Method and apparatus for deriving a parameter value indicative of arrhythmia risk

Abstract

A data-processing apparatus for deriving a parameter value indicative of a subject's risk of a future arrhythmia comprises a processing system (6), configured to, for a plurality of frames (500) of a sequence of two-dimensional magnetic field data frames, each representing the rate of change of a magnetic field in the subject's heart at a respective cardiac-cycle instant within a cardiac cycle, define a vector (510) extending between a minimum-rate-of-change position and a maximum-rate-of-change position. The system (6) is further configured to calculate a frame-pair angle change for a plurality of successive pairs of frames, by calculating the absolute value of the difference in angle (512) of the vector defined for each frame of the pair, and calculate the parameter value by taking a combination of at least two of the frame-pair angle changes.

Description

[0001] 410.163722 / 02

[0002] Method and Apparatus for deriving a parameter value indicative of arrhythmia risk

[0003] BACKGROUND OF THE INVENTION

[0004] This invention relates to a processing system, a method, a system and software for deriving a parameter value indicative of a subject’s risk of a future arrhythmia.

[0005] Sudden cardiac death due to sudden onset of malignant ventricular arrhythmias (mVA) is a leading cause of mortality worldwide, despite the availability of effective prophylactic treatments such as implantable cardioverter defibrillators (ICDs). An ICD is a small battery-powered and surgically implanted device that monitors an individual’s heart rhythm and, if it detects the onset of an mVA, provides an electrical shock or other pacing intervention to restore the correct cardiac rhythm and prevent progression to cardiac arrest and death.

[0006] Unfortunately, current tools to assess mVA risk perform poorly, thus hindering the ability of clinicians to accurately target available prophylactic treatments towards those that will benefit.

[0007] As a result, many patients who do need protection are not identified for prophylactic therapy, thus leading to potentially avoidable mortality due to arrhythmic sudden cardiac death, and many patients who are currently selected to receive prophylactic therapy do not benefit since in practice they would not have gone on to experience a future mVA event in any case and therefore the treatment was never required. Typically, only around 15% of people who are selected to receive an ICD using the current guidelines receive benefit in the form of a potentially life-saving intervention from their device. This inability to accurately predict a risk of a future mVA therefore results in unproductive healthcare expense and avoidable exposure of patients to procedural risks.

[0008] There is therefore an urgent need for improved methods to assess individual risk of sudden onset of mVA to enable clinicians to target available prophylactic treatments more accurately to both reduce mortality due to arrhythmic sudden cardiac death and more efficiently use limited healthcare resources. The present invention seeks to address these shortcomings.

[0009] SUMMARY OF THE INVENTION

[0010] From a first aspect, the invention provides a data-processing apparatus for deriving a parameter value indicative of a subject’s risk of a future arrhythmia, the data- processing apparatus comprising a processing system, the processing system configured to: for a plurality of frames of a sequence of two-dimensional magnetic field data frames, each representing the rate of change of a magnetic field in the subject’s heart at a respective cardiac-cycle instant within a cardiac cycle, define a vector extending between a minimum-rate-of-change position in the two-dimensional frame, at which the rate of change of the magnetic field has a minimum value, and a maximum-rate-of- change position in the two-dimensional frame, at which the rate of change of the magnetic field has a maximum value; for a plurality of successive pairs of frames in the sequence, calculate a framepair angle change, by calculating the absolute value of the difference in angle of the vector defined for the earlier frame of the pair and the vector defined for the later frame of the pair; and calculate the parameter value by taking a combination of at least two of the frame-pair angle changes.

[0011] From a second aspect, the invention provides a method (e.g. a computer-implemented method) for deriving a parameter value indicative of a subject’s risk of a future arrhythmia, the method comprising: for a plurality of frames of a sequence of two-dimensional magnetic field data frames, each representing the rate of change of a magnetic field in the subject’s heart at a respective cardiac-cycle instant within a cardiac cycle, defining a vector extending between a minimum-rate-of-change position in the two-dimensional frame, at which the rate of change of the magnetic field has a minimum value, and a maximum-rate-of- change position in the two-dimensional frame, at which the rate of change of the magnetic field has a maximum value; for a plurality of successive pair of frames in the sequence, calculating a framepair angle change, by calculating the absolute value of the difference in angle of the vector defined for the earlier frame of the pair and the vector defined for the later frame of the pair; and calculating the parameter value by taking a combination of at least two of the frame-pair angle changes.

[0012] From a third aspect, the invention provides a system for deriving a parameter value indicative of a subject’s risk of a future arrhythmia, the system comprising: a magnetic field data collection module, arranged to obtain magnetic field data indicative of the rate of change of a magnetic field in the heart of a subject; and a data-processing apparatus as described herein above, and further below, wherein the two-dimensional frames of magnetic field data are derived from the magnetic field data obtained by the magnetic field data collection module.

[0013] From a fourth aspect, the invention provides software, or a signal or tangible medium bearing said software, comprising instructions which, when executed by a processing system, cause the processing system to: for a plurality of frames of a sequence of two-dimensional magnetic field data frames, each representing the rate of change of a magnetic field in the subject’s heart at a respective cardiac-cycle instant within a cardiac cycle, define a vector extending between a minimum-rate-of-change position in the two-dimensional frame, at which the rate of change of the magnetic field has a minimum value, and a maximum-rate-of- change position in the two-dimensional frame, at which the rate of change of the magnetic field has a maximum value; for a plurality of successive pair of frames in the sequence, calculate a framepair angle change, by calculating the absolute value of the difference in angle of the vector defined for the earlier frame of the pair and the vector defined for the later frame of the pair; and calculate the parameter value by taking a combination of at least two of the frame-pair angle changes.

[0014] It will be understood that a cardiac-cycle instant refers not to a particular time instant, but to a particular instant within (or fraction through) the recognised and well-known cardiac cycle, e.g. the start, end or middle of the QRS complex, or a fractional distance through the cardiac cycle, e.g. 0%, 30%, 50%, 100% etc. Changes in the angle of the vector are calculated for successive pairs of frames in the sequence, i.e. for pairs of frames that are adjacent in the sequence. A sequence will be understood as an ordered plurality of such frames. Thus, the sequence comprises a plurality of frames. The sequence may be effectively understood as a time-series, since each frame corresponds to a particular instant through the cardiac cycle, and they are ordered, albeit each frame may be made up of averaged data from several different time points or instants (each of which correspond to the same cardiac instant). Alternatively, each frame may correspond to a specific time-instant (i.e. being based on data collected only at that time instant). A particular frame, numbered n, which is not at an end of the sequence used for the calculation may be used both as the later frame of a pair, to calculate a first angle difference and also as the earlier frame of another pair (i.e. the next pair) to calculate a second angle difference. This may be the case for all frames apart from the two end frames, i.e. when all frames in the sequence are used.

[0015] The plurality of frames (for which vectors are calculated) may be considered as a set of frames of a sequence. The frame-pair angle changes may be calculated for each pair of the set. The parameter value may be calculated by taking a combination of each frame-pair angle change for the set (i.e. combining all the angle changes).

[0016] The parameter value is calculated by taking a combination of at least two of the framepair angles, by this it will be understood that there is a particular sequence of frames (i.e. this at least two referenced frames) on which all of the stated calculation stages are carried out. It will be understood that this referenced sequence may be a subset of a total sequence of data / frames that is obtained. Some of the recited steps, e.g. deriving the vector, may be carried out on additional frames which are not then included in the combination.

[0017] The vector may be defined for each frame of the sequence. A frame-pair angle change may be calculated for each successive pair of the pair of frames in the sequence. Each of the stated steps may be carried out for each frame or pair of frames of the sequence.

[0018] The vector for each frame is defined between a two-dimensional position of maximum value (i.e. a first extremum, a first pole) and a two-dimensional position of minimum value (i.e. a second extremum, a second pole). There may be a particular point in each two-dimensional frame which is the extremum (i.e. a single point of maximum / minimum value), in which case this may provide the maximum / minimum- rate-of-change position, or where multiple points each share the same maximum / minimum value the maximum / minimum-rate-of-change position may be a single position out of these plurality of points (e.g. a central point). The minimum value will be understood as referring to the lowest value for the rate of change, i.e. including negative values. It thus may be the largest negative value for the rate of change, where some values are negative. It thus extends between effectively the poles of the data (albeit of rate-of-change of magnetic field rather than the magnetic field itself). The angle of this vector may be referred to as the dipole angle.

