Medical equipment positioning method based on magnetic and impedance sensors

Abstract

Provided herein are systems and methods for identifying the positioning of electrodes of a catheter in three-dimensional space. The system and method initially predict the positioning of physical electrodes and / or physical magnetic sensors of a catheter in three-dimensional space. Impedance and / or magnetic response are predicted for the predicted positioning. Actual measurements / responses (e.g., measured responses) are then obtained for the physical electrodes and / or physical sensors. Based on the predicted responses and measured responses, the system and method generate the calculated positioning of electrodes and / or sensors in three-dimensional space. The system and method utilize information from both the predicted responses and the measured responses to generate the calculated positioning, which can have greater accuracy than the positioning generated by the predicted responses or measured responses.

Description

[0001] Cross-references

[0002] This application claims the benefit of the following application dates: U.S. Provisional Application No. 62 / 756,941, filed on November 7, 2018; U.S. Provisional Application No. 62 / 756,915, filed on November 7, 2018; U.S. Provisional Application No. 62 / 756,926, filed on November 7, 2018; U.S. Provisional Application No. 62 / 756,931, filed on November 7, 2018; and U.S. Provisional Application No. 62 / 756,936, filed on November 7, 2018, the entire contents of each of the above applications are incorporated herein by reference. Technical Field

[0003] The present disclosure generally relates to positioning a medical device in a patient reference frame using a medical device model that estimates the shape of the medical device in the patient reference frame in conjunction with measurements from impedance electrodes and magnetic sensors of the medical device. Background Art

[0004] A variety of systems are known for determining the position and orientation (P&O) of a medical device in the human body, for example, for visualization and navigation purposes. One such system is referred to as an electrical impedance-based positioning system. An electrical impedance-based system typically includes one or more pairs of body surface electrodes (e.g., patches) outside the patient's body, a reference sensor (e.g., another patch) attached to the patient's body, and one or more sensors (e.g., electrodes) attached to the medical device. These pairs can be adjacent, arranged linearly, or associated with corresponding axes of a coordinate system for such positioning systems. The system can determine P&O by applying current to the electrode pairs, measuring the corresponding voltages induced at the device electrodes (i.e., relative to the reference sensors), and then processing the measured voltages.

[0005] Another system is called a magnetic field-based positioning system. This type of system typically includes one or more magnetic field generators that are attached to or placed near a patient bed or other components of the operating environment; and one or more magnetic field detection coils coupled to a medical device. Alternatively, the field generator can be coupled to the medical device, and the detection coil can be attached to or placed near a component of the operating environment. The generator provides a controlled low-intensity AC magnetic field in the area of ​​interest (i.e., the anatomical region). The detection coil generates a corresponding signal indicating one or more characteristics of the sensing field. The system then processes these signals to generate one or more P&O readings associated with the coil (and thus the medical device). The P&O readings are typically about the field generator, and thus the field generator acts as the actual "origin" of the coordinate system of the magnetic field-based positioning system. Unlike electrical impedance-based systems, magnetic field-based systems have a patient-independent coordinate system in the case where the coordinate system is relative to the patient to whom the body surface electrodes are applied.

[0006] Both electrical impedance-based and magnetic field-based positioning systems offer advantages. For example, electrical impedance-based systems provide the ability to simultaneously position (i.e., provide P&O readings) a relatively large number of sensors on multiple medical devices. However, because electrical impedance-based systems utilize the flow of electrical current in the human body, such systems may be subject to electrical interference. As a result, the geometry and representation presented based on the position measurement may appear distorted relative to the actual image of the examined area of ​​interest. On the other hand, magnetic field-based coordinate systems do not depend on the characteristics of the patient's anatomy and generally provide improved accuracy. However, magnetic field-based positioning systems are generally limited to tracking relatively few sensors.

[0007] Efforts have been made to provide a system that combines the advantages of an electrical impedance-based positioning system (e.g., positioning of multiple electrodes) with the advantages of a magnetic field-based coordinate system (e.g., independence from the patient's anatomical structure, higher accuracy). In an embodiment, such a system can be provided by aligning the coordinate system of an electrical impedance-based positioning system with the coordinate system of a magnetic field-based positioning system. In such an arrangement, the position of an electrode in an impedance-based coordinate system can be identified in conjunction with identifying the position of one or more magnetic sensors in a magnetic-based coordinate system. In an embodiment, at least a portion of the electrode and the magnetic sensor can be co-located to define a reference pair. This co-location allows the determination of a transformation (e.g., a transformation matrix) between coordinate systems. Once the transformation is determined, the transformation can be applied to the positioning of any electrode to align these positionings in a magnetic-based coordinate system. Therefore, an electrical impedance-based electrode can be identified in the coordinate system of a magnetic field-based positioning system, thereby improving the positioning accuracy of the electrode. While providing improved electrode positioning, the determination of the transformation between an impedance-based coordinate system and a magnetic-based impedance system and the subsequent alignment of the electrode positioning to a magnetic coordinate system may not be able to resolve a variety of impedance offsets and / or drifts associated with the electrode.

[0008] Previous systems that utilize electrode information (e.g., impedance measurements) and magnetic sensor information to provide improved electrode positioning in three-dimensional space (e.g., in a patient's body) rely primarily on impedance-based measurements. That is, magnetic sensor information (e.g., magnetic sensor measurements) provides additional accuracy. This can be described as an impedance-based positioning arrangement. Due to the distortion and temporal instability of impedance measurements, such arrangements may suffer from instability. In addition, in some cases, previous impedance-based positioning arrangements cannot resolve multiple errors within the system. In addition, such systems may not be able to consider other system inputs (e.g., patient movement, shape of medical equipment, etc.) that may affect the calculated positioning or position of the electrode. In short, the registration of an impedance-based system to a magnetic-based system may not include additional information that can be observed and / or inferred and can improve the overall identification of the catheter and / or electrode position in three-dimensional space. Summary of the invention

