Electrode configurations for electrophysiological measurements

Abstract

A concept is proposed for generating and using a predictive model that allows prediction of the quality (e.g., accuracy and / or reliability) of an electrode configuration (i.e., arrangement) for electrophysiological measurements. It is proposed to use training data for determining parameter values ​​of the predictive model such that the predictive model is configured to accurately predict a quality value (e.g., a measure of accuracy) of an electrode arrangement or pattern for electrophysiological measurements of a subject. Thus, embodiments facilitate evaluation of electrode configurations and / or recommendation of changes to electrode configurations to allow for obtaining more reliable and / or accurate electrophysiological measurements.

Description

[Technical field]

[0001] The present invention relates to the field of electrophysiological measurements in a subject, and more particularly to electrode configurations for electrophysiological measurements. [Background technology]

[0002] Electrophysiological measurements are typically obtained via an array of electrodes placed on regions of the subject's (i.e., patient's) body. The signals acquired by such electrodes consist of a mixture of (unwanted) signals and noise sources that are superimposed on the electrophysiological signal of interest (e.g., electrocardiogram, ECG). To facilitate / assist clinical interpretation, for example, the signal of interest is extracted (i.e., separated from noise and unwanted signals) by dedicated signal processing methods.

[0003] Electrophysiological measurements have been demonstrated to be more accurate than standard diagnostics and facilitate additional benefits (e.g., allowing long-term remote monitoring). Summary of the Invention [Problem to be solved by the invention]

[0004] Furthermore, electrophysiological measurements are not very sensitive to accurate positioning of the electrodes for reliable measurements. However, for example, the ECG signal may be weak compared to the noise it is coupled with. This makes it difficult to extract the signal of interest. Therefore, optimal positioning of the electrodes may affect the accuracy and / or reliability of the electrophysiological measurements obtained. [Means for solving the problem]

[0005] The invention is defined by the claims.

[0006] According to one aspect, there is provided a method for generating a predictive model configured to predict a quality value of an electrode configuration (i.e., a positional arrangement or relative positioning of electrodes) for electrophysiological measurements, the method comprising: inputting a plurality of training electrode signals for a plurality of known electrode configurations and with a plurality of known quality values ​​into a predictive model; receiving a plurality of predicted quality values ​​from the predictive model; and determining parameter values ​​of the predictive model based on differences between the plurality of known quality values ​​and the plurality of predicted quality values. has.

[0007] By way of example, the proposed concept may provide a predictive model that allows prediction of the quality (e.g., accuracy and / or reliability) of an electrode configuration (i.e., arrangement) for electrophysiological measurements. That is, the embodiments propose to use training data to determine parameter values ​​of the predictive model, such that the predictive model is configured to accurately predict a quality value (e.g., a measure of accuracy) of an electrode arrangement or pattern for electrophysiological measurements of a subject. In this way, the embodiments may facilitate evaluation of and / or recommendation of changes to the electrode configuration to allow obtaining more reliable and / or more accurate electrophysiological measurements.

[0008] Reference to an electrode configuration is intended to refer to a positional arrangement, pattern, or relative placement of a plurality of electrodes. Thus, the electrodes of a first electrode configuration are positioned on the subject relative to one another (e.g., spaced apart from one another by the same or varying distances) to be in a first set of locations. The electrodes of a second, different electrode configuration are then positioned on the subject relative to one another to be in a second, different set of locations. In this manner, the first electrode configuration and the second, different electrode configuration have different positional arrangements of the electrodes (i.e., different relative placements or locations of the electrodes).

[0009] As an example, the proposed embodiments provide a model that aids in determining optimal electrode positions for extracting fetal heart rate (fHR) (particularly relevant during the antepartum period (i.e., before birth) when fetal position frequently changes). Although techniques are known for investigating optimal electrode positioning for monitoring fHR (e.g., using the results of ECG analysis), there is no consensus on a uniquely preferred configuration. However, the proposed embodiments can aid in determining optimal electrode configurations for acquiring and analyzing fetal ECG. Specifically, based on raw signals from multi-channel recordings, the quality of fetal ECG acquisition using a certain electrode configuration can be predicted before acquiring the signals. This allows optimal use of a small set of electrodes and automatic selection of an optimal electrode subset from a larger electrode grid. Specifically, some embodiments may be based on an understanding of blind source separation (BSS) techniques that are typically used for fetal ECG analysis.

[0010] Optimal electrode configuration (i.e., electrode positioning or placement) can minimize the number of components required to represent noise and interference sources, thus enabling high quality signal measurements with a reduced number of electrodes. Indeed, the use of fewer electrodes facilitates the implementation of continuous monitoring (e.g., by reducing power consumption, reducing the complexity of the electronic design, and / or reducing the required data volume and / or bandwidth).

[0011] The predictive models generated according to the proposed embodiments allow for an automatic assessment of the quality of electrode configurations for obtaining and / or monitoring electrophysiological measurements (e.g., fECG in pregnant women). In particular, the use of predictive models according to the embodiments can assist in (i) identifying an optimal electrode configuration among a set of available configurations, or (ii) determining whether a particular electrode configuration meets a required quality measure (e.g., whether the electrode configuration is sufficiently accurate for signal analysis and interpretation). Based on a multi-channel recording system, the results of such an automatic assessment are determined via analysis of the raw signal data, without requiring any pre-processing of the signals and / or without resource-intensive processing to extract and / or analyze the signals.