[0019] The processing system may take any suitable form. It may comprise one or more of: a central processing unit, a graphics processing unit, a microcontroller, an ASIC, an FPGA, and any other discrete or integrated components or circuitry. It may comprise memory storing software instructions for instructing a processor to perform some or all of the steps disclosed herein. The processing system may be a single processing unit or device, or it may be distributed — e.g. comprising one or more servers. Processing may be carried out on a single device or may be shared across multiple devices in any appropriate way. For instance, one device may generate magnetic field data and a different device (possibly remote) may derive the two-dimensional frames and / or the parameter value from the generated data. In some embodiments, sampled magnetic field data may be collected by a first device (e.g. an inductive coil) and sent to a remote computer or server which applies some or all of the processing operations. The processing system may be a local processing system, or may be a remote processing system (i.e. local to the operator or other system components or remote from the operator or other system components) or it may include both a local part and a remote part. The processing system may be a cloud computing system.

[0020] The software may be provided on a non-transitory computer-readable medium, which may be a solid-state memory or a magnetic or optical storage medium. It may be stored in volatile memory (e.g. RAM) or non-volatile memory (e.g. flash memory, EEPROM, ROM). The software may be written in any language (i.e. computer language), e.g. C, C++, Java, etc. It may be compiled into machine code. The data-processing apparatus may further comprise one or more microprocessors. It may comprise memory, buses, internal peripherals, input / output peripherals, network interfaces, a power supply, etc. These components may be connected to the processing system.

[0021] The heart referred to may be a human heart, i.e. the subject may be a human.

[0022] Thus, it will be seen that, in accordance with the invention, by taking a combination of multiple angle changes in the defined vector at a series of points through the cardiac cycle to generate a numerical value of a parameter, spatio-temporal variations in the magnetic field emitted by a subject’s heart are assessed to derive an indication of the risk of that heart experiencing a cardiac event in future. This numerical parameter may provide a useful clinical insight for a clinician to take into account in assessing whether a particular patient would be likely to benefit from receiving prophylactic treatment e.g. through the fitting of an implantable cardioverter defibrillator (ICD). The clinician thus may take the derived value of this numerical parameter into account, alongside other clinical factors, in order to decide whether a particular patient is a good candidate to receive prophylactic treatment.

[0023] The Applicant has appreciated that taking a combination (e.g. a weighted sum) of angle changes in the vector defined between magnetic rate-of-change poles (i.e. at a series of cardiac-cycle instants) measures the temporal instability in the angle of this vector (the dipole angle) throughout the cardiac-cycle period across which the combination is calculated (e.g. across the QRS interval) and that this correlates strongly with a subject’s future arrhythmia risk, and therefore provides a useful clinical indication.

[0024] Deriving the parameter value using data representing the rate of change of a magnetic field is advantageous, compared to performing calculations based on magnitude data of the magnetic field itself, because this increases the impact of higher frequency components on the derived parameter value, which the Applicant has appreciated seem to be particularly relevant to arrhythmic risk. Furthermore, the rate-of-change data effectively has any background magnetic fields which are substantially constant filtered out of it automatically. This particular method of effectively quantifying the temporal instability of the dipole angle of the heart’s magnetic field is particularly advantageous in terms of its implementation since it does not require precise alignment or positioning relative to the subject. The parameter value is derived using change of the angle of the vector between magnetic poles (A0) so exact angular alignment of the sensor header to the subject is not necessary. Furthermore, the precise location of the heart relative to the magnetic field data collection device does not need to be known in order to carry out the calculation, provided that the heart is sufficiently close to the data collection device.

[0025] The combination may be a combination of the frame-pair angle changes for all pairs of the plurality of two-dimensional magnetic field data frames.

[0026] The combination may be a linear combination. A linear combination will be understood as a sum, which may be weighted or unweighted. Thus, in some embodiments, the (linear) combination is a weighted sum (i.e. in which each term, i.e. each frame-pair angle change, is multiplied by a corresponding weighting value. The use of weightings may allow certain frame-pair angle changes to have greater influence than others on the resulting parameter value, which, by using a suitable weighting, may improve accuracy of the output parameter value.

[0027] In some embodiments, the weighting of a particular frame-pair angle change of the combination (i.e. the weighting of the sum) is based on a relative magnitude value for at least one frame of the respective pair of frames from which the frame-pair angle change is calculated, wherein the relative magnitude value is indicative of the relative magnitude of the frame, i.e. relative to other frames in the sequence (optionally all of the other frames in the sequence).

[0028] In some embodiments, the processing system is further configured to, for each of the frames used to derive the frame-pair angle changes used in the combination (e.g. for each of the plurality of successive pairs of frames for which angle changes are calculated, optionally for each of the two-dimensional magnetic field data frames in the sequence), determine the relative magnitude value. The weighting of the sum advantageously helps to improve the accuracy of the derived parameter value since it reduces the effect on the value of angle changes which occur in low magnitude frames. It has been appreciated that when amplitudes (i.e. of rate-of- change of the magnetic field) are low, the angle of the defined vector can become more variable (i.e. jittery), which, if uncorrected, would cause a significant increase in the derived parameter value. The weighting therefore effectively reduces the impact of lower accuracy dipole angle determination due to low or variable signal to noise ratio which occurs in the lower magnitude frames. As explained in further detail below, this weighting also helps the parameter to distinguish between rapid rotation of the dipole angle, which is generally understood to be a normal behaviour, expected in a healthy heart, and a slower rotation across a larger number of frames (or of course movement back and forth of the angle) all of which result in a higher parameter value.

[0029] In some embodiments, the relative magnitude value for a particular frame is the magnitude difference divided by a scaling value, the magnitude difference being the difference between the value of the data frame at the maximum-rate-of-change position and the value of the data frame at the minimum-rate-of-change position, and the scaling value being the maximum magnitude difference of magnitude differences for all frames corresponding to (i.e. from which are calculated) the (at least two) framepair angle changes used in calculating the parameter value (e.g. in the plurality of successive pairs of frames in the sequence, optionally in the sequence of two- dimensional magnetic field data frames). This therefore scales the angle change for a particular pair of frames based on their magnitude in relation to the highest magnitude frame in the whole sequence that is used to calculate the parameter value.

[0030] The weighting for a pair of frames (i.e. a pair from which an angle change is derived) may be determined in any suitable manner from the relative magnitude value of each frame in that pair - e.g. by selecting one of these as the weighting for that pair, or combining them in a suitable manner, such as by averaging. In some embodiments, the weighting for a (i.e. each) particular pair of adjacent (i.e. successive) frames is the relative magnitude value of the later frame of the pair. Alternatively, the weighting for a particular pair of adjacent (i.e. successive) frames may be the relative magnitude value of the earlier frame of the pair. In some embodiments, the processing system is configured to receive magnetic field data (e.g. magnetocardiographic data) from a magnetic field data collection module and to derive from the magnetic field data the two-dimensional frames of magnetic field data; or the processing system is configured to receive the two-dimensional frames of magnetic field data from a magnetic field data collection module (and optionally the magnetic field data collection module is arranged to derive from the magnetic field data two-dimensional frames of magnetic field data, indicative of the rate of change of magnetic field in the subject’s heart at respective cardiac-cycle instants within a cardiac cycle). Similarly, in some embodiments, the method further comprises receiving (e.g. by the magnetic field data collection module or the processing system) magnetic field data relating to the subject’s heart, and deriving from the magnetic field data a plurality of two-dimensional frames of magnetic field data indicative of the rate of change of magnetic field in the subject’s heart at respective cardiac-cycle instants within a cardiac cycle.