[0009] Various embodiments herein provide systems, methods and / or non-transitory computer-readable media storing instructions (i.e., utilities) for identifying the positioning of electrodes of catheters in three-dimensional space (e.g., a patient's body or a patient's reference system). Initially, the utility involves predicting the positioning of physical electrodes and / or physical magnetic sensors of a physical medical device disposed in three-dimensional space. The utility predicts the response or measurement (hereinafter referred to as "response") of the electrodes and / or sensors based on the predicted positioning of the electrodes and / or sensors. In addition, the utility obtains actual measurements / responses from the electrodes and / or sensors of the physical catheter. For example, when an applied electric potential field is applied to the three-dimensional space, the utility can obtain or measure an impedance response from the physical electrodes. Similarly, when a magnetic field is applied to the three-dimensional space, the utility can obtain or measure a magnetic response. Based on the predicted response and the measured response, the utility can update the positioning of the electrodes or sensors in the three-dimensional space. Such updated positioning can utilize information from both the predicted response and the measured response to generate the positioning of the electrodes and / or sensors (e.g., calculated positioning), wherein the calculated positioning has greater accuracy than the positioning generated by the predicted response or the measured response.

[0010] In an embodiment, the utility relates to a positioning arrangement that integrates predicted and measured impedance responses of electrodes from a medical device with other observed parameters (such as predicted and measured position and orientation responses of a magnetic sensor) to estimate the position (e.g., latent state) of a physical medical device (e.g., a physical catheter) disposed within a patient reference system. In an embodiment, the utility utilizes a catheter model of a physical catheter to predict the positioning of physical electrodes and / or magnetic sensors in three-dimensional space. The catheter model models the physical catheter, wherein the spacing of model electrodes and / or model sensors of the catheter model corresponds to the spacing of electrodes and / or sensors of the physical catheter. In such embodiments, the catheter model defines the model electrodes and / or model sensors in the catheter reference system. In an embodiment, a catheter transformation transforms the positioning of the model electrodes and / or sensors from the catheter reference system to three-dimensional space (e.g., a patient reference system) to predict the positioning of the model electrodes and / or sensors in three-dimensional space. In an embodiment, the transformation is a rigid body transformation.

[0011] In an embodiment, the utility applies the impedance model to the model electrode location in three-dimensional space to predict the impedance response of the model electrode location. In an embodiment, the impedance model models the potential field applied to the three-dimensional space via the physical surface patch electrodes. In such embodiments, independent impedance fields can be mapped to the driven patch pairs to estimate the impedance response or measurement of any location within the potential field.

[0012] In an embodiment, the utility applies the magnetic model to the model sensor location in three-dimensional space to predict the magnetic response of the model sensor location. In an embodiment, the magnetic model includes a magnetic patient reference sensor disposed in three-dimensional space. In such embodiments, the origin defined by the patient reference sensor can be related to the origin of the applied magnetic field. The measured responses of the physical sensors and the model sensors can be used together with the magnetic model to locate and orient the electrodes and sensors of the catheter model in three-dimensional space.

[0013] In an embodiment, the utility integrates (e.g., fuses) the predicted and measured impedance responses from electrodes and / or external patches with additional observation parameters (including, for example, predicted and measured position and orientation responses from magnetic sensors) to estimate the potential state (e.g., position) of a medical device set in three-dimensional space. In an embodiment, a model (e.g., a catheter model, a catheter transformation, an impedance model, and / or a magnetic model) is a variable model in which the variables of the model represent state variables of a state-space system. Such a state-space system allows multiple models to be updated based in part on the measured response of a physical system. In an embodiment, one or more models are used in combination to define a composite model of a medical device in three-dimensional space. In such an embodiment, an estimator system can estimate the potential (e.g., hidden) variables of each model to iteratively improve the correspondence of the model with the physical system they represent. In an embodiment, the estimator is an extended Kalman filter. In any embodiment utilizing an estimator, a state space estimate of possible states can be generated from a composite model. A variety of constraints can be applied to the state space estimate to penalize unlikely states. The most likely state (e.g., mean and covariance) of the state space estimate can be mapped to the measured response to produce a corrected state space estimate. Updated positioning of electrodes and / or sensors may be generated from the corrected state-space estimates.

[0014] Various embodiments described herein provide systems, methods and / or non-transitory computer readable media storing instructions (i.e., utilities) for estimating the shape of a deformable catheter in a three-dimensional space (e.g., a patient reference space). A catheter model is used to estimate the shape of a deformable catheter. The catheter model includes definitions of two or more model segments corresponding to two or more segments of the deformable catheter. Typically, the length of each model segment is defined as the positioning of electrodes and / or magnetic sensors along the length. The spacing of the electrodes and / or magnetic sensors in the definition corresponds to the spacing of the physical electrodes and / or magnetic sensors of the corresponding physical catheter (i.e., the deformable catheter). Each model segment may have one or more variable shape parameters defining the curvature of the segment. That is, the model segment may define a first variable shape parameter for a first segment and a second variable shape parameter for a second segment, wherein the variable shape parameter describes the curvature of the model segment. In one arrangement, each of the model segments includes a variable curvature parameter and a torsion parameter. These parameters may vary within a predetermined range (e.g., in a computer model), which may be predetermined and / or dependent on the physical properties of the modeled catheter. In addition, the parameters of each model segment may be different. The shape parameters are changed by the computer to generate multiple potential catheter shapes. Each potential shape can include model electrode positioning and / or model magnetic sensor positioning. That is, each model segment can define the positioning of one or more electrodes and / or magnetic sensors along the length of the model segment. In one arrangement, the potential catheter shape defines the state distribution of the potential shape. In conjunction with generating potential catheter shapes, the impedance and / or magnetic response (e.g., measured response) of the electrodes and / or magnetic sensors for the deformable catheter arranged in three-dimensional space can be obtained. For example, a medical positioning system can measure these responses. Using one of the selected catheter shapes and measured responses, the utility is operable to update the variable shape parameters so that the catheter model is more closely adapted to the shape of the deformable catheter. The updated shape parameters can be used to generate a catheter shape that can be output to a display. Such updates can be substantially continuous. For example, the shape parameters and / or generated catheter shapes can be updated 30, 50 or even 100 times per second.