[0012] The electrodes used can be of any type: contact Ag-AgCl, capacitive or dry. Since the most advanced and accurate algorithms for analyzing fetal ECG are based on BSS techniques (e.g., principal component analysis, independent component analysis and singular value decomposition), embodiments support the assessment of the quality of the electrode configuration through the evaluation of multiple signal features reflecting linear and non-linear correlations between channels.

[0013] It is proposed to provide a predictive model that can predict the quality of an electrode configuration for electrophysiological measurements (e.g., electrocardiogram, electromyogram, electrogastrogram or electrohysterogram). Such a model can be used to determine whether an electrode configuration is good enough for accurate or reliable signal acquisition and analysis. Predictions from such a model are based on selected features extracted from the raw signals.

[0014] In some embodiments, the predictive model may use a classifier or regressor. For example, the predictive model may use a support vector machine (e.g., for classification) and / or a support vector regression (e.g., for regression), the latter providing a quality score that is again used for classification by threshold and decision making. Lasso or bridge regression, Gaussian mixture models or k-nearest neighbors may also be used, as well as many other neural network architectures. Such techniques allow for the determination of confidence levels to aid in decision making.

[0015] The plurality of training electrode signals can include training electrode signals having one or more signal variables. The predictive model is then configured to represent interactions between the one or more signal variables and a plurality of known quality values. By way of example, the one or more signal variables can include channel kurtosis, channel skewness, channel variance, inter-channel correlation, inter-channel mutual information, inter-channel spectral coherence, channel entropy, or inter-channel cross-entropy.

[0016] In some embodiments, the quality value may include at least one of an F-score, a root mean square error, an area under the ROC curve, accuracy, sensitivity, specificity, positive predictive value, and a correlation coefficient. Thus, various measures of quality may be used by embodiments, thus allowing for improved understanding of the performance and / or accuracy of an electrode configuration (e.g., through the use of multiple and / or different quality measures).

[0017] Determining parameter values ​​of the predictive model comprises adjusting parameter values ​​of the predictive model to reduce a difference between known quality values ​​associated with training electrode signals for known electrode configurations and predicted quality values ​​received from the predictive model for known electrode configurations. Thus, a relatively simple, iterative-based training / learning concept is used by the proposed embodiments, which in turn can reduce the cost and / or complexity of implementation.

[0018] According to another aspect of the invention, there is provided a method for processing electrode signals from an electrode configuration for electrophysiological measurements, said method comprising the steps of generating a predictive model according to a proposed embodiment, obtaining electrode signals from an electrode configuration for obtaining electrophysiological measurements of a subject, and obtaining a prediction of a quality value of an electrode configuration based on inputting the electrode signals into the generated predictive model.

[0019] Some embodiments further comprise identifying modifications to the electrode configuration based on the predicted quality value of the electrode configuration. Thus, embodiments may facilitate dynamic modification and / or optimization of the electrode configuration.

[0020] In some embodiments, the method further comprises classifying the electrode configuration into one of a plurality of categories based on the predicted quality value of the electrode configuration. For example, classifying the electrode configuration comprises comparing the predicted quality value of the electrode configuration with at least one threshold value and classifying the electrode configuration into one of a plurality of categories based on the result of the comparison. Thus, a relatively simple quality assessment concept is used by the proposed embodiments. This in turn can reduce the cost and / or complexity of implementation.

[0021] According to yet another aspect of the present invention, there is provided a method for processing electrode signals from an electrode configuration for electrophysiological measurements using a predictive model configured to predict a quality value of an electrode configuration, the method comprising the steps of obtaining electrode signals from an electrode configuration for obtaining electrophysiological measurements of a subject, and obtaining a prediction of a quality value of the electrode configuration based on inputting the electrode signals into the generated predictive model.

[0022] The quality value may comprise at least one of an F-score, a root mean square error, an area under a receiver operating characteristic (ROC) curve, accuracy, sensitivity, specificity, a positive predictive value, and a correlation coefficient.

[0023] Embodiments may further comprise identifying modifications to the electrode configuration based on the predicted quality value of the electrode configuration.

[0024] According to another aspect of the invention there is provided a computer program product comprising computer program code means which, when executed on a computing device having a processing system, causes said processing system to perform all of the steps of a method according to the proposed embodiments. Thus, according to an example embodiment according to one aspect of the invention there is provided a computer program comprising code means for implementing a method according to the proposed embodiments, when said computer program is executed on a processing system.

[0025] According to yet another aspect of the present invention, there is provided a system for processing electrode signals from an electrode configuration for electrophysiological measurements using a predictive model configured to predict a quality value of an electrode configuration for electrophysiological measurements, the system comprising a modeling unit configured to generate a predictive model configured to predict a quality value of an electrode configuration for electrophysiological measurements, an interface configured to obtain electrode signals from an electrode configuration, and a processing unit configured to obtain a prediction of a quality value of the electrode configuration based on inputting the electrode signals into the generated predictive model.

[0026] The proposed embodiments may provide and / or use a predictive model configured to predict a quality value of an electrode configuration for electrophysiological measurements. Thus, the embodiments are of particular use / benefit in combination with any medical device and / or procedure that deals with multi-channel surface biopotential measurements. By way of example only, the embodiments are particularly beneficial for fetal monitoring. However, the embodiments are applicable to other subject monitoring applications that rely on electrophysiological measurements.

