of the sensed return electromagnetic signal.

[0230] The signal extraction procedure further has the step 34 of detecting a second input signal from the sensed return electromagnetic signal, and the second input signal is based on the sensed amplitude of the sensed return electromagnetic signal.

[0231] The signal extraction procedure 30 further has a step 36 of applying a signal generation procedure having a step of generating a plurality of candidate signals, each candidate signal being formed from a different linear combination of the first and second input signals.

[0232] The signal extraction procedure further has a step 36 of applying a signal selection procedure for selecting one of the candidate signals, the signal selection procedure being based on a predefined criterion related to one or more signal characteristics of the input signal, the criterion being configured to separate signals related to a specific physiological source of the body, and the selected signal forms the output signal.

[0233] In a further series of embodiments, the method further has a step of performing physical inductive sensing and a step of acquiring an inductive sensing signal.

[0234] In particular, according to one or more embodiments, the method further applying an electromagnetic excitation signal to the body using a resonator circuit including a loop antenna, and sensing the return electromagnetic signal from the body using the loop antenna based on detecting a change in an electrical characteristic of the resonator circuit.

[0235] An example according to a further aspect provides a computer program product including code means configured to cause a processor to execute, when executed by the processor, a method according to any example or embodiment outlined above or described below or according to any claim of the present application.

[0236] As described above, the system uses a processor to perform data processing. The processor is implemented in various ways using software and / or hardware to perform the various necessary functions. The processor typically uses one or more microprocessors that can be programmed using software (e.g., microcode) to perform the necessary functions. The processor is implemented as a combination of dedicated hardware for performing some functions and one or more programmed microprocessors and associated circuitry for performing other functions.

[0237] Examples of circuit configurations applicable in various embodiments of the present disclosure include, but are not limited to, conventional microprocessors, application specific integrated circuits (ASICs), and field programmable gate arrays (FPGAs).

[0238] In various implementations, the processor is associated with one or more storage media such as volatile and non-volatile computer memories such as RAM, PROM, EPROM, and EEPROM. The storage media is encoded with one or more programs that perform the necessary functions when executed in one or more processors and / or controllers. The various storage media may be fixed within the processor or controller or may be portable so that the stored one or more programs can be loaded into the processor.

[0239] Modifications to the disclosed embodiments will be understood and effected by those skilled in the art in practicing the claimed invention, by considering the drawings, the disclosure, and the appended claims. In the claims, the term "comprising" does not exclude other elements or steps, and the singular does not exclude the plural.

[0240] A single processor or other unit performs the functions of several items recited in the claims.

[0241] The mere fact that certain measures are recited in mutually different dependent claims does not indicate that a combination of these measures cannot be used advantageously.

[0242] The computer program is stored / distributed by a suitable medium such as an optical storage medium or a solid state medium supplied together with or as part of other hardware, but may also be distributed in other forms such as via the Internet or other wired or wireless communication systems.

[0243] It should be noted that when the term "adapted" is used in the claims or the description, the term "adapted" is intended to be equivalent to the term "configured".

[0244] Reference signs in the claims should not be construed as limiting the scope.

Claims

1. A system for use in inductive sensing for processing a return electromagnetic signal from the body in response to the application of an electromagnetic excitation signal to the body, the system receives a signal input indicative of the sensed return electromagnetic signal, the return electromagnetic signal corresponding to a signal sensed by a loop antenna of a resonator circuit based on detecting a change in an electrical characteristic of the resonator circuit when the resonator circuit is driven to generate the electromagnetic excitation signal, the system detects a first input signal based on the frequency of the sensed return electromagnetic signal from the sensed return electromagnetic signal, detects a second input signal based on the sensed amplitude of the sensed return electromagnetic signal from the sensed return electromagnetic signal, applies a signal generation procedure having generating a plurality of candidate signals, each candidate signal being formed from a different linear combination of the first and second input signals, applies a signal selection procedure for selecting one of the candidate signals, performs a signal extraction procedure, the signal selection procedure is based on a predefined criterion related to one or more signal characteristics of the input signals, the criterion being configured to separate a signal related to a specific physiological source of the body, and the selected candidate signal forms an output signal, system.

2. The system includes a resonator circuit including a loop antenna, signal generation means for exciting the loop antenna to generate the electromagnetic excitation signal, and signal sensing means for sensing the return electromagnetic signal from the body using the loop antenna based on detecting a change in an electrical characteristic of the resonator circuit, The system according to claim 1, further comprising an inductive sensing device.

3. The signal generation procedure is based on the use of an independent component analysis method, the system according to claim 1 or 2.

4. The signal generation procedure is based on the use of a predefined set of signal combination ratios for forming the plurality of candidate signals, the system according to any one of claims 1 to 3.

5. The criterion of the signal selection procedure includes one or both of the frequency of the candidate signal and the maximum and minimum numbers of the candidate signal in a given time window, the system according to any one of claims 1 to 4.

6. The system according to any one of claims 1 to 5, wherein the signal extraction procedure further comprises generating an information output indicating a specific physiological phenomenon based on the selected candidate signal.

7. The system according to any one of claims 1 to 6, wherein the signal extraction procedure has a further step of applying a band-pass filter to the input signal before the signal generation procedure.

8. The system according to any one of claims 1 to 7, wherein the signal extraction procedure has a further signal processing step applied immediately after detection of the input signal, and the further signal processing step suppresses movement artifacts in each of the input signals.

9. The further signal processing step comprises receiving an input indicating the fundamental frequency of the movement to be suppressed, and applying a notch filter to the input signal, the notch filter having an adaptable frequency setting, the notch filter being applied to the input signal at one or more multiples of the fundamental frequency. The system according to claim 8.

10. The system according to any one of claims 1 to 9, wherein the signal selection procedure selects a candidate signal determined to indicate the respiratory rate of the subject in at least one mode.

11. The system according to any one of claims 1 to 10, wherein the system performs at least two executions of the signal extraction procedure in at least one mode, and the signal selection procedure selects signals related to different first and second physiological phenomena in the first and second executions.

12. The system according to claim 11, wherein a band-pass filter is applied to the sensed input signal prior to the signal generation procedure, at least in the second execution.

13. The system according to claim 11 or 12, wherein the signal selection procedure selects a signal related to the respiratory rate of the subject in the first execution and a signal related to the heart rate of the subject in the second execution.

14. A method for use in inductive sensing for processing a return electromagnetic signal from a body in response to the application of an electromagnetic excitation signal to the body, the method comprising: Receiving a signal input indicative of the sensed return electromagnetic signal, wherein the return electromagnetic signal corresponds to a signal sensed by a loop antenna of the resonator circuit based on a step of detecting a change in electrical characteristics of the resonator circuit when the resonator circuit is driven to generate the electromagnetic excitation signal; Performing a signal extraction procedure, the performing of the signal extraction procedure comprising: Detecting, from the sensed return electromagnetic signal, a first input signal based on the frequency of the sensed return electromagnetic signal; Detecting, from the sensed return electromagnetic signal, a second input signal based on the sensed amplitude of the sensed return electromagnetic signal; Applying a signal generation procedure having a step of generating a plurality of candidate signals, each candidate signal being formed from a different linear combination of the first and second input signals; Applying a signal selection procedure for selecting one of the candidate signals, the signal selection procedure being based on a predefined criterion related to one or more signal characteristics of the input signals, the criterion separating signals related to a particular physiological source of the body, and the selected signal forming an output signal. Claim 15 A computer program comprising code means which, when executed on a processor, cause the processor to perform the method according to claim 14.

Images