[0015] In one arrangement, the selected catheter shape model is used to predict the positioning of electrodes and / or magnetic sensors in three-dimensional space. In such an arrangement, the catheter model can be transformed from the catheter reference system to the three-dimensional space to predict the positioning of the model electrodes and / or model sensors in the three-dimensional space. A predicted response is generated for the predicted positioning of the model electrodes and / or model sensors. Such predicted responses can be generated by an impedance model that models the impedance field of the three-dimensional space and / or a magnetic model that models the magnetic field of the three-dimensional space. Based on the predicted response and the measured response, the utility can update the positioning of the electrodes or sensors in the three-dimensional space (for example, generate a calculated positioning). The utility can utilize information from both the predicted response and the measured response to generate the calculated positioning of the electrodes and / or sensors of the catheter. The calculated positioning generally has a greater accuracy than the positioning generated by the predicted response or the measured response. In addition, the predicted response and the measured response can be utilized to update the variable shape parameters.

[0016] In an embodiment, the utility integrates (e.g., fuses) predicted and measured responses to estimate hidden variables of the system. Such hidden variables may include the position of the catheter in the three-dimensional space of the catheter and the variable parameters of the catheter model. In one arrangement, the variable parameters of the model segment of the catheter model represent the state variables of the state vector. Such an arrangement allows multiple parameters to be updated based in part on the measured response of the physical system. In such an arrangement, the estimator system can estimate potential (e.g., hidden) variables to iteratively improve the correspondence of the catheter model with the physical catheter they represent. In an embodiment, the estimator is an extended Kalman filter. In any embodiment utilizing the estimator, a state space estimate (e.g., catheter shape) of possible states can be generated. The most likely shape can be represented by an average state distribution. The mean of the state space estimate can be mapped to the measured response to generate a corrected state space estimate. The calculated positioning of the electrode and / or sensor can be generated from the corrected state space estimate. Similarly, updated shape parameters can be generated from the corrected state space estimate.

[0017] In one arrangement, each model segment of the catheter model includes at least one electrode and / or at least one sensor. Such an arrangement ensures that the measured response from the corresponding segment of the physical catheter can be used to adjust the variable parameters of each model segment. In one arrangement, the model segments are continuous. The continuous model segments can define the entire deformable portion of the catheter. In one arrangement, each model segment is defined as a moving coordinate system. In a specific arrangement, the moving coordinate system is a Frenet coordinate system.

[0018] Various embodiments described herein provide systems and methods for determining shape parameters of a deformable catheter. The system and method apply known forces and orientations to the catheter. Such systems and methods can be implemented in bench tests. In one arrangement, the deformable catheter is maintained at a known roll angle relative to the central axis of the catheter (e.g., the catheter shaft). The movable slide contacts the distal end of the deformable catheter at a known contact angle. The movable slide is advanced a predetermined distance and / or until a predetermined force setting value is reached. At this point, the three-dimensional positioning of the electrode and / or magnetic sensor can be obtained (e.g., using three-dimensional imaging). The three-dimensional positioning of the electrode and / or sensor can be associated with known forces, roll angles, and contact angles to determine the shape parameters of one or more segments of the catheter for a known displacement. The process can be repeated for multiple arrangements of roll angles, contact angles, displacements, and / or forces to determine the profile of the shape parameters.

[0019] Various embodiments herein provide systems, methods and / or non-transitory computer-readable media storing instructions (i.e., utilities) for identifying the positioning of electrodes of a catheter within a three-dimensional space (e.g., a patient's body or a patient reference frame) while taking into account respiratory artifacts. That is, the inventors have recognized that during a medical procedure (e.g., a cardiac medical procedure), in vivo impedance measurement errors covariate significantly due to breathing. That is, breathing causes time-varying artifacts relative to spatially varying impedance measurements within a patient reference frame (e.g., on or within a patient's chest). Time-varying artifacts occur during each respiratory cycle due to increases and decreases in the volume changes of the patient's chest. More specifically, changes in volume change the patient's physiological state and thereby change the impedance measurement of the impedance potential field within the patient's reference frame. Therefore, taking into account respiratory artifacts allows for improved accuracy of electrode positioning in the patient's reference frame (e.g., determined by impedance measurements).

[0020] In one arrangement, the utility predicts a respiratory artifact of an impedance field (e.g., covering all or a portion of a patient reference frame) based on the phase angle and amplitude of the patient's current respiratory cycle. In addition, the utility predicts one or more spatially related impedance values ​​for a predicted location of one or more physical electrodes (e.g., catheter electrodes) of a physical medical device (e.g., a physical catheter) disposed within the patient reference frame. The respiratory artifact and the predicted spatially related impedance value together define a predicted impedance value for the predicted location of the catheter electrode. The utility then obtains an observed or measured impedance value of the catheter electrode. For example, such measured impedance values ​​can be obtained from an impedance-based medical positioning device. Based on the predicted impedance value and the measured impedance value, the utility can calculate the location of the electrode in the patient reference frame. Such calculated location can utilize information from both the predicted impedance value and the measured impedance value to produce a more accurate location of the electrode. Such calculated location can have greater accuracy than the location generated by the predicted impedance value or the measured impedance value. In addition, the predicted impedance value and the measured impedance value can be utilized to update the phase and / or amplitude used to predict subsequent respiratory artifacts. Additionally, the utility may output the calculated position to a display, such as in or on a rendering of a catheter disposed within a patient.

[0021] In one arrangement, the breathing model predicts breathing artifacts. In such an arrangement, the breathing model is defined as a quasi-periodic function where phase and amplitude are variables of the model. In one arrangement, the quasi-periodic function is equal to zero when the phase angle is zero. In another arrangement, the phase and amplitude are hidden variables of the breathing model. In such an arrangement, the phase and amplitude are state variables that can be estimated in the estimation system even though these variables are never directly observed. In one implementation, a Kalman filter is used to estimate the state variables.