[0027] Still further advantages of the present invention will be appreciated to those of ordinary skill in the art upon reading and understand the following detailed description of the invention.

[0028] These and other aspects of the invention will be apparent from and elucidated with reference to the embodiments described hereinafter. [Brief description of the drawings]

[0029] For a better understanding of the present invention, and to show more clearly how the same may be carried into effect, reference will now be made, by way of example only, to the accompanying drawings in which: [Figure 1] FIG. 1 is a simplified flow diagram of a method for generating a predictive model according to an embodiment. [Diagram 2] FIG. 2 is a simplified flow diagram of a method for processing electrode signals from an electrode configuration for electrophysiological measurements according to an embodiment. [Figure 3A] FIG. 3A is a graph showing the performance of SVM classification in terms of sensitivity, specificity and F-score of class (good or bad configuration) prediction for increasing numbers of features, using RMSE as a metric. [Figure 3B] FIG. 3B is a graph showing the performance of SVM classification in terms of sensitivity, specificity and F-score of class (good or bad configuration) prediction for increasing numbers of features, using the F-score as an index. [Figure 3C] FIG. 3C is a graph showing the performance of SVM classification in terms of sensitivity, specificity and F-score of class (good or bad configuration) prediction for increasing number of features, using a combination of F-score and RMSE as metrics. [Figure 4A] FIG. 4A is a graph showing the performance of the SVR predictions in terms of the obtained RMSE. [Figure 4B] FIG. 4B is a graph showing the performance of SVR prediction of the obtained F-scores. [Figure 4C] FIG. 4C is a graph showing the performance of SVR prediction for the combined F-score and RMSE obtained. [Diagram 5] FIG. 5 shows a preferred electrode configuration for fetal head-and-feet presentation using five electrodes (as predicted by a predictive model according to an embodiment). [Figure 6] FIG. 6 shows an exemplary electrode grid showing a set of 16 possible electrode locations. [Figure 7] FIG. 7 shows eight different five-electrode configurations of the electrode grid of FIG. [Figure 8] FIG. 8 is a simplified block diagram of a computer in which one or more portions of an embodiment may be used. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0030] The present invention will now be described with reference to the drawings.

[0031] It should be understood that while the detailed description and specific examples set forth exemplary embodiments of the devices, systems and methods, they 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 devices, systems and methods of the present invention will become better understood from the following description, appended claims and accompanying drawings. It should be understood that the drawings are merely schematic and are not drawn to scale. It should also be understood that the same reference numerals are used throughout the drawings to denote the same or similar parts.

[0032] The present invention provides a concept for modeling and / or predicting the quality (eg, accuracy and / or reliability) of electrode configurations (ie, arrangements) for electrophysiological measurements.

[0033] For example, embodiments may provide a predictive model configured to assess the quality of an electrode configuration. Based on such assessment, it is determined whether the electrode configuration meets one or more predefined requirements for reliable signal acquisition and / or analysis. The model may use one or more features extracted from the raw electrode signals.

[0034] The proposed predictive models can be implemented in several ways. For example, a support vector machine (SVM) is used for classification and a support vector regression (SVR) is used for regression. For example, such regression can determine a quality score for the electrode configurations, and such a quality score can be used for classification and / or decision making (e.g., by comparison with one or more thresholds).

[0035] The embodiments may assist or facilitate the determination of an optimal electrode configuration (i.e., spatial arrangement) for one or more specific requirements. For example, the embodiments may be used to identify an optimal electrode configuration for acquiring a fetal ECG. Using raw signals from a multi-channel recording, the quality of a planned / subsequent electrophysiological measurement may be predicted prior to acquisition. In particular, the embodiments allow the identification of a best / optimum electrode configuration from a set of available electrode configurations. Additionally or alternatively, the embodiments may determine whether an electrode configuration is good enough to acquire a signal (e.g., meets accuracy and / or reliability requirements).

[0036] A large number of feature sets have been tested and predictive models have been used for feature selection.

[0037] Thus, the proposed embodiments are used to provide electrode configuration recommendations that help improve the accuracy and / or reliability of electrophysiological measurements.

[0038] For example, embodiments may facilitate the generation of personalized / customized recommendations for electrode placement / arrangements tailored / compatible with specific situations and / or requirements, e.g., using predictive models according to proposed embodiments to provide sophisticated electrode configuration recommendations.

[0039] For example, the exemplary embodiments may be used in connection with healthcare-related services that involve the acquisition and / or analysis of electrophysiological measurements.

[0040] The embodiments according to the present disclosure relate to various systems, adaptations and / or methods for predicting or assessing the quality value of an electrode configuration for electrophysiological measurements. According to the proposed concept, several possible solutions may be implemented separately or together, i.e., these possible solutions are described individually below, but two or more of these possible solutions can be implemented in any combination.

[0041] By way of example, a first simplified embodiment will now be described with reference to FIG.