[0031] The received magnetic field data may be rate-of-change data, representing the rate of change of a magnetic field within the subject’s heart (i.e. at a respective cardiac cycle instant). Alternatively, the received magnetic field data may be magnitude data, representing the magnitude of a magnetic field within the subject’s heart (i.e. at a particular cardiac-cycle instant), and therefore indirectly indicative of the rate of change of the magnetic field. The method may further comprise differentiating the magnitude data to derive rate-of-change data, representing the rate of change of a magnetic field within the subject’s heart. The processing system may likewise be arranged to carry out such a step.

[0032] It will be appreciated that references to receiving data may (unless otherwise specified) comprise receiving the data from an external source, e.g. over a cable or network interface, or may comprise receiving the data effectively internally within the processing system, following some previous processing by the processing system, or receiving the data by retrieving it from memory within the data-processing apparatus (e.g. within the processing system). It will therefore be appreciated that according to the claimed method and processing system it is not important whether the preprocessing of raw data to derive the magnetic field data and / or the two-dimensional frames of magnetic field data is carried out by the same processor which carries out the subsequent parameter value calculation or is carried out by one or more separate processors, and the processed data then provided to the processing system for area calculation. Thus, the modules referred to, i.e. the magnetic field data collection module and the heart rate data collection module (described below), are conceptual definitions, but need not necessarily correspond to units that are entirely separate to the processing system, for example the processing system may carry out some of the processing that is referred to as being carried out by the module, and vice versa.

[0033] In some embodiments, the method comprises obtaining from the subject (e.g. using a magnetic field data collection module) magnetic field data indicative of the rate of change of magnetic field in the subject’s heart. As set out above, this data indicative of the rate of change of the magnetic field may be data representative of the rate of change (i.e. obtained directly) or may be data representative of the magnitude of the magnetic field, which may be processed (e.g. differentiated) to give the rate of change data.

[0034] In some embodiments, the two-dimensional magnetic field data frames are derived from data collected by a magnetocardiographic (MCG) data collection device. Thus, in some embodiments, the magnetic field data collection device is a magnetocardiographic (MCG) data collection device.

[0035] It is advantageous that clinically useful information is extracted based on magnetocardiography (MCG) data, since magnetocardiography is non-invasive, rapid and (for at least some types of MCG sensor) low cost to deploy. MCG is furthermore advantageous because it’s non-contact and radiation free, and presents no known risk to the subject on whom the measurements are carried out. It is also able to record data through clothing, and the signals are not distorted by a subject’s body composition (as experienced with ECG).

[0036] As set out above, in some embodiments, the method comprises obtaining from the subject magnetic field data representative of the rate of change of a magnetic field in the subject’s heart. The magnetic field data collection device (e.g. MCG device) may comprise one or more inductive coils. Such a device straightforwardly obtains data representative of the rate of change of magnetic fields, thus requiring less subsequent processing of the data in order to implement the method described above. Such devices are also available at very low cost, and do not require costly shielding in order to work effectively. Induction-coil-based magnetometers are lower cost both to produce and to run than other known magnetometer devices. For example, traditional SQUID devices require liquid helium as a coolant and therefore have high cost-per-use, and some quantum magnetometer sensors are known which may have a lower cost-per- use but which are very expensive to produce.

[0037] In other embodiments, the method comprises obtaining from the subject magnetic field data representative of the magnitude of a magnetic field in the subject’s heart, and differentiating the obtained data to derive data representative of the rate of change of a magnetic field in the subject’s heart.

[0038] In some embodiments, the sequence of two-dimensional magnetic field data frames is derived from data obtained using a sensor array. Thus, in some embodiments, the magnetic field data collection module comprises a sensor array (e.g. an array of inductive coils). In some embodiments, the method comprises interpolating data from at least two sensors in the array (optionally for all sensors in the array) in order to derive each of the two-dimensional magnetic field data frames. The magnetic field data collection module or the processing system (or an additional processing system) may be arranged to carry out this interpolation.

[0039] In some embodiments, the magnetic field data collection device is arranged to collect magnetic field data at a sample rate of over 200 Hz, optionally over 600 Hz. The Applicant has appreciated that this is a sufficiently high sample rate to capture substantially all of the frequency information which is relevant to predicting a future arrhythmia risk. The magnetic field data collection device may be arranged to collect magnetic field data at a sample rate of over 1000 Hz, optionally over 2000 Hz.

[0040] In some embodiments, the magnetic field data collection device is arranged to collect magnetic field data at a sample rate of approximately 2100 Hz. It has been appreciated that this sample rate is particularly advantageous because when divided by the frequency of mains electricity for the UK (which is 50 Hz) and for the US (which is 60 Hz), this gives an integer value. As a result, contributions to the magnetic field data arising due to mains electricity where the measurement is carried out in one of these countries can be removed particularly straightforwardly, e.g. by applying windowed filters. In some embodiments, the method further comprises applying filtering to received (or obtained) magnetic field data and / or applying filtering to the two-dimensional magnetic field data frames. Thus, deriving from the magnetic field data the plurality of two- dimensional frames of magnetic field data indicative of the rate of change of magnetic field in the subject’s heart at respective cardiac-cycle instants within a cardiac cycle may comprise applying one or more filters to the data, for example a bandpass filter up to 100 Hz, and / or a time-window filter.

[0041] The two-dimensional frames of magnetic field data may be averaged over a plurality of cardiac cycles (i.e. to give a frame of magnetic field data for a particular cardiac-cycle instant which is an average of data collected at the same corresponding instant within two or more cardiac cycles), or may relate to a single cardiac cycle. The two- dimensional frames of magnetic field data may be averaged over the plurality of cardiac cycles by summing data from (corresponding points in) each respective cardiac cycle. For example, it may be averaged over at least 100 cardiac cycles, optionally at least 300, further optionally at least 600. This may improve the reliability of the obtained values, by improving the signal to noise ratio of the data. If data from several cardiac cycles is summed, the inherent random noise doesn’t add coherently over multiple heartbeats, unlike the heart signals which do sum coherently, thus becoming larger than the noise signal when summed over a sufficiently high number of heartbeats. Thus, the two-dimensional frames of magnetic field data are said to correspond to a particular cardiac-cycle instant, i.e. a particular point through the cardiac cycle, since a particular frame corresponds to (approximately) the same point through the cardiac cycle although all the data comprised in the frame may not necessarily correspond to a single point in time.

[0042] Thus, in some embodiments, the method further comprises averaging magnetic field data relating to at least two different time-instants, corresponding to respective cardiac cycles, to provide each two-dimensional magnetic field data frame of the sequence, wherein for each respective frame, each time-instant (i.e. from different cardiac cycles) corresponds to the same cardiac-cycle instant (i.e. within their respective cardiac cycle). The processing system may similarly be arranged to carry out this timeaveraging. This time-averaging, i.e. over several cardiac cycles, may improve the quality of the data and therefore the reliability of the derived parameter value. This averaging may be achieved in any suitable manner. For example, where the heart rate is approximately constant throughout data collection, the data for averaging may be identified simply based on the frame number and / or aligning points a certain time separation apart in the data (i.e. based on a detected heart rate). Alternatively, each heartbeat may be identified by identifying a particular corresponding point of each heartbeat (i.e. a fiduciary point), such as the start of the respective heartbeat or the peak of the R wave of each heartbeat, and then the heartbeat data may be defined as a certain time window each side of this point (e.g. 1 second either side). The alignment point (i.e. fiduciary point) may be identified from the magnetic field data itself (e.g. for at least certain types of magnetometer, including SQUID magnetometers), or alternatively may be identified from heart rate data (e.g. ECG data). This is advantageous since particularly clean magnetic field data may be needed to identify the correct fiduciary point in the heartbeat (e.g. the peak of the R wave). Identifying each heartbeat by a recognisable point within it (e.g. the peak of the R wave), and then classifying a specific time window around that point as a single heartbeat, accounts for any variation in heart rate (i.e. the separation between beats) which may occur during data collection. It will be understood that a fixed time window around the fiduciary point can be used to define each heartbeat since the heartbeat itself is generally of a very consistent length.