[0022] In one arrangement, the utility uses a catheter model of a physical catheter to predict the positioning of electrodes in a patient reference frame. The catheter model models the physical catheter, wherein the spacing of the model electrodes and / or sensors of the catheter model corresponds to the spacing of the electrodes and / or sensors of the physical catheter. In such embodiments, the catheter model defines the model electrodes in the catheter reference frame. In an embodiment, the catheter transformation transforms the positioning of the model electrodes and / or sensors from the catheter reference frame to the patient reference frame to predict the positioning of the model electrodes (e.g., the model positioning). The utility applies an impedance model to the model electrode positioning to predict the spatially related impedance values ​​of the model electrode positioning. In an embodiment, the impedance model models the potential field applied to the patient reference frame by the physical surface patch electrodes. In such embodiments, independent impedance fields can be mapped to the driven patch pairs to estimate the impedance response or measurement at any position within the potential field.

[0023] In one arrangement, the utility integrates (e.g., fuses) a predicted impedance response including a respiratory artifact and a measured impedance response from an electrode to estimate a potential state (e.g., position) of a medical device disposed within a patient reference space. In an embodiment, the model (e.g., a catheter model, an impedance model, and / or respiration) is a variable model in which the variables of the model represent state variables of a state-space system. Such a state-space system allows multiple models to be updated based in part on the measured response of a physical system. In an embodiment, one or more models are used in combination to define a composite model of a medical device in three-dimensional space. In such an embodiment, an estimator system can estimate potential (e.g., hidden) variables of each model to iteratively improve the correspondence of the model to the physical system they represent. In an embodiment, the estimator is an extended Kalman filter.

[0024] Various embodiments described herein provide systems, methods and / or non-transitory computer-readable media storing instructions (i.e., utilities) for predicting impedance values ​​or measurements in three-dimensional space. In a broad sense, the utility defines an impedance potential field and its measurement characteristics so that impedance measurements can be estimated for any location within the potential field. The utility defines a transformation or impedance model that estimates electrode impedance measurements (e.g., location-impedance values) in three-dimensional space. The model can develop over time based on actual impedance measurements of electrodes located in three-dimensional space. The utility drives multiple patch electrodes to generate an impedance field to three-dimensional space (e.g., a patient reference frame). For example, such patch electrodes can be applied to the patient's body externally (e.g., surface patch electrodes). Each pair of surface patch electrodes (e.g., source-sink) can be driven to generate an impedance field in three-dimensional space. For example, in a six-patch electrode system, six separate combinations of paired patch electrodes can be driven for each impedance measurement. When each pair of patches is driven, one or more electrodes set in the impedance field can measure impedance. Additionally, for each set of driven patch pairs, there are many independent impedance fields between the non-driven patch pairs. That is, the non-driven patch pairs define independent impedance potential fields within the system. These independent impedance potential fields can be estimated and mapped to impedance measurements of electrodes at locations within the impedance field to define an impedance field. Such mapping of independent impedance potentials to measured impedances defines a model of the impedance field.

[0025] Once the impedance model is defined, an impedance value can be generated or predicted for the predicted positioning of the electrode in the impedance field. In one arrangement, the positioning of the physical electrode of the catheter can be predicted in three-dimensional space using a catheter model that models the catheter disposed in three-dimensional space. In such an arrangement, actual impedance measurements or values ​​for the physical electrodes can be obtained. Then, the measured impedance values ​​and the predicted impedance values ​​can be used to generate updated impedance values ​​and / or positioning of the electrodes. Such updated impedance values ​​can have improved accuracy compared to the predicted values ​​or measured values. In addition, the impedance model can be updated using the predicted values ​​and measured values. For example, these values ​​can update the definition of an independent impedance field.

[0026] In one arrangement, the independent impedance fields are defined as combinations of basis functions. In a particular arrangement, the impedance fields are defined as linear combinations of harmonic basis functions. The basis functions may include weighting factors that may be adjusted in a random process. In a further arrangement, the definition of the independent impedance fields may be further constrained. In another arrangement, the definition of the independent impedance fields may include error terms. Such error terms may include distance-related modeling errors and / or respiration-related modeling errors.

[0027] Various embodiments herein provide systems, methods and / or non-transitory computer-readable media storing instructions (i.e., utilities) for predicting magnetic values ​​of coordinates in a patient reference system while continuously updating these values ​​based on the patient's motion. The utility utilizes a time-varying patient reference sensor transform (e.g., a patient reference sensor model) that aligns the position and orientation of a patient reference sensor attached to the patient's body with the patient reference system. The transform allows continuous tracking of the patient's body's motion (e.g., relative to the patient's body relative to the initial or nominal position of the patient reference system). The utility further utilizes a time-varying magnetic transform (e.g., a magnetic model) that transforms between the patient reference system and the magnetic reference system of a magnetic-based medical positioning system. The transform predicts the magnetic values ​​of coordinates in the patient reference system. The utility is operable to apply a time-varying patient reference sensor transform to the coordinates (e.g., the predicted magnetic sensor positioning in the patient's body) to align the coordinates with the patient reference system and adjust the positioning of the coordinates based on the patient's motion. This generates patient coordinate system coordinates (e.g., the predicted positioning of the magnetic sensor in the patient reference system). The time-varying magnetic transformation can be applied to the patient coordinate system coordinates to identify the magnetic values ​​of the coordinates in the magnetic reference system. The utility can periodically or continuously update the movement of the patient reference sensor based on the time-varying transformation and / or the measurement of the corresponding magnetic sensor in the patient reference system. Similarly, the magnetic value of the coordinates can also be continuously updated. In an embodiment, such updates can occur 20 times per second, fifty times per second, or even 100 times per second. In such embodiments, the update is substantially continuous for the user, for example, viewing the output of the corresponding medical device on a display.

[0028] In one arrangement, the utility further includes predicting the response of a magnetic sensor of a catheter disposed in a patient reference system. In such an arrangement, a catheter model corresponding to the catheter can be used to predict the positioning of the magnetic sensor in the patient reference system. The position can define the coordinates to which the transformation is applied. That is, a time-varying transformation is applied to the predicted positioning of the magnetic sensor to predict the magnetic value (e.g., predicted value) of the magnetic sensor. A magnetic-based medical positioning system can then obtain magnetic measurements of the magnetic sensor. Magnetic measurements (e.g., observed measurements) and predicted values ​​can be used to calculate the positioning of the magnetic sensor in the patient reference system and / or update the time-varying transformation.