[0042] FIG. 1 is a simplified flow diagram of a method for generating a predictive model according to one embodiment. The predictive model is configured to predict a quality value of an electrode configuration (i.e., the positional arrangement or relative position of these electrodes) for electrophysiological measurements. Specifically, in this example, the predictive model uses a support vector machine (e.g., for classification) and support vector regression (e.g., for regression). Support vector regression provides a quality score that is again used for classification and decision making (e.g., via comparison of the quality score to a threshold).

[0043] The method begins at step 110 with inputting (to the predictive model) a number of training electrode signals for a number of known electrode configurations, the training electrodes being associated with a number of known quality values, i.e., training data having inputs with known associated outputs is provided to the predictive model.

[0044] As an example, the plurality of training electrode signals may comprise training electrode signals having one or more signal variables, such as channel kurtosis, channel skewness, channel variance, inter-channel correlation, inter-channel mutual information, inter-channel spectral coherence, channel entropy, or inter-channel cross-entropy, such that the prediction model is configured to represent an interaction between the one or more signal variables and the plurality of known quality values.

[0045] The quality value may also comprise at least one of the following: F-score, root mean square error, area under the ROC curve, accuracy, sensitivity, specificity, positive predictive value, and correlation coefficient, i.e., various measures of quality may be used.

[0046] In response to the input, a number of predicted quality values ​​are received from the prediction model at step 120. These predicted quality values ​​are generated by the prediction model based on the input training electrode signals, such that a difference between the output predicted quality value and the known quality value provides an indication of the accuracy of the prediction model and / or identifies how the prediction model needs to be modified to generate more accurate predictions.

[0047] Thus, in step 130, parameter values ​​of the predictive model are determined based on the difference between a plurality of known quality values ​​and a plurality of predicted quality values. Specifically, in this exemplary embodiment, determining parameter values ​​of the predictive model step 130 comprises adjusting parameter values ​​of the predictive model 132 to reduce the difference between known quality values ​​associated with training electrode signals for a known electrode configuration and predicted quality values ​​received from the predictive model for this known electrode configuration. In this manner, an iterative based training / learning algorithm can be used to determine optimal parameter values.

[0048] As a further example, an exemplary embodiment of a method for processing electrode signals from an electrode configuration for electrophysiological measurements is now described with reference to Fig. 2. In this example, a predictive model generated according to the proposed embodiment is utilized to obtain a prediction of a quality value of said electrode configuration, i.e., the proposed predictive model (such as the predictive model generated in the embodiment of Fig. 1) can be used to determine a quality value of the electrode configuration.

[0049] The method starts with step 210 of generating a predictive model according to the proposed embodiment, where, as shown in Fig. 2, step 210 comprises the method of Fig. 1. The result of step 210 is thus the provision of a predictive model for predicting a quality value of an electrode configuration for electrophysiological measurements.

[0050] Next, in step 220, electrode signals from the electrode configuration are acquired, and then in step 230, these acquired electrode signals are input into a predictive model.

[0051] A prediction of the quality value of the electrode configuration is obtained from the predictive model at step 240. The prediction of the quality value is generated by the predictive model based on the input electrode signals.

[0052] In the exemplary embodiment of FIG. 2, the predicted quality value is used in additional steps 250 and / or 260. Step 250 comprises identifying modifications to an electrode configuration based on the predicted quality value of the electrode configuration. Such modifications are communicated to an external system and / or a user as suggestions for improving the electrode configuration (e.g., for improving signal accuracy). Step 260 comprises classifying the electrode configuration into one of a plurality of classifications based on the predicted quality value of the electrode configuration. For example, step 260 of classifying an electrode configuration can comprise comparing the predicted quality value of an electrode configuration with at least one threshold value and classifying the electrode configuration into one of a plurality of classifications based on the result of the comparison. Even though classification of electrode configurations has been described in detail, alternative embodiments can use a regressor, in which electrode configurations are evaluated, for example, using a score.

[0053] As a further example and explanation, the application of the proposed concepts to predict the quality value of an electrode configuration for extracting fHR is now discussed.

[0054] The synthetic training data set was generated by the public FECGSYN simulator (currently detailed at the following internet address: http: / / fernandoandreotti.github.io / fecgsyn / ). This simulator makes it possible to mix sources related to fetal and maternal cardiac activity, together with muscle activity, electronic noise and volume conductors. The fetal position (e.g. head or feet) is also taken into account.

[0055] Simulations with 16 electrodes were performed to generate an adequate data set for training and testing the predictive model. In total, 200 subjects were simulated and 8 electrode configurations were evaluated for each of the subjects. The electrode configurations evaluated are shown below with reference to FIG. 7. With all signal components other than the fetal ECG signal considered as noise, the signal-to-noise ratio (SNR) quality value was less than −20 dB, which is comparable to the real acquisition.

[0056] Multiple feature sets are tested and predictive models are used for feature selection.For each electrode configuration, a known algorithm proposed by Varanini et al. ("A multi-step approach for non-invasive fetal ECG analysis", Computing in Cardiology 2013 (pp. 281-284)) was used to estimate the fetal R-peak.