[0043] In some embodiments, the parameter value may be derived using a sequence of two- dimensional magnetic field data frames wherein at least two (optionally a plurality, or all) of the frames in the sequence correspond to cardiac-cycle instants within the QRS period of the cardiac cycle, i.e. all of the frame-pair angle changes used in calculating the parameter value may be calculated from (pairs of) frames which correspond to cardiac instants within the QRS period. Thus, the sampling for data used to derive the frames, and therefore the parameter value, is done across at least the QRS period. The frames in the sequence may substantially span the QRS period of the cardiac cycle, i.e. be from throughout the QRS period. Thus, the parameter value may be derived using a sequence of two-dimensional magnetic field data frames wherein the frames in the sequence correspond to cardiac-cycle instants from within both the QR period of the cardiac cycle and RS period of the cardiac cycle. Data from this period of the cardiac-cycle has been appreciated as being particularly important to determining arrhythmia risk. Thus, the set or subset referenced above may be the set or subset of frames (or data) from within, or corresponding to (i.e. spanning), the QRS period of the cardiac cycle.

[0044] In some embodiments, the magnetic field data and / or the frames are correlated with the cardiac cycle (i.e. to identify which instant they correspond to). This enables filtering of the magnetic field data, or the frames, according to the cardiac-cycle instant that they correspond to (e.g. to take into account only data from a certain part of the cardiac cycle).

[0045] The two-dimensional magnetic field data frames may be correlated in some suitable manner with the cardiac cycle. This correlation or alignment may be carried out by the processing system and similarly may form part of the claimed method. This may be carried out by correlating the magnetic field data with the cardiac cycle. Further details of this process are described below.

[0046] The method may further comprise deriving heart rate data from obtained or received magnetic field data. This advantageously does not require any further data to achieve the correlation, however it can be complex to achieve.

[0047] Alternatively, the method may further comprise receiving or obtaining heart rate data from the subject (e.g. during collection of the magnetic field data), for example using an electrocardiogram (ECG) device. Thus, in some embodiments, the processing system is further configured to receive heart rate data from the subject and to correlate the two-dimensional magnetic field data frames with the cardiac cycle using the heart rate data.

[0048] In some embodiments the system further comprises a heart rate detection module, arranged to detect heart rate data (e.g. the heart rate) of a subject to enable correlation of the magnetic field data with respect to the cardiac cycle. The heart rate detection module may be an electrocardiogram (ECG) device. The heart rate may be detected over the same time period (at least overlapping time periods) as the magnetic field data. Preferably, the heart rate data is collected at the same time as (i.e. simultaneously with) collecting the magnetic field data (e.g. across several cardiac cycles). The method may further comprise using the heart rate data to isolate magnetic field data from a particular heartbeat cycle (or a plurality of heartbeat cycles). The heart rate data may allow the phases within a heart cycle to be identified (and their corresponding data isolated), such as to determine the magnetic field behaviour of the heart in a particular part of the cardiac cycle, e.g. the QRS period. For example, the different phases of the cardiac cycle are apparent from an ECG trace, thus allowing a particular period, such as the QRS period, to be identified accurately and straightforwardly. The heart rate data also enables rate-adjustment of the magnetic field data, as described above.

[0049] According to the invention, a parameter value is derived, where the value of the parameter is indicative of a subject’s risk of experiencing an arrhythmia in future. A higher value of the parameter may be indicative of a higher risk of an arrhythmia in future.

[0050] A threshold value may be selected, wherein if the parameter has a value above the threshold value, then this is considered to be indicative of a high risk of an arrhythmia occurring for the subject in future.

[0051] As described above, the parameter value may be derived using a sequence of two- dimensional magnetic field data frames wherein some (or all) of the frames in the sequence correspond to cardiac-cycle instants within the QRS period of the cardiac cycle.

[0052] In some embodiments, the method comprises deriving a secondary parameter value, based on the (first) parameter value, wherein the secondary parameter value is derived by dividing the (first) parameter value by a time value, indicative of the length of the QRS period, i.e. for the cardiac cycle(s) over which the magnetic field data (from which the frames are derived) is collected, for the particular subject. The processing system may similarly be configured to carry out these steps. Thus, where the data relates to a single cardiac cycle, the time value may be the length of the QRS period (e.g. in seconds) for that cardiac cycle. Alternatively, for data which is averaged over more than one cardiac cycle, the time value may be the average length of the QRS period across that set of cardiac cycles, or across a sub-set, or the length of the QRS period in a particular representative one of the cardiac cycles. This secondary parameter value therefore effectively accounts for differences in the length of the cardiac cycle (i.e. heartbeat) between different subjects, and therefore may in some cases provide a more reliable indication of future arrhythmia risk.

[0053] In some embodiments, the processing system is arranged to output the parameter value. Outputting the value may comprise displaying the value on a display screen and / or transmitting the value to an external device or location (e.g. saving it to cloud storage). Thus, in some embodiments, the system (i.e. the system comprising the magnetic field data collection module) further comprises an output module (e.g. a display device) arranged to output or display the output parameter value.

[0054] In some embodiments, the data-processing apparatus further comprises an output module. The processing system may be arranged to output the parameter value and the output module may be arranged to display the value.

[0055] In some embodiments, the processing system is configured to derive a risk indicator, based on (at least) the calculated parameter value (and optionally also based on one or more additional factors). For example, the processing system may compare the calculated parameter value to one or more threshold values, and may, for example, categorise a patient’s risk of a future arrhythmia as low, medium or high. The processing system may be arranged to output the risk indicator (e.g. to the output module). The output module may be arranged to display the risk indicator.

[0056] Thus, in some embodiments, the output module may be arranged to display both the parameter value and to display the risk indicator.

[0057] It will be appreciated that for any steps described above that the processing system may be configured to perform, the invention similarly extends to software, or a signal or tangible medium bearing said software, comprising instructions which, when executed by a processing system, cause the processing system to perform such steps. Furthermore, for any steps which may form part of the method of the invention, the invention further extends to a processing system or data-processing apparatus configured to carry out such steps, and vice versa any steps which the processor system or data processing apparatus may be configured to carry out may likewise form steps of the method of the invention. Features of any aspect or embodiment described herein may, wherever appropriate, be applied to any other aspect or embodiment described herein. Where reference is made to different embodiments or sets of embodiments, it should be understood that these are not necessarily distinct but may overlap.

[0058] BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Certain preferred embodiments of the invention will now be described, by way of example only, with reference to the accompanying drawings, in which:

[0060] Figure 1 is a schematic diagram of a system according to an embodiment of the present invention;

[0061] Figure 2 is a flow diagram representing a method according to an embodiment of the present invention;

[0062] Figure 3 is a schematic side view of a magnetic field data collection device, which may be used in the method of Figure 2;

[0063] Figure 4 is a schematic diagram of an array of sensors which may be used in the arrangement of Figure 3;

[0064] Figure 5 is a plot illustrating an exemplary two-dimensional magnetic field data frame as used in the described method;

[0065] Figure 6 is a graph of magnitude of the rate-of-change value, seen along the direction indicated by the vector in Figure 5;

[0066] Figure 7 is a graph representing the behaviour of a weighting value across the QRS interval of the heart cycle, which may be used in the method of Figure 2;

[0067] Figure 8 is a graph showing an ECG trace;

[0068] Figures 9 and 10 are graphs demonstrating example behaviour of the dipole angle across the QRS period of the cardiac cycle for two subjects;

[0069] Figure 11 is a table showing the mean parameter value and secondary parameter value for a plurality of subjects, separated according to whether or not the subjects received therapy from their implantable cardioverter defibrillators;

[0070] Figure 12 represents an ROC curve for the parameter value data of Figure 11;

[0071] Figure 13 represents an ROC curve for the secondary parameter value data of Figure 11 ;

[0072] Figure 14 is a table showing the mean parameter value and secondary parameter value for a plurality of subjects, separated according to whether or not the subjects received a shock from their implantable cardioverter defibrillators; Figures 15 represents an ROC curve for the parameter value data of Figure 14; and

[0073] Figure 16 represents an ROC curve for the secondary parameter value data of Figure 14.