[0029] In one arrangement, the utility integrates (e.g., fuses) predicted magnetic values ​​and measured magnetic values ​​to improve the transformation. In an embodiment, the time-varying transformation (e.g., patient reference sensor model and magnetic model) is a variable model, wherein the parameters of the model are state variables of the state vector. Such variable systems allow multiple models to be updated based in part on the measured response of the physical system. In such an arrangement, the estimator system can estimate the potential (e.g., hidden) variables of each model to iteratively improve the correspondence of the model with the physical system they represent. In one arrangement, the estimator is an extended Kalman filter.

[0030] The foregoing and other aspects, features, details, utilities and advantages of the present invention will become apparent from a reading of the following description and claims, and from a reading of the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 A schematic block diagram of a system for determining the position of a medical device using impedance and magnetic measurements is shown.

[0032] Figure 2 A schematic block diagram showing an embodiment of an electrical impedance based positioning system.

[0033] Figures 3A-3D Shows suitable for Figure 2 An exemplary external impedance patch pair for use with a system.

[0034] Figure 4 An embodiment of a magnetic field based positioning system is shown.

[0035] Figure 5A A set of models for describing composite models according to the present disclosure is shown.

[0036] Figure 5B The prediction of the catheter shape and the transformation of the catheter shape to the patient reference frame are shown.

[0037] Figure 5CPredictions of predicted positioned measurements in the patient reference frame and observed measurements in the patient reference frame are shown.

[0038] Fig. 6A A catheter model with a magnetic sensor and multiple electrodes is shown.

[0039] Figure 6B A physical catheter and the corresponding catheter shape model are shown.

[0040] Figures 7A-7C Shown are the state distribution, the regularization function, and the application of the regularization function to the state distribution.

[0041] Figures 8A-8D Various views of the catheter shape model showing a planar catheter.

[0042] Fig. 9 A test system for determining shape parameters relative to deformation is shown.

[0043] Fig.10 The curvature of the proximal-distal section of the planar catheter is shown.

[0044] Figures 11A-11C Various transformations are shown depicted in the patient reference frame.

[0045] Fig. 12A and Fig. 12B The catheter model is shown oriented in the patient reference frame.

[0046] Fig.13A Mathematical diagram showing patch electrodes.

[0047] Fig. 13B Shows independent potential fields.

[0048] Fig.14 Shows the constraints defined by a set of external impedance patch pairs.

[0049] Fig.15 The breathing waveform is shown.

[0050] Fig.16 The state distribution is shown.

[0051] Fig.17 A one-dimensional comparison of the observed state and the measured state is shown.

[0052] Fig.18 The constraining manifold and the distribution of states that deviate from the manifold are shown.

[0053] Fig.19 A block diagram illustrating an example of a computer-readable medium in communication with a processing resource of a computing device according to an embodiment of the present disclosure.

[0054] Fig. 20A flow chart associated with determining a potential state of a system to identify electrode positioning is shown in accordance with an embodiment of the present disclosure.

[0055] Fig.21A and Fig. 22 B shows Fig. 20 A call graph of the interactions of modules.

[0056] Fig. 22 Show Fig.21A The call graph of the routine. DETAILED DESCRIPTION

[0057] Reference is now made to the drawings, wherein like reference numerals are used to identify like or similar components throughout the several views. Figure 1 is a schematic diagram of a system 10 in which a medical device, such as a guidewire, catheter, introducer (eg, sheath), incorporating a magnetic position sensor 28 and an electrode 30 may be used.

[0058] Before proceeding to a detailed description of the embodiments of the present disclosure, a description of an exemplary environment in which such devices and sensors may be used will first be set forth. Figure 1 As shown, the system 10 includes a main electronic control unit 12 (e.g., a processor) having various input / output mechanisms 14, a display 16, an optional image database 18, an electrocardiogram (ECG) monitor 20, a positioning system (such as a medical positioning system 22), a medical positioning system-enabled elongated medical device 24, a patient reference sensor 26, a magnetic position sensor 28, and an electrode 30. For simplicity, one magnetic position sensor 28 and one electrode 30 are shown, however, more than one magnetic position sensor 28 and / or more than one electrode 30 may be included in the system 10.

[0059] The input / output mechanism 14 may include conventional devices for interfacing with a computer-based control unit including, for example, one or more of a keyboard, mouse, tablet, foot pedal, switches, etc. The display 16 may also include conventional devices, such as a computer monitor.

[0060] Various embodiments described herein may find use in navigation applications that use real-time and / or pre-acquired images of regions of interest. Thus, the system 10 may optionally include an image database 18 to store image information related to the patient's body. The image information may include, for example, a region of interest surrounding a destination site of the medical device 24 and / or a plurality of regions of interest along a navigation path that is expected to be traversed by the medical device 24. The data in the image database 18 may include known image types, including (1) one or more two-dimensional static images acquired at respective times in the past; (2) a plurality of related two-dimensional images acquired in real time from an image acquisition device (e.g., fluoroscopic images from an x-ray imaging device), wherein the image database serves as a buffer (real-time fluoroscopic); and / or (3) a sequence of related two-dimensional images defining a cine-loop, wherein each image in the sequence has at least an ECG timing parameter associated therewith sufficient to allow playback of the sequence based on real-time ECG signals obtained from the ECG monitor 20. It should be understood that the foregoing embodiments are merely examples and are not limiting in nature. For example, the image database may also include three-dimensional image data. It should be further understood that the images may be acquired by any imaging modality now known or later developed (eg, X-ray, ultrasound, computed tomography, magnetic resonance imaging, etc.).

[0061] The ECG monitor 20 is configured to continuously detect electrical timing signals of the heart organ by using a plurality of ECG electrodes (not shown), which may be externally fixed to the patient's body. The timing signals correspond in particular to a particular phase of the cardiac cycle. Typically, the ECG signal may be used by the control unit 12 for ECG synchronization playback of previously captured image sequences (image playback) stored in the database 18. Both the ECG monitor 20 and the ECG electrodes may include conventional components.