[0057] From the resulting estimation, two metrics were calculated to evaluate the quality of the estimation: F-score and root mean square error (RMSE). These metrics evaluate the ability to detect the R-peak (F-score) as well as the timing error of the detected R-peak (RMSE) with respect to the ground truth. A good fetal ECG estimation shows a high F-score and a low RMSE. An SVR can be trained to predict these individual metrics or on their analytical combination, e.g. RMSE / F-score. An SVM can be trained to directly predict whether a configuration is good enough (binary classification), e.g. corresponding to “RMSE<50ms” and “F-score>90%”. Such a threshold, or an alternative threshold based on a combination of the two metrics, can also be applied to the prediction of the SVR to obtain a (binary) classification and distinguish between good and poor electrode configurations. A binary classification may be preferred to predict a good or bad outcome for a given electrode configuration.

[0058] For both SVM and SVR, forward 10-fold cross-validation is a valid option for feature selection.

[0059] 3A-3C show the performance of SVM classification in terms of sensitivity, specificity and F-score of class (good or bad configuration) prediction for increasing number of features, using either F-score or RMSE as the sole metric (FIGS. 3A and 3B) or a combination of them (FIG. 3C).

[0060] Specifically, Figure 3A is a graph showing performance on a test set versus the number of features selected, where the classifier is trained using the RMSE metric. Figure 3B is a graph showing performance on a test set versus the number of features selected, where the classifier is trained using the F-score metric. Figure 3C is a graph showing performance on a test set versus the number of features selected, where the classifier is trained using a hybrid (i.e., combination) of RMSE and F-score. In each graph of Figures 3A-3C, the classification sensitivity, specificity, and F-score are provided for each number of features.

[0061] 4A-4C show the predictive performance of SVR for the resulting F-score and RMSE separately (FIGS. 4A and 4B) or in combination (FIG. 4C). The performance of SVR is evaluated in terms of prediction error versus actual values.

[0062] Specifically, FIG. 4A is a graph showing prediction error of SVR in a test set versus number of features selected, where the regression model is trained using the RMSE metric. FIG. 4B is a graph showing prediction error of SVR in a test set versus number of features selected, where the regression model is trained using the F-score metric. FIG. 4C is a graph showing prediction error of SVR in a test set versus number of features selected, where the regression model is trained using a mixture (i.e., combination) of RMSE and F-score. In each graph of FIG. 4A-4C, prediction error is provided for each number of features.

[0063] Based on the results obtained, a good compromise for restricting the set of features and achieving good predictions is the following set of signal variables: Channel kurtosis; Channel distortion; Channel variance; correlation between channels; Mutual information between channels; and Spectral coherence between channels is given by:

[0064] Additional features related to channel entropy and cross-entropy between channels may be used, as well as other measures of connectivity.

[0065] Thus, by way of example only, embodiments may provide a method and / or system for determining an optimal electrode configuration in multi-channel electrophysiological monitoring. Such a method may involve determining the following signal variables from raw electrode signals (i.e., recorded signals having a mixture of desired signals (e.g., fetal ECG), unwanted signals (e.g., maternal ECG, muscle EMG, etc.), and noise) recorded by any set of channels of the electrode configuration: The kurtosis of the channel, Channel distortion, Channel diversification, Correlation between channels, Mutual information between channels, Spectral coherence between channels, Entropy and the Similar Correntropy The method includes deriving one or more of:

[0066] The derived variables are used as inputs to a proposed predictive model for predicting a quality value of an electrode configuration, which provides an indication of the quality (e.g., accuracy or reliability) of the electrode configuration prior to actually acquiring an electrophysiological measurement, e.g., a desired ECG signal.

[0067] Two exemplary applications of the proposed concepts are now described in the context of the analysis of fetal ECG.

[0068] Example 1 In this example, it is assumed that the number of electrodes is limited to the minimum number that still allows to perform a reliable fetal ECG analysis, where the proposed concepts are used to guide the operator or clinician to improved / correct positioning of the electrodes.

[0069] For example, a visible and / or audible output signal may be provided to indicate whether the electrode configuration meets accuracy / reliability requirements and / or to suggest one or more modifications to the electrode configuration, thus aiding in the acquisition of a proper fetal ECG. Based on such output signals, the operator may reposition the electrodes until positive feedback (e.g., a predefined output signal) is provided by the implemented predictive model (i.e., trained classifier). To assist the operator, guidance may be provided to initiate and / or assist the electrode positioning process.

[0070] Referring to Figure 5, a preferred electrode configuration for fetal head and feet position using five electrodes (as predicted by a predictive model according to one embodiment) is shown. The five electrodes are positioned on the mother's abdomen. The arrows indicate the four resulting differential channels.

[0071] As can be seen from the electrode configurations shown in FIG. 5, the simulations indicate that the preferred (eg, optimal) electrode configurations vary depending on, for example, head or feet position of the fetus.

[0072] Example 2 In this example, more electrodes are considered. Figure 6 shows an electrode grid showing a set of 16 possible electrode locations, where the spacing between adjacent electrode locations is 8 cm. This set of possible electrode locations therefore allows for multiple electrode configurations including fewer than 16 electrodes.

[0073] Given the various possible electrode positions shown in Figure 6, periodic testing of several different electrode configurations is performed, which yields the most reliable electrode configuration for analysis of the fetal ECG.

[0074] Figure 7 shows eight different five-electrode configurations for the electrode grid of Figure 6. With five electrodes in each electrode configuration, four channels are provided.

[0075] If the electronics for signal conditioning are limited to a small number of electrodes, a multiplexer can be used to switch between different electrode configurations to be tested. The configuration that gives the best prediction can then be selected to analyze the fetal ECG until the next test is performed. In this way, changes (e.g., degradation) in the fetal ECG analysis due to fetal movement or changing electrode impedance can be overcome (e.g., by dynamically identifying the optimal electrode configuration from the obtained predictions).