[0074] DETAILED DESCRIPTION

[0075] Figure 1 is a system 1 according to an embodiment of the present invention, arranged to obtain magnetocardiography data from a subject and to derive from that data a parameter value indicative of a risk for that subject of a future occurrence of a malignant cardiac arrhythmia (mVA). Such an mVA is a sudden onset condition which can be potentially fatal.

[0076] The system 1 includes a magnetocardiography (MCG) data collection device 2 and an electrocardiography (ECG) data collection device 4, which are each arranged to collect data from a subject, in a known manner. The magnetocardiography (MCG) data collection device 2 will be understood as an exemplary magnetic field data collection module, and the electrocardiography (ECG) data collection device 4 will be understood as an exemplary heart rate data collection module. The data collected by each is supplied to a processing system 6, as represented by the arrows seen in Figure 1. This connection may be direct, or it may be indirect, i.e. the data collected may be stored and then separately uploaded to the processing system 6. The data transfer may be via a computer network to a computer / server on a local network or it may be to a ‘cloud’ server at a remote location accessed via the internet.

[0077] The processing system 6 processes the received data according to the method described below, so as to derive a parameter indicative of a risk for that subject of a future occurrence of a malignant cardiac arrhythmia (mVA). This parameter is then output by the processing system 6. This may be output to any suitable location, including for example, a display device 8, which may, for example, be the screen of a computer or mobile electronic device.

[0078] As described above, the processing system 6 may be a single processing system or a distributed processing system. It may be a partly (or fully) cloud-based processing system. The processing system 6 may be remote or local (i.e. to the data collection devices). Thus, it may be an on-board computer integrated with the MCG device 2 or the ECG device 4 (or both) or it may be a separate processing system 6 (e.g. offline and not connected to the data collection modules).

[0079] During normal operation of a mammalian heart, rhythmic electrical signals pass through the heart to co-ordinate the contractions of the cardiac muscles which pump oxygenated blood around the body. The correct timing and location of these electrical currents in the heart is critical to the correct function of the heart. Defects of cardiac electrical function are described as ‘arrhythmias’, they can result in reduced cardiac function, or even total loss of cardiac function. A severe arrhythmia that results in loss of cardiac function is a ‘cardiac arrest’ which typically results in death.

[0080] It is widely thought that some people have an underlying elevated risk of arrhythmias, due to often undetected, localised defects in the timing and location of the cardiac current impulses that pass through the cardiac muscle (known as ‘proarrhythmic substrates’). These defects, most commonly ‘re-entrant currents’, typically result from changes in the myocardial tissue due to disease or inherited conditions.

[0081] Importantly, these localised electrical defects / proarrhythmic substrates can often be asymptomatic and undetected using existing tools such as electrocardiography (ECG). However, when suitably triggered, they can nucleate a sudden catastrophic breakdown in cardiac electrical function, a malignant arrhythmia, leading to sudden cardiac arrest (SCA) and, if untreated, sudden cardiac death (SCD).

[0082] Examples of morphological defects that act as proarrhythmic substrates include localised scaring and tissue remodelling due to ischemia and various forms of cardiomyopathy where the cardiac muscle has become stretched, thickened, stiffened or inflamed. Examples of triggers that, when combined with the presence of a proarrhythmic substrate in the heart, can result in SCA include exercise, alcohol, recreational drug use, electrolyte imbalance and dehydration.

[0083] By monitoring the electrical signals in the heart, in particular using the MCG data collection device 2, using the method described below, the likely presence of such (otherwise undetected) defects can be more accurately inferred and therefore a subject’s risk of arrhythmias can be more accurately predicted. Figure 2 is a flow diagram illustrating a method according to an embodiment of the present invention, which may be carried out by the system illustrated in Figure 1.

[0084] In a first step, 200, the MCG data collection device 2 and the ECG data collection device 4 are operated to collect data from the subject over several cardiac cycles.

[0085] Next, at stage 202, the data collected by each magnetic detector of the MCG data collection device 2 is averaged over at least two cardiac cycles. This produces a single “averaged” trace corresponding to each detector element in the array.

[0086] In order to average the data over several cardiac cycles, the data collected by the ECG data collection device 4 is used to determine the point in the cardiac cycle which each part of the data corresponds to. Then, data relating to the same point through the cardiac cycle (e.g. sample 100) from different heartbeats is identified, i.e. so as to match up data from different heartbeat cycles which relates to the same time point through the respective sample. An average of these corresponding data sets is then taken (e.g. by taking the mean). In this particular example, the peak of the R-wave, as seen in the ECG data, is used as a fiduciary reference point, to match up corresponding points in the MCG data from different heartbeats.

[0087] Also in stage 202 (before or after the averaging across cardiac cycles, as appropriate), for each frame of the collected data, the data received by each of the magnetic detectors in the MCG data collection device 2 is interpolated, to give one interpolated, two-dimensional data frame corresponding to each sampling time in the sampling window. An example two-dimensional magnetic field data frame is illustrated in Figure 5, and described in greater detail below.

[0088] The result of these two steps is a sequence of two-dimensional frames of magnetic field data, each representing a spatial and temporal average for a particular point through the cardiac cycle, i.e. a cardiac-cycle instant. In this example, the sampling frequency is 2100 Hz. Thus, the series of collected frames represent cardiac-cycle instants every 0.5 ms through the cardiac cycle. The values of each data-frame represent the rate of change of the magnetic fields in the detected region, at that particular cardiac-cycle instant. Next, at step 204, certain data is extracted for each frame of the data (although this may only be done for a subset of the total data collected, i.e. just for a portion of the cardiac cycle). In particular, a vector, referred to as the magnetic vector, is extracted which extends between a maximum value of the rate-of-change data and a minimum value of the rate-of-change data (i.e. the two poles of the magnetic field data), and also a magnitude value is extracted, which is the absolute difference between the maximum value and the minimum value (i.e. between the magnitude of the two poles of that frame). This process is described further below with reference to Figure 5.

[0089] Then, at step 206, the angle of the magnetic vector is determined for each frame. Since the calculation is ultimately determined based on the change in angles of this vector between adjacent frames, the absolute value of the angle can be determined relative to any reference line, provided that this is consistent across all of the frames. This also means that the angle of the scanner relative to the subject does not need to be consistent across different subjects since the absolute angle is not important in this method.

[0090] At step 208, the absolute change in angle of the magnetic vector between each pair of successive vector pairs in the sequence is calculated, i.e. A0n= 0n- 0n-i etc.

[0091] At step 210, a weighting is derived for each angle change A0n. This weighting is calculated by dividing the magnitude difference for the second frame of the pair (i.e. the “n” frame) by the maximum magnitude difference observed across the (averaged) cardiac cycle, i.e. in all of the frames or selected subset of frames. The magnitude difference is explained below with reference to Figure 6, and the weighting is explained in greater detail below with reference to Figure 7.

[0092] Finally, at step 212, a value for the parameter is calculated by calculating the weighted sum of the angle changes, where each angle change is weighted by multiplying it by the corresponding weighting derived for that angle change at step 210. In particular, in this example, the average is calculated over frames falling specifically within the QRS period of the cardiac cycle, as explained in greater detail below. The parameter may be referred to as a “Peak Rotation Score”. In a further optional step 214, a secondary parameter may be calculated. This secondary parameter, which may be referred to as an “Angular Dynamics score”, is derived from the “Peak Rotation Score”. The secondary parameter value is derived by dividing the parameter value derived in step 212 by a time value, indicative of the length of the QRS period e.g. averaged over the cardiac cycle(s) over which the magnetic field data (from which the frames are derived) is collected. In this example, the QRS duration is the average QRS duration derived from the ECG data, which is collected simultaneously with the MCG data (i.e. across the same heartbeats).

[0093] In this particular example, the earlier steps 202-208 are also carried out only on those frames within the QRS period of the cardiac cycle. However, in other embodiments, one or more of the earlier steps may be carried out on additional data, e.g. all the data, across the whole cardiac cycle, or a fraction of the cardiac cycle, and then the sum (at stage 212) to derive the parameter value is carried out only on a subset of the data frames, relating to the QRS period.