[0062] Another medical positioning system sensor, a patient reference sensor (PRS) 26 (if provided in the system 10), can be configured to provide a position reference of the patient's body to allow motion compensation for patient body motion (such as motion caused by breathing). Such motion compensation is described in more detail in U.S. patent application Ser. No. 12 / 650,932, entitled "Compensation of Motion in a Moving Organ Using an Internal Position Reference Sensor," which is incorporated herein by reference in its entirety as if fully set forth herein. The PRS 26 can be attached to the patient's sternum or other location. The PRS 26 can be configured to detect one or more characteristics of the magnetic field in which it is located, wherein the medical positioning system 22 determines a positioning reading (e.g., a P&O reading) indicating the position and orientation of the PRS in a magnetic reference coordinate system.

[0063] The medical positioning system 22 is configured to function as a positioning system and thus determine position (positioning) data relative to one or more magnetic position sensors 28 and / or electrodes 30 and output corresponding positioning readings. In an embodiment, the medical positioning system 22 may include a first medical positioning system or an electrical impedance-based medical positioning system 22A that determines the electrode positioning in a first coordinate system, and a second medical positioning system or a magnetic field-based medical positioning system 22B that determines the magnetic position sensor in a second coordinate system. In an embodiment, the positioning readings may each include at least one or both of position and orientation (P&O) relative to a reference coordinate system (e.g., a magnetic-based coordinate system or an impedance-based coordinate system). For some types of sensors, the P&O may be expressed in five degrees of freedom (five DOF) as a three-dimensional (3D) position (e.g., coordinates in three perpendicular axes X, Y, and Z) and a two-dimensional (2D) orientation (e.g., pitch and yaw) of the electromagnetic position sensor 28 relative to the magnetic field of a magnetic field generator or transmitter and / or relative to the electrode 30 in the electric field applied by an electric field generator (e.g., a set of electrode patches). For other sensor types, P&O can be expressed as 3D position (eg, X, Y, Z coordinates) and 3D orientation (eg, roll, pitch, and yaw) using six degrees of freedom (six DOF).

[0064] The impedance-based medical positioning system 22A determines electrode positioning based on capturing and processing signals received from the electrodes 30 and external electrode patches when the electrodes are placed in a controlled electric field (eg, an electric potential field), such as generated by the electrode patches. Figure 22 is a schematic overview of an exemplary electrical impedance-based medical positioning system ("MPS system") 22A. The MPS system 22A may include a medical positioning system as known in the art (including, for example, EnSite® commercially available from St. Jude Medical, Inc.). TM Electroanatomical mapping systems) or a variety of visualization, mapping and navigation components as generally seen by reference to U.S. Patent No. 7,263,397 to Hauck et al., entitled “Method and Apparatus for Catheter Navigation and Location and Mapping in the Heart” or U.S. Patent Application No. 2007 / 0060833A1 to Hauck, entitled “Method of Scaling Navigation Signals to Account for Impedance Drift in Tissue,” both patents are owned by the common assignees of the present invention and both are incorporated herein by reference in their entirety.

[0065] The medical positioning system 22A includes a schematic diagram of a heart 52 of a patient 54. The system includes the ability to determine the catheter electrode location (i.e., position and orientation) as the catheter distal end moves around and within the chambers of the heart 52. To this end, three sets of body surface electrodes (patches) are shown: (1) electrodes 56, 58 (X axis); (2) electrodes 60, 62 (Y axis); and (3) electrodes 64, 66 (Z axis). In addition, a body surface electrode ("abdominal patch") 68 is schematically shown. All surface electrodes are connected to a switch 70. Of course, other surface electrode configurations and combinations are also suitable for use with the present invention, including fewer electrodes (e.g., three electrodes), more electrodes (e.g., twelve electrodes), or different physical arrangements, for example, a linear arrangement rather than an orthogonal arrangement.

[0066] Medical device 24 is shown as a catheter having a distal electrode 30. Catheter 24 may have additional electrodes in addition to electrode 30 (eg, catheter tip electrodes and / or ring electrodes) and one or more magnetic position sensors (not shown). Figure 2 Also shown is a second separate catheter 74 having a fixed reference electrode 76 that can be fixed to the heart for calibration purposes. In many cases, a coronary sinus electrode or other fixed reference electrode 76 in the heart 52 can be used as a reference for measuring voltage and displacement.

[0067] It should be understood that the catheter 24 may also include other electrodes, and in other embodiments, such as in EP or RF ablation embodiments, the other electrodes may be used for any number of diagnostic and / or therapeutic purposes. For example, such electrodes, and therefore such catheters, may be used to perform ablation procedures, cardiac mapping, electrophysiological (EP) studies, and other diagnostic and / or therapeutic procedures. The embodiments are not limited to any one type of catheter or catheter-based system or procedure.

[0068] Figure 2 Further shown are a computer system 78, a signal generator 80, an analog-to-digital converter 82, and a low-pass filter 84. The computer system 78 includes a processing device configured to perform various functions and operations described herein. The computer system 78 can be configured to control the signal generator 80 according to a predetermined strategy to selectively excite various pairs (dipoles) of surface electrodes. In operation, the computer system 78 can (1) obtain raw patch data (i.e., voltage readings) via the filter 84 and the analog-to-digital converter 82, and (2) use the raw patch data (in combination with the electrode measurements) to determine the raw uncompensated electrode location coordinates of the catheter electrode (such as electrode 30) located within its heart or chamber in a three-dimensional coordinate system (e.g., an impedance-based coordinate system). The computer system 78 can be further configured to perform one or more compensation and adjustment functions and output the location of one or more electrodes, such as electrode 72, in the coordinate system 14. Motion compensation may include, for example, compensation for respiration-induced patient body motion as described in U.S. Patent Application Serial No. 12 / 980,515, entitled “Dynamic Adaptive Respiration Compensation with Automatic Gain Control,” the entire contents of which are incorporated herein by reference.