[0076] From the proposed concepts and the foregoing description of various exemplary embodiments, it will be appreciated that various benefits and / or advantages may be provided.

[0077] For example, optimal / preferred combinations of channels may be pre-selected and / or electrode repositioning may be guided prior to use (e.g., without injecting current, without the use of dedicated sensors, or without the need for extraction of vital signs).

[0078] Also, signal variables (e.g., kurtosis, skewness, variance, correlation, mutual information, spectral coherence, entropy and cross-entropy, and other measures of connectivity) are calculated as an initial step to detect / extract desired signal measurements (e.g., ECG or heart rate) from electrophysiological signals without the need for pre-processing (e.g., noise removal, crosstalk compensation, etc.).

[0079] It is therefore understood that the proposed concepts facilitate the evaluation and / or recommendation of electrode configurations. Thus, embodiments can provide personalized (i.e., user-specific) electrode configuration recommendations, where the recommendations are based on information from a predictive model.

[0080] The proposed methods and systems (including those of Figures 1-2) may be implemented in hardware or software, or a mixture of both (e.g., as firmware executing on a hardware device). To the extent that an embodiment is partially or wholly implemented in software, the functional steps illustrated in the process flow charts are performed by a suitably programmed physical computing device, such as one or more central processing units (CPUs) or graphic processing units (GPUs). Each process, and the steps of the individual components illustrated in the flow charts, may be performed by the same or different computing devices.

[0081] 8 illustrates an example of a computer 500 in which one or more components of the embodiments may be used. The various operations described above may utilize the functionality of the computer 500. For example, one or more components of a system for processing electrode signals from an electrode configuration for electrophysiological measurements using a predictive model may be incorporated into any of the elements, modules, applications and / or components described herein. In this regard, it should be understood that the functional blocks of the system may be executed on a single computer or may be distributed across several computers and locations (e.g., connected via the Internet).

[0082] The computer 500 includes, but is not limited to, PCs, workstations, laptops, PDAs, palm devices, servers, storages, etc. In general, with respect to hardware architecture, the computer 500 includes one or more processors 510, memory 520, and one or more I / O devices 570 communicatively coupled via a local interface (not shown). The local interface can be, but is not limited to, one or more buses, or other wired or wireless connections, as known in the art. The local interface can have additional elements to enable communication, such as controllers, buffers (caches), drivers, repeaters, and receivers. Additionally, the local interface can include address, control, and / or data connections to enable appropriate communication between the aforementioned components.

[0083] The processor 510 is a hardware device for executing software that may be stored in the memory 520. The processor 510 may be virtually any custom or commercially available processor, central processing unit (CPU), digital signal processor (DSP) or co-processor among several processors associated with the computer 500, and the processor 510 may be a semiconductor-based microprocessor (in the form of a microchip) or a microprocessor.

[0084] The memory 520 may include any one or combination of volatile memory elements (e.g., random access memory (RAM), such as dynamic random access memory (DRAM), static random access memory (SRAM)) and non-volatile memory elements (e.g., read only memory (ROM), erasable programmable read only memory (EPROM), electronically erasable programmable read only memory (EEPROM), programmable read only memory (PROM), tape, CD-ROM, disk, diskette, cartridge or cassette, etc.). Additionally, the memory 820 may incorporate electronic, magnetic, optical and / or other types of storage media. It should be noted that the memory 520 may have a distributed architecture, where various components are located remotely from each other, but they may be accessed by the processor 510.

[0085] The software in memory 520 includes one or more separate programs, each having an ordered list of executable instructions for implementing logical functions. The software in memory 520 includes a suitable operating system (O / S) 550, a compiler 540, source code 530, and one or more applications 560 according to an exemplary embodiment. As illustrated, application (App) 560 includes a number of functional components for implementing the features and operations of the exemplary embodiment. Application 560 of computer 500 may represent various applications, computing units, logic circuits, functional units, processes, operations, virtual entities, and / or modules according to an exemplary embodiment, although application 560 is not meant to be limiting.

[0086] Operating system 550 controls the execution of other computer programs and provides scheduling, input / output control, file and data management, memory management, and communication control and related services. It is contemplated by the inventors that application 560 for implementing the exemplary embodiments is applicable to all commercially available operating systems.

[0087] The application 560 may be a source program, an executable program (object code), a script, or any other entity having a set of instructions to be executed. In the case of a source program, the program is typically translated via a compiler (e.g., compiler 540), assembler, interpreter, or the like, which may or may not be included in memory 520, so as to operate properly in conjunction with the operating system 550. Furthermore, the application 560 may be written in an object-oriented programming language with classes of data and methods, or a procedural programming language with routines, subroutines, and / or functions, including, for example, but not limited to, C, C++, C#, Pascal, BASIC, API calls, HTML, XHTML, XML, ASP Script, JavaScript, FORTRAN, COBOL, Perl, Java, ADA, and NET, etc.