[0094] The frames corresponding to the QRS period of the cardiac cycle may be identified based on data collected by the ECG data collection device 4, as explained in greater detail below with reference to Figure 8.

[0095] Several of these steps are explained in greater detail below with reference to Figures 3 to 8.

[0096] The process of obtaining data, as set out at step 200, and of averaging and interpolating this data, as set out at step 202, can be further understood with reference to Figures 3 and 4.

[0097] Magnetocardiography relies on the fact that electrical charges moving through the heart generate an associated weak, time varying magnetic field, with the magnitude of the magnetic field being proportional to the size of the electrical current that is flowing and being inversely proportional to the distance from the flowing charge to the detector. The direction of the magnetic field is determined by the direction of charge flow (direction of B field determined using the ‘right-hand grip rule’). This time-varying magnetic field produced by the heart’s electrical currents extends outside of the chest and can be measured using the MCG data collection device 2, which is a magnetic field detector.

[0098] Figure 3 represents an exemplary MCG data collection device 2, in use collecting magnetic field data from the chest region of a subject 10. The MCG data collection device 2, also referred to as a magnetometer system, comprises an induction coil 40 coupled to a detection circuit 41 that may contain a number of components.

[0099] The detection circuit 41 may comprise a low impedance pre-amplifier, such as a microphone amplifier, that is connected to the coil 40, and one or more filters, e.g. one or more a low pass filters, one or more high pass filters, one or more band pass filters, and / or one or more notch filters e.g. to remove line noise (e.g. 50 or 60 Hz and harmonics). The current output from the coil 40 is processed and converted to a voltage by the detection circuit 41 and provided to an analogue to digital converter (ADC) 42 which digitises the analogue signal from the coil 40 and provides it to a data acquisition system 43.

[0100] A biological signal that is correlated to the heartbeat (e.g. an ECG trigger from the ECG data collection device 4) from the test subject may be used as a detection trigger for the digital signal acquisition, and the digitised signal over a number of trigger pulses is then binned into appropriate signal bins, and the signal bins overlaid or averaged, by the data acquisition unit 43. Other arrangements would, however, be possible.

[0101] The coil 40 and detection circuit 41 may be arranged such that the coil 40 and the preamplifier of the detection circuit 41 are arranged together in a sensor head or probe which is then joined by a wire to a processing circuit that comprises the remaining components of the detection circuit 41. Connecting the sensor head 20 (probe) and the processing circuit by wire allows the processing circuit to be spaced from the sensor head (probe) in use. With this magnetometer, the sensor head (probe) will be used as a magnetic probe by placing it in the vicinity of the magnetic fields of interest.

[0102] In this example, the MCG data collection device 2 (in particular the inductive coil 40 shown in Figure 3) in fact comprises an array 60 of detection coils 61 (magnetic sensors), which are positioned spread apart from each other, as shown in Figure 4, and which may be placed over a subject's chest to measure the magnetic field of a subject's heart at sampling positions over the subject's chest.

[0103] Figure 4 shows an exemplary regular rectangular array of detection coils 61. In this example, each coil 61 of the array 60 is configured as described with respect to Figure 3. The output signals from the respective coils 61 can then be combined (e.g. by interpolating) and used appropriately to generate a magnetic scan of the subject's heart. In this particular example, interpolation of data from the different coils 61 is achieved by first using a coarse 2D cubic spline interpolation followed by a fine Radial Basis Function (RBF) interpolation. It will be appreciated that in other examples, the interpolation may be done differently. Other array arrangements could be used, if desired, such as hexagonal arrays, circular arrays, irregular arrays, etc.

[0104] The array of these magnetic detectors is used to generate a sequence of two- dimensional magnetic field data frames (an example data frame of which is shown in Figure 5), showing how the magnitude and distribution of the cardiac magnetic field across the chest varies during the course of a cardiac cycle.

[0105] In this example, the magnetic detectors 61 detect the rate of change of the magnetic field. It has been appreciated that such detectors may generally be created and operated at much lower cost than detectors which detect the magnitude of the magnetic field. Furthermore, the rate of change of magnetic field data is then immediately available for the further calculations described above in further detail. In other examples, magnetic detectors may be used which detect the absolute magnitude of the B-field itself. Rate of change data may then be obtained by differentiation of the magnetic field magnitude data. However, in this case highly effective magnetic shielding is required since the weak MCG signal of the heart is much smaller than other environmental magnetic fields, including the earth’s magnetic field.

[0106] In practice, the two-dimensional magnetic field data frame 500, i.e. MCG map - obtained using the system described with reference to Figures 3 and 4 - of a healthy individual at a time point in the cardiac cycle corresponding to the main ventricular contraction (electrical depolarisation of the left ventricle measured as the ‘QRS complex’ in an ECG) generally comprises effectively a magnetic dipole with a first area 502 of magnetic field coming out of the plane of the chest (a positive pole, shown by a first shaded region in Figure 5) and a second area 504 oriented into the plane of these chest (a negative pole, shown by a first shaded region in Figure 5).

[0107] The positions and magnitudes of these magnetic poles 502, 504 vary during the cardiac cycle in response to the changing magnitude and direction of the cardiac current, and by deriving a parameter based on their movements, according to the present invention, disruption to these cardiac currents can effectively be quantified giving an indication of whether an arrhythmia is likely for a given subject in future (since this risk is closely linked to the behaviour of the cardiac currents).

[0108] At step 204, vector and magnitude data are extracted from each frame 500 in the part of the cardiac cycle which is of interest, or optionally for all frames 500 collected or derived.

[0109] In order to do this, both the maximum-rate-of-change position 506 and the minimum- rate-of-change position 508 are identified for a given frame. This may be done in either order or simultaneously. The maximum-rate-of-change position 506 is the point (or a point, since there may be more than one point of the same maximum magnitude) within the two-dimensional frame at which the rate of change of the magnetic field has a maximum value. Where a region all has the same maximum value, the maximum- rate-of-change position 506 is a point within that region, e.g. a central point. Similarly, the minimum-rate-of-change position 508 is a point in the two-dimensional frame at which the rate of change of the magnetic field has a minimum value.

[0110] Next, a vector 510 is defined which extends from the a minimum-rate-of-change position 508 to the maximum-rate-of-change position 506. This vector 510 may be referred to as a dipole vector. It will be understood that the vector used for the method could alternatively extend from the maximum-rate-of-change position 506 to the minimum-rate-of-change position 508, i.e. in the opposite direction.

[0111] Then, at step 206, the angle 0, indicated by reference numeral 512 of the vector 510 is determined for each frame. In this example, the angle 0 512 of the vector 510 is determined relative to the horizontal direction 514 of the frame. However, it will be understood that since relative angle changes are used in determining the parameter, any suitable reference line may be used for defining the angle, provided it is used consistently across all the frames.

[0112] Step 204, discussed above, also includes extracting magnitude data from each frame. This is explained further with reference to Figure 6.

[0113] Figure 6 is a graph of magnitude of the rate-of-change value (i.e. the value of each point of the MCG map), seen along a line defined by the direction of the vector 510 in Figure 5 (i.e. a line extending the vector 510 in each direction), with left to right on the graph being along the direction in which the arrow points. The values at the poles 504, 502 can be seen by one positive peak and one negative peak. At step 204 the magnitude of this difference 516 is calculated for a particular frame, by subtracting the negative peak value 503 (the minimum) in that frame (i.e. the value at the minimum- rate-of-change-position 508) from the positive peak value 505 (the maximum) in that frame (i.e. the value at the maximum-rate-of-change-position 506).

[0114] At step 210, the weighting value for each frame, referred to as the relative magnitude value, is calculated based on these derived maximum difference values 516. This is done by dividing the maximum difference value 516 for a given frame by the maximum difference value observed across all of the data frames of interest. In this particular example, the data frames of interest are those in the QRS interval, and therefore the maximum difference value here refers to the maximum difference value observed across the QRS period (i.e. the maximum across all frames within the QRS period). Generally speaking, this would be expected to be the maximum also across the whole cardiac cycle, since the most significant current changes (and therefore magnetic field changes) occur in the QRS period, but it has been appreciated that there is a risk that outside the QRS period interference in the collected data could give rise to a larger difference value than in the QRS period.