[0069] Each body surface (patch) electrode is independently coupled to a switch 70, and the electrode pair is selected by software running on a computer system 78, which couples the patch to a signal generator 80. A pair of electrodes, such as Z-axis electrodes 64 and 66, can be excited by a signal generator 80 to generate an electric field in the body and heart 52 of the patient 54. In one embodiment, the electrode excitation process occurs quickly and sequentially because different groups of patch electrodes are selected, and one or more unexcited (in an embodiment) surface electrodes are used to measure voltage. During the delivery of an excitation signal (e.g., a current pulse), the remaining (unexcited) patch electrodes can be referenced to the abdominal patch 68, and the voltage applied to these remaining electrodes is measured by an analog-to-digital converter 82. In this way, the surface patch electrodes are divided into a driving electrode group and a non-driving electrode group. A low-pass filter 84 can process voltage measurements. The filtered voltage measurements are converted to digital data by an analog-to-digital converter 82 under the management of the software and transmitted to a computer 78 for storage. This collection of voltage measurements is referred to as "patch data" in this article. The software has access to every individual voltage measurement made at each surface electrode during each stimulation of each pair of surface electrodes.

[0070] The patch data is used together with the measurements made at the electrode 30 to determine the relative positioning of the electrode 30 in a so-called patient-based coordinate system or patient reference system 6. That is, since the patch is applied directly to the patient, the patient defines the reference system for the impedance measurement. Except when a specific surface electrode pair is driven (in an embodiment), the potential across each of the six orthogonal surface electrodes can be acquired for all samples. In one embodiment, sampling is generally avoided when the surface electrode acts as a source or sink in the driven pair, because the potential measured at the driven electrode during this period may be deviated by the effects of electrode impedance and high local current density. However, in alternative embodiments, sampling can occur at all patches (even driven patches).

[0071] Typically, in one embodiment, three nominally orthogonal electric fields are generated by a series of driven and sensed electric dipoles to achieve the catheter positioning function in a biological conductor. Alternatively, these orthogonal fields can be decomposed and any pair of surface electrodes (e.g., non-orthogonal) can be driven as dipoles to provide effective electrode triangulation. Figures 3A-3D A plurality of exemplary non-orthogonal dipoles are shown, labeled D0, D1, D2, and D3, set in an impedance-based coordinate system 2. Figures 3A-3D In the diagram, the X-axis surface electrode is marked as X A and X B , mark the Y-axis surface electrode as Y A and Y B , and the Z-axis electrode is marked as Z A and ZB For any desired axis, the potentials measured across the intracardiac electrodes 30 produced by a set of predetermined drive (source-sink) configurations can be combined algebraically to produce an effective potential that is the same as that obtained by simply driving a uniform current along an orthogonal axis. The surface electrodes 56, 58, 60, 62, 64, 66 (see Figure 2 ) as dipole source and drain relative to a ground reference (e.g., abdominal patch 68), while the unexcited body surface electrodes measure the voltage relative to the ground reference. The measuring electrodes 30 placed in the heart 52 are also exposed to the electric field from the current pulses and are measured relative to ground (e.g., abdominal patch 68). In practice, one or more catheters within the heart may contain multiple electrodes, and each electrode potential may be measured individually. As previously described, alternatively, at least one electrode may be fixed to the inner surface of the heart to form a fixed reference electrode 76, which may also be measured relative to ground.

[0072] The data sets from each of the surface electrodes and the internal electrodes are all used to determine the location of the measurement electrode 30 within the heart 52. After the voltage measurement is made, the surface electrodes of different pairs are excited by the current source, and the voltage measurement process of the remaining patch electrodes and the internal electrodes is performed. This sequence occurs rapidly, for example, on the order of about 100 times per second in an embodiment. Generally, the voltage on the electrodes within the heart has a linear relationship with the position between the patch electrodes that establish the field within the heart, as described more fully in the above-mentioned U.S. Patent No. 7,263,397.

[0073] The magnetic-based medical positioning system 22B determines the magnetic position sensor position (e.g., P&O) in a magnetic coordinate system based on capturing and processing signals received from the magnetic position sensor 28 when the sensor is placed in a controlled low-intensity alternating current (AC) magnetic (e.g., magnetic field). Each magnetic position sensor 28, etc., may include a coil, and from an electromagnetic perspective, when the coil is in a magnetic field, the changing magnetic field or AC magnetic field can induce a current in the coil. Therefore, the magnetic position sensor 28 is configured to detect one or more characteristics (e.g., flux) of the magnetic field in which it is placed and generate a signal indicative of those characteristics, which is further processed by the medical positioning system 22B to obtain a corresponding P&O of the magnetic sensor 28 relative to, for example, a magnetic field generator.

[0074] Figure 4 8 is a schematic diagram of an exemplary magnetic field-based medical positioning system 22B in a fluoroscopy-based imaging environment, which is labeled as system 88. A magnetic field generator or magnetic transmitter assembly (MTA) 90 and a magnetic processing core 92 for determining position and orientation (P&O) readings generally define the magnetic field-based positioning system 22B. The MTA 90 is configured to Figure 4 A magnetic field is generated in and around the patient's chest in a predetermined three-dimensional space marked as a motion frame 94 in the figure. Magnetic field sensors coupled to a device 24 (e.g., a catheter or another medical device) are configured to sense one or more characteristics of the magnetic field, and when the sensor is in the motion frame 94, each sensor generates a corresponding signal provided to the magnetic processing core 92. The processing core 92 responds to these detected signals and is configured to calculate the corresponding three-dimensional position and orientation (P&O) readings of each magnetic field sensor. Therefore, the MPS system 22B enables real-time tracking of each magnetic field sensor in a three-dimensional space that forms a magnetic-based coordinate system 4. The position of the sensor can be shown on the display 96 relative to, for example, only a heart model or geometric shape. Additional exemplary embodiments of magnetic field-based medical positioning systems are described in commonly owned U.S. Patent No. 7,386,339 and U.S. Patent Application No. 2013 / 0066193, which are incorporated herein by reference in their entirety. It should be understood that numerous variations are possible, as seen, for example, by reference to U.S. Patent Nos. 7,197,354 and 6,233,476, which are also incorporated herein by reference in their entirety. Figure 2 Unlike the discussed electrical impedance based systems, which have an origin based on the patient reference system 6 when body surface electrodes are applied directly to the patient, the origin of the magnetic field based system is typically based in or on the MTA 90 (e.g., as shown by the dashed line) and is independent of the patient. In other words, the patient coordinate system (e.g., patient reference system) 6 and the magnetic based coordinate system 4 have different origins.