[0088] The I / O devices 570 may include input devices such as, but not limited to, a mouse, keyboard, scanner, microphone, camera, etc. Additionally, the I / O devices 570 may also include output devices such as, but not limited to, a printer, display, etc. Finally, the I / O devices 570 may further include devices that communicate both input and output, such as, but not limited to, a network interface card (NIC) or modulator / demodulator (for accessing remote devices, other files, devices, systems, or networks), radio frequency (RF) or other transceivers, telephone line interfaces, bridges, routers, etc. The I / O devices 570 may also include components for communicating over various networks, such as, for example, the Internet or an intranet.

[0089] If computer 500 is a PC, a workstation, an intelligent device, or the like, the software in memory 520 may further include a basic input output system (BIOS) (omitted for brevity). The BIOS is a set of essential software routines that initializes and checks the hardware at start-up, starts O / S 550, and assists in transferring data between hardware devices. The BIOS is stored in any kind of read-only memory, such as, for example, ROM, PROM, EPROM, or EEPROM, so that the BIOS can be executed when computer 500 starts up.

[0090] When computer 500 is operating, processor 510 is configured to execute software stored in memory 520, to communicate data to and from memory 520, and to generally control the operation of computer 500 in accordance with the software. Applications 560 and O / S 550 are read, in whole or in part, by processor 510, possibly buffered within processor 510, and then executed.

[0091] It should be noted that when application 560 is implemented in software, application 560 may be stored on virtually any computer-readable medium for use by or in connection with any computer-related system or method. In the context of this specification, a computer-readable medium may be an electronic, magnetic, optical or other physical device or means that can contain or store a computer program for use by or in connection with a computer-related system or method.

[0092] The application 560 may be embodied in any computer readable medium for use by or in connection with an instruction execution system, apparatus, or device, such as a computer-based system, a processor-containing system, or other system that can fetch instructions from such an instruction execution system, apparatus, or device and execute the instructions. In the context of this specification, a "computer readable medium" may be any means that can store, communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device. The computer readable medium may be, for example, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, device, or propagation medium.

[0093] The present invention may be a system, a method and / or a computer program product. The computer program product may include a computer readable storage medium having computer readable program instructions for causing a processor to perform aspects of the present invention.

[0094] According to an embodiment, a computer readable storage medium stores a computer program having computer program code configured to cause one or more physical computer devices to perform the above-mentioned methods when the program is executed on one or more physical computer devices.

[0095] The storage medium may include, for example, volatile and non-volatile computer memory such as RAM, PROM, EPROM, and EEPROM. Various storage media may be installed in a computing device or may be transportable such that one or more programs stored thereon may be read by a processor.

[0096] All or part of the schematic model according to the embodiment may be stored on a storage medium. Data according to the embodiment may be stored on the same storage medium or on a different storage medium. The schematic model and / or the data may be transmitted as a signal modulated onto an electromagnetic carrier wave. The signal may be defined according to a standard for digital communications. The carrier wave may be an optical carrier wave, a radio frequency wave, a millimeter wave, a short-range communication wave. It may be wired or wireless.

[0097] To the extent that the embodiments are implemented partially or entirely in hardware, the blocks shown in the figures may be separate physical components, or logical subdivisions of a single physical component, or may all be implemented in an integrated manner in one physical component. The functionality of a block shown in the figures may be divided among multiple components in implementation, or the functionality of multiple blocks shown in the figures may be combined into a single component in implementation. Hardware components suitable for use in embodiments of the present invention include, but are not limited to, conventional microprocessors, application specific integrated circuits (ASICs), and field-programmable gate arrays (FPGAs). One or more of the blocks may be implemented as a combination of dedicated hardware to perform some functions and one or more programmed microprocessors and associated circuitry to perform other functions.

[0098] As used herein, memory includes one or more of a non-transitory computer readable medium, a magnetic disk or other magnetic storage medium, an optical disk or other optical storage medium, a RAM, a ROM, or other electronic memory device or a chip or set of operatively interconnected chips, or an Internet / Intranet server where stored instructions are retrieved over the Internet / Intranet or a local area network, etc. Additionally, as used herein, a processor includes one or more of a microprocessor, a microcontroller, a graphics processing unit (GPU), an ASIC, and an FPGA, etc.; a user input device includes one or more of a mouse, a keyboard, a touch screen display, one or more buttons, one or more switches, and one or more toggles, etc.; a display device includes one or more of a liquid crystal display (LCD), a light emitting diode (LED) display, a plasma display, a projection display, and a touch screen display, etc.; and a database includes one or more memories.

[0099] The present invention has been described with reference to the preferred embodiment. Modifications and alterations may occur to others upon reading and understanding the above detailed description. It is intended that the present invention be construed as including all such modifications and alterations insofar as they come within the scope of the appended claims and their equivalents.

[0100] It is also understood that the disclosed methods may be computer-implemented methods. As such, the concept of a computer program product having code means for performing any of the described methods when said program is executed on a processing system is also proposed.

[0101] Those skilled in the art can readily develop a processor to perform any of the methods described herein, and each step of the flowchart may therefore represent a different action performed by a processor, or may be performed by a respective module of the processing processor.

[0102] The proposed system may utilize a processor to process data. The processor may be implemented in various ways using software and / or hardware to perform the various functions required. The processor typically uses one or more microprocessors that are programmed using software (e.g., microcode) to perform the required functions. The processor may be implemented as a combination of dedicated hardware to perform some functions and one or more programmed microprocessors and associated circuitry to perform other functions.