[0115] For a given frame, the weighting is therefore proportional to the difference between the lowest rate of change value occurring in that frame and the highest rate of change occurring in that frame. The weighting for a pair of frames (i.e. from which an angle change is derived) may be determined in any suitable manner from the weighting value of each frame in that pair - e.g. by selecting one of these as the weighting for that pair, or combining them in a suitable manner, such as by averaging. For a healthy heart, the current builds up and then diminishes during the QRS period of each heart cycle in a predictable and well understood manner (and therefore so too does the magnetic field produced by this current). Therefore, if the magnitude of the magnetic field is observed, this will increase positively to a peak, and then decrease back to a baseline value, through the QRS period, whilst also moving around since the direction of the current changes through the heart cycle.

[0116] As a result, when the rate of change of the magnetic field is observed, the magnitude of the peak difference 516 is expected to initially be relatively large, since the field is increasing rapidly and so there is a large increase at the positive pole and a large decrease at the negative pole. The peak difference 516 slowly drops as the magnetic field builds more slowly, and then drops to close to (or exactly) zero as the magnetic field peaks approximately at the midpoint of the heart cycle, and so the field at each pole stays approximately constant. The magnetic field then starts to decrease, so there is a large decrease at the positive pole and a large increase at the negative pole, across a relatively short time, giving a high difference in peak rate. The peak difference therefore increases again, albeit the poles are reversed, because the previously “positive” pole now has a negative rate of change and vice versa. Thus, a rapid rotation of the dipole angle is expected to occur in the rate of change data, at this point in the cycle. This behaviour is represented in the example of Figure 9, which is discussed further below. This behaviour causes the peak difference value 516, for a healthy heart, to vary across the QRS interval in line with the behaviour shown in Figure 7.

[0117] Since the weighting for each pair is based on the relative magnitude value, and this relative magnitude value is proportional to the peak difference value 516 for at least one frame of the respective pairs (e.g. the later frame in the pair), the weighting values through the QRS interval are expected to have approximately the shape of the curve 700 shown in Figure 7, for a healthy heart. This curve shows the relative magnitude value along the y-axis 704, over time (shown along the x-axis 702), for a single representative QRS interval of the heartbeat cycle. The relative magnitude value has a maximum value of 1, which occurs where the peak difference for a frame is equal to the maximum peak difference. It can therefore be seen that changes to the dipole angle 512 occurring at points of the QRS interval with high rates of change are weighted more highly in the sum than changes occurring at the point at which the dipole angle shifts around, and the magnitudes are low. This is advantageous since this rotation behaviour is expected for a healthy heart and therefore its contribution to the parameter value, indicative of issues, should be minimal. Furthermore, there is more noise in the obtained data and in the derived dipole angles in this part of the heart cycle, and therefore a risk of angle changes detected in this part of the cycle of skewing the parameter value, and decreasing its accuracy. The weighting reduces the significance of contributions from this part of the cycle, thereby improving accuracy.

[0118] Then, at stage 212, the parameter value is calculated by taking the weighted sum of the angle changes over a set of frames. In this embodiment, the weighted sum may be represented as: where A0nis the angle change from the n-1 frame (0n-i) to the n frame (0n), such that as described above.

[0119] Peak height difference for the n frame is the maximum difference value 516 for frame n (i.e. the later frame of the pair).

[0120] As can be seen, the sum is taken across all frames lying within the QRS period, i.e. for all frame numbers n where n is within the QRS period.

[0121] At step 214, a secondary parameter is calculated, which is the parameter value, derived according to the equation above, divided by the length of the QRS period: It will be appreciated that in order to sum values specifically over the QRS period (i.e. excluding other parts of the heartbeat cycle) and to determine the length of the QRS period in order to derive the secondary parameter value, the QRS period of the heartbeat cycle (i.e. cardiac cycle) must be identified within the collected data (e.g. identifying which frames correspond to points within the QRS period).

[0122] It is possible to identify the QRS period directly from the magnetic field data collected from the patient. But this is difficult and requires complex calculation, particularly where rate-of-change of magnetic field data is collected.

[0123] Alternatively, in this particular example, an ECG data collection module 4 is arranged to collect data from the subject simultaneously with collection of the magnetic field data. As described below, it is well known how to identify the QRS period (also referred to as the QRS complex) in ECG data, and this can therefore be used to identify which frames correspond to the QRS period.

[0124] Figure 8 shows a typical ECG trace and the conventional labelling of the typical elements present in the ECG trace. Similar elements also occur in the MCG trace and the correspondence between the two has led to researchers using the same labelling convention.

[0125] As shown in Figure 8, the ECG trace comprises a repeating P-P interval comprising a so-called P-wave, followed by a P-R (or P-Q) segment (where the combination of the P-wave and the P-R (or P-Q) segment is referred to as the P-R (or P-Q) interval), followed by a QRS complex, followed by an S-T segment, followed by a T-wave (where the combination of the S-T segment and the T-wave is referred to as the S-T interval, and the combination of the QRS complex and the S-T interval is referred to as the Q-T interval), followed by a T-P segment.

[0126] On average, the QRS period usually lasts for approximately 80-150 ms. In this example, the magnetic field data is collected by the MCG data collection device at a rate of approximately 2100 Hz. Thus, the QRS period corresponds to approximately 160-320 samples, and therefore the parameter value may be derived by summing over across approximately 160-320 frames. Figures 9 and 10 are graphs illustrating behaviour of the dipole angle (y-axis) over time (along the x-axis), across the QRS period of the cardiac cycle. Figure 9 represents the behaviour of the dipole angle for a healthy patient, with low risk of future occurrence of an arrhythmia. As described above, the dipole angle is generally fairly constant, other than a fairly rapid shift in the middle of the QRS period. This therefore represents healthy behaviour for the dipole angle, and correspondingly the parameter value derived for this behaviour will be low because the angle change happens in the period of time when the weighting value is low, as seen in Figure 7.

[0127] It will be appreciated that were the same total angle change of 180° to occur, but much more slowly, across the whole QRS period, then the parameter value would be much higher since the angle changes would be occurring at points where the weighting value is much higher, and therefore such a behaviour may be identified as indicating an arrhythmia risk. For example, if the current diffuses very slowly after it has built up (e.g. due to defects in heart tissue) then the dipole angle would change more gradually, across a greater portion of the cardiac cycle, leading to an increased parameter value.

[0128] Figure 10 represents the behaviour of the dipole angle for a patient with a high risk of arrhythmia. It can be seen that the dipole angle is very unstable, and changes a lot across the cardiac cycle. This would therefore result in a much higher parameter value, once all these angle changes are summed across the cardiac cycle or a portion of the cardiac cycle. The consistent variation of the dipole angle throughout the QRS interval suggests that the electrical current is constantly changing direction as it passes around non-conducting regions of the myocardium.

[0129] Figures 11-16 show the results of scan data collected from 90 subjects each of whom were fitted with implantable cardioverter defibrillators (ICDs). The data is analysed to determine how effectively the parameter value (also referred to as the “rotation score”) and the secondary parameter value (also referred to as the “angular dynamics” score) correlate with the subject receiving therapy from the ICD (Figures 11-13) or receiving a shock from the ICD (Figures 14-16). An ICD can deliver two kinds of therapy, depending on the nature of the electrical signal that it applies. One of these is a series of low-voltage electrical pulses, known as pacing, which can sometimes be sufficient to return the heart to a normal rhythm. The ICD may deliver, for example, anti- tachycardia pacing (ATP), which is faster than the cycle length of the ventricular arrhythmia, or bradycardia pacing. The other type of therapy is an electric shock - either a low-level electric shock, which may be referred to as cardioversion, or a large electric shock known as defibrillation. Both ATP and shock are considered to be capable of terminating a life-threatening arrhythmia. The patients categorised as receiving “therapy” from their device have received any of these kinds of therapy, whereas the “shock” patients have received only the kind of therapy considered as a “shock”.