[0103] Examples of circuitry that may be used in various embodiments of the present disclosure include, but are not limited to, conventional microprocessors, ASICs, and FPGAs.

[0104] In various implementations, the processor may be associated with one or more storage media, e.g., volatile and non-volatile computer memories such as RAM, PROM, EPROM, and EEPROM. The storage media may be encoded with one or more programs that, when executed on one or more processors and / or controllers, perform the required functions. The various storage media may be attached within the processor or controller, or may be transportable such that one or more programs stored on the storage media are read into the processor.

[0105] Variations to the disclosed embodiments can be understood and implemented by those skilled in the art in practicing the claimed invention, from a study of the drawings, the disclosure and the appended claims. In the claims, the word "comprising" does not exclude other elements or steps, nor does it exclude a plurality of them, even if a plurality is not stated. A single processor or other unit may fulfill the functions of several items recited in the claims. The mere fact that certain means are recited in mutually different dependent claims does not indicate that a combination of these means cannot be used to advantage. The computer program may be stored / distributed on a suitable medium, for example an optical storage medium or a solid-state medium provided together with or as part of other hardware, but may also be distributed in other forms, for example via the Internet or other wired or wireless telecommunication systems. When the word "compatible with" is used in the claims or the description, this word "compatible with" is intended to be equivalent to the word "comprised with". Any reference signs in the claims should not be interpreted as limiting the scope.

Claims

1. 1. A method for generating a predictive model configured to predict a quality value of an electrode configuration for electrophysiological measurements, the electrode configuration being a positional arrangement of a plurality of electrodes, the method comprising: inputting a plurality of training electrode signals for a plurality of known electrode configurations and with a plurality of known quality values ​​into the predictive model; receiving a plurality of predicted quality values ​​from the predictive model; and determining parameter values ​​of the predictive model based on differences between the plurality of known quality values ​​and the plurality of predicted quality values. The method according to claim 1,

2. The method of claim 1 , wherein the predictive model uses a classifier or a regression.

3. 3. The method of claim 1 or 2, wherein the plurality of training electrode signals comprises training electrode signals comprising the one or more signal variables, and the predictive model is configured to represent an interaction between the one or more signal variables and the plurality of known quality values.

4. 4. The method of claim 3, wherein the one or more signal variables comprise channel kurtosis, channel skewness, channel variance, inter-channel correlation, inter-channel mutual information, inter-channel spectral coherence, channel entropy, or inter-channel cross-entropy.

5. 2. The method of claim 1, wherein the quality value comprises at least one of an F-score, a root mean square error, an area under a ROC curve, an accuracy, a sensitivity, a specificity, a positive predictive value, and a correlation coefficient.

6. The step of determining parameter values ​​of the predictive model comprises:

2. The method of claim 1, comprising adjusting parameter values ​​of the predictive model to reduce a difference between known quality values ​​associated with training electrode signals for known electrode configurations and predicted quality values ​​received from the predictive model for the known electrode configurations.

7. 1. A method for processing electrode signals from an electrode configuration for electrophysiological measurements, the method comprising: Generating a predictive model according to claim 1; acquiring electrode signals from an electrode configuration for obtaining electrophysiological measurements of a subject; and obtaining a prediction of a quality value of the electrode configuration based on inputting the electrode signals into the generated prediction model. The method comprising:

8. The method of claim 7 , further comprising identifying modifications to the electrode configuration based on the predicted quality value of the electrode configuration.

9. 9. The method of claim 7 or 8, further comprising classifying the electrode configuration into one of a plurality of classifications based on the predicted quality value of the electrode configuration.

10. The step of classifying the electrode configurations comprises: comparing the predicted quality value of the electrode configuration to at least one threshold; and classifying the electrode configuration into one of a plurality of classifications based on a result of the comparison.

10. The method of claim 9, having the following structure:

11. 1. A method for processing electrode signals from an electrode configuration for electrophysiological measurements using a predictive model configured to predict a quality value of an electrode configuration, the electrode configuration being a positional arrangement of a plurality of electrodes, the method comprising: acquiring electrode signals from an electrode configuration for obtaining electrophysiological measurements of a subject; and obtaining a prediction of a quality value of the electrode configuration based on inputting the electrode signals into the prediction model. The method comprising:

12. 12. The method of claim 11, wherein the quality value comprises at least one of an F-score, a root mean square error, an area under the ROC curve, an accuracy, a sensitivity, a specificity, a positive predictive value, and a correlation coefficient.

13. The method of claim 11 or 12, further comprising identifying modifications to the electrode configuration based on the predicted quality value of the electrode configuration.

14. The method of claim 11 , further comprising classifying the electrode configuration into one of a plurality of classifications based on the predicted quality value of the electrode configuration.

15. 1. A system for processing electrode signals from an electrode configuration for electrophysiological measurements using a predictive model configured to predict a quality value of an electrode configuration, the electrode configuration being a positional arrangement of a plurality of electrodes, the system comprising: a modeling unit configured to generate a predictive model configured to predict a quality value of an electrode configuration for an electrophysiological measurement; an interface configured to acquire an electrode signal from the electrode configuration; and a processing unit configured to obtain a prediction of a quality value of the electrode configuration based on inputting the electrode signals into the generated prediction model. The system has:

Images