[0130] Figure 11 is a table showing the mean “rotation score” (parameter value) and “angular dynamics” score (secondary parameter value) for a total of 90 trial subjects (first column 800), for the 11 subjects out of that total for whom therapy was delivered by the ICD (second column, 802) and for the 79 patients to whom no therapy was delivered (third column, 804). The fourth column, 806, provides the p-value for each of the rotation score and the angular dynamics score, based on this data set. The p-value is a well understood statistical measure that indicates whether or not an effect is statistically significant.

[0131] Figure 12 is a receiver operating characteristic (ROC) curve for therapy vs the “rotation score”. An ROC curve will be understood as a graphical plot of sensitivity (i.e. true positive rate) as a function of specificity (i.e. false positive rate), or in this case 1 minus the specificity.

[0132] The area under the curve value (AUC) for such a curve gives an indication of the classifier’s performance - in the example of Figure 12, this is of the rotation score classifying for a particular subject receiving therapy from their ICD in future. The AUC score for the curve of Figure 12 is 0.727.

[0133] Figure 13 is similarly an ROC curve, in this case for the angular dynamics score. The AUC score for the curve of Figure 13 is 0.687.

[0134] Figure 14 shows a table similar to that of Figure 11, but in this case with the subjects classified based on whether or not they received a shock from their ICD. Figure 15 is an ROC curve for the rotation score classifying for a particular subject receiving a shock from their ICD in future. The AUC score for the curve of Figure 15 is 0.809. Figure 16 is similarly an ROC curve, in this case for the angular dynamics score. The AUC score for the curve of Figure 16 is 0.814.

[0135] It will be appreciated by those skilled in the art that the invention has been illustrated by describing one or more specific embodiments thereof, but is not limited to these embodiments; many variations and modifications are possible, within the scope of the accompanying claims.

Claims

Claims1. A data-processing apparatus for deriving a parameter value indicative of a subject’s risk of a future arrhythmia, the data-processing apparatus comprising a processing system, the processing system configured to: for a plurality of frames of a sequence of two-dimensional magnetic field data frames, each representing the rate of change of a magnetic field in the subject’s heart at a respective cardiac-cycle instant within a cardiac cycle, define a vector extending between a minimum-rate-of-change position in the two-dimensional frame, at which the rate of change of the magnetic field has a minimum value, and a maximum-rate-of- change position in the two-dimensional frame, at which the rate of change of the magnetic field has a maximum value; for a plurality of successive pairs of frames in the sequence, calculate a framepair angle change, by calculating the absolute value of the difference in angle of the vector defined for the earlier frame of the pair and the vector defined for the later frame of the pair; and calculate the parameter value by taking a combination of at least two of the frame-pair angle changes.

2. The data-processing apparatus of claim 1, wherein the combination is a linear combination.

3. The data-processing apparatus of claim 1 or 2, wherein the combination is a weighted sum, wherein the weighting of a particular frame-pair angle change of the combination is based on a relative magnitude value for at least one frame of the respective pair of frames from which the frame-pair angle change is calculated, wherein the relative magnitude value is indicative of the relative magnitude of the frame relative to other frames in the sequence.

4. The data-processing apparatus of claim 3, wherein the relative magnitude value for a particular frame is the magnitude difference divided by a scaling value, the magnitude difference being the difference between the value of the data frame at the maximum-rate-of-change position and the value of the data frame at the minimum- rate-of-change position, and the scaling value being the maximum magnitude difference of magnitude differences for all frames corresponding to the frame-pair angle changes used in calculating the parameter value.

5. The data-processing apparatus of claim 3 or 4, wherein the weighting for a particular pair of successive frames is the relative magnitude value of the later frame of the pair.

6. The data-processing apparatus of any preceding claim, wherein the two- dimensional frames of magnetic field data are averaged over several cardiac cycles.

7. The data-processing apparatus of any preceding claim, wherein the parameter value is derived using a sequence of two-dimensional magnetic field data frames wherein all of the frames in the sequence correspond to cardiac-cycle instants within the QRS period of the cardiac cycle.

8. The data-processing apparatus of claim 7, wherein the frames in the sequence used to derive the parameter value span the QRS period.

9. The data-processing apparatus of any preceding claim, wherein the processing system is further configured to receive heart rate data from the subject and to correlate the two-dimensional magnetic field data frames with the cardiac cycle using the heart rate data.

10. The data-processing apparatus of any preceding claim, wherein the processing system is further configured to derive a secondary parameter value, wherein the secondary parameter value is derived by dividing the parameter value by a time value, the time value indicative of the length of the QRS period of the cardiac cycle.

11. The data-processing apparatus of any preceding claim, further comprising an output module, wherein the processing system is arranged to output the parameter value and wherein the output module is arranged to display the value.

12. The data-processing apparatus of any preceding claim, wherein the sequence of two-dimensional magnetic field data frames is derived from data obtained using a sensor array and wherein the processing system is configured to interpolate data from at least two sensors in the array in order to derive each of the two-dimensional magnetic field data frames.

13. A system for deriving a parameter value indicative of a subject’s risk of a future arrhythmia, the system comprising: a magnetic field data collection module, arranged to obtain magnetic field data indicative of the rate of change of a magnetic field in the heart of a subject; and the data-processing apparatus of any preceding claim, wherein the two- dimensional frames of magnetic field data are derived from the magnetic field data obtained by the magnetic field data collection module.

14. The system of claim 13, wherein the magnetic field data collection device is a magnetocardiographic data collection device.

15. The system of claim 13 or 14, wherein the magnetic field data collection device comprises one or more inductive coils.

16. The system of any of claims 13 to 15, wherein the magnetic field data collection module comprises a sensor array17. The system of any of claims 13 to 16, wherein the magnetic field data collection device is arranged to collect magnetic field data at a sample rate of over 200 Hz.

18. The system of claim 17, wherein the magnetic field data collection device is arranged to collect magnetic field data at a sample rate of approximately 2100 Hz19. The system of any of claims 13 to 18, wherein the system further comprises a heart rate detection module, arranged to detect heart rate data of a subject to enable correlation of the magnetic field data with respect to the cardiac cycle.

20. The system of claim 19, wherein the heart rate detection module is an electrocardiogram device.

21. The system of any of claims 13 to 20, wherein the system further comprises an output module arranged to display the output parameter value22. A computer-implemented method for deriving a parameter value indicative of a subject’s risk of a future arrhythmia, the method comprising: for a plurality of frames of a sequence of two-dimensional magnetic field data frames, each representing the rate of change of a magnetic field in the subject’s heart at a respective cardiac-cycle instant within a cardiac cycle, defining a vector extending between a minimum-rate-of-change position in the two-dimensional frame, at which the rate of change of the magnetic field has a minimum value, and a maximum-rate-of- change position in the two-dimensional frame, at which the rate of change of the magnetic field has a maximum value; for a plurality of successive pair of frames in the sequence, calculating a framepair angle change, by calculating the absolute value of the difference in angle of the vector defined for the earlier frame of the pair and the vector defined for the later frame of the pair; and calculating the parameter value by taking a combination of at least two of the frame-pair angle changes.

23. The method of claim 22, further comprising averaging magnetic field data relating to at least two different time-instants, corresponding to respective cardiac cycles, to provide each two-dimensional magnetic field data frame of the sequence, wherein for each respective frame, each time-instant corresponds to the same cardiaccycle instant.

24. Software, or a signal or tangible medium bearing said software, comprising instructions which, when executed by a processing system, cause the processing system to: for a plurality of frames of a sequence of two-dimensional magnetic field data frames, each representing the rate of change of a magnetic field in the subject’s heart at a respective cardiac-cycle instant within a cardiac cycle, define a vector extending between a minimum-rate-of-change position in the two-dimensional frame, at which the rate of change of the magnetic field has a minimum value, and a maximum-rate-of- change position in the two-dimensional frame, at which the rate of change of the magnetic field has a maximum value; for a plurality of successive pair of frames in the sequence, calculate a framepair angle change, by calculating the absolute value of the difference in angle of thevector defined for the earlier frame of the pair and the vector defined for the later frame of the pair; and calculate the parameter value by taking a combination of at least two of the frame-pair angle changes.

Images