[0075] like Figure 4As further shown in , a patient reference sensor (PRS) 26 can be applied to the patient. In an embodiment, the PRS 26 can be attached to the patient's sternal manubrium. However, other patient locations for the PRS 26 are also possible. In an embodiment, the PRS 26 is a magnetic sensor configured to detect one or more characteristics of a magnetic field in which it is disposed, wherein the medical positioning system 22B determines a positioning reading (e.g., a P&O reading) indicating the position and orientation (e.g., in a magnetic-based coordinate system) of the PRS 26. For the present application, the PRS defines an origin (e.g., PRF 0, 0, 0) in a patient reference coordinate system or patient reference frame 6 (PRF). The origin may deviate from the actual location of the sensor. That is, a predetermined offset (e.g., x, y, and z) can be applied to the PRS measurement, which corresponds to an estimated distance between the placement of the sensor on the patient and the desired origin. For example, the origin can be offset from the sensor so that it is within the heart of a patient for cardiac applications. In addition, two or more PRSs can be applied to provide additional orientation information for the PRF 6. In any embodiment, because the PRS 26 is attached to the patient and moves with the patient's movement, the origin of the PRF 6 also moves with it. Such movement may be due to the patient's breathing and / or the patient's body movement (drift, roll, etc.). Therefore, the origin of the PRF 6 depends on the patient's position and can be updated over time. More specifically, a measurement of the PRS can be determined in a magnetic field coordinate system, and the measurement can be used as the origin of the PRF (e.g., by adjustment).

[0076] As previously described, impedance-based medical positioning systems and magnetic-based medical positioning systems have different advantages and disadvantages. For example, impedance-based systems provide the ability to simultaneously locate a relatively large number of electrodes. However, since impedance-based systems utilize the flow of current in the human body, the system may suffer from measurement inaccuracies due to offsets and / or drifts caused by a variety of physiological phenomena (e.g., local conductivity changes, sweat / patch interactions, etc.). In addition, impedance-based systems may be subject to electrical interference. As a result, electrode positioning, rendering, geometry, and / or representation based on such impedance-based measurements may be distorted. On the other hand, magnetic-based systems do not depend on the characteristics of the patient's anatomy and are believed to provide higher accuracy. However, magnetic position sensors are typically limited to tracking relatively few sensors.

[0077] Efforts have been made to provide a system that combines the advantages of an electrical impedance-based positioning system (e.g., the positioning of multiple electrodes) with the advantages of a magnetic field-based coordinate system (e.g., independence from the patient's anatomical structure, higher accuracy). In an embodiment, such a system can be provided by aligning the coordinate system of an electrical impedance-based positioning system with the coordinate system of a magnetic field-based positioning system. In such an arrangement, the positioning of an electrode in an impedance-based coordinate system can be identified in conjunction with the positioning of one or more magnetic sensors in a magnetic-based coordinate system. In an embodiment, at least a portion of an electrode and a magnetic sensor can be co-located to define a reference pair. The co-location allows the determination of a transformation (e.g., a transformation matrix) between coordinate systems. Once the transformation is determined, the transformation can be applied to the positioning of any electrode to align these positionings in a magnetic-based coordinate system. Therefore, an electrical impedance-based electrode can be identified in the coordinate system of a magnetic field-based positioning system, thereby improving the positioning accuracy of the electrode. Such a system is described in co-owned U.S. Patent Publication No. 2013 / 0066193, incorporated herein as described above.

[0078] While providing improved electrode positioning, the determination of the transformation between an impedance-based coordinate system and a magnetic-based impedance system and the subsequent registration of the electrode positioning to the magnetic coordinate system may not account for the various impedance shifts and / or drifts associated with the electrodes. That is, impedance-based systems may suffer from nonlinear shifts and / or drifts due to physiological phenomena. Along these lines, previous efforts have been made to identify shifts and / or drifts and apply corrections to the registration. Such systems are set forth in commonly owned U.S. Patent Publication No. 2016 / 0367168, which is incorporated herein by reference in its entirety. Typically, such systems determine the transformation between an impedance-based system and a magnetic-based system and apply corrections to electrode positioning.

[0079] Previous systems that utilize electrode information (e.g., impedance measurements) and magnetic sensor information to provide improved electrode positioning in three-dimensional space (e.g., in a patient's body) rely primarily on impedance-based measurements. That is, magnetic sensor information (e.g., magnetic sensor measurements) provides additional accuracy. This can be described as an impedance-based positioning arrangement. Due to the distortion of impedance measurements and temporal instability, such arrangements may suffer from instability. In addition, in some cases, previous impedance-based positioning arrangements cannot resolve multiple errors within the system. For example, the transformation between an impedance-based coordinate system and a magnetic-based impedance system may underestimate the errors or uncertainties in electrode and / or magnetic sensor measurements. As a further example, such systems may not be able to consider other system inputs (e.g., patient motion, shape of medical devices, etc.) that may affect the calculated positioning or position of the electrode. In short, the registration of an impedance-based system to a magnetic-based system may not include additional information that can be observed and / or inferred and can improve the overall identification of the catheter and / or electrode position in three-dimensional space.

[0080] In order to provide an improved system for determining the positioning of electrodes in a three-dimensional space such as within a patient, the present disclosure relates to a positioning arrangement (e.g., a sensor fusion process or algorithm) that continuously integrates (e.g., fuses) impedance measurements from electrodes and external patches with position and orientation measurements from magnetic sensors to estimate the potential state (e.g., position) of a medical device set within a patient reference frame. The potential state is used to track catheter electrodes within the patient's body, just as if there were a magnetic sensor on each catheter electrode, thereby achieving both accuracy and stability. More broadly, the proposed arrangement expands the number of observation parameters for positioning electrodes within the patient reference frame without relying on a direct transformation between an impedance-based coordinate system and a magnetic-based impedance coordinate system based on the presence of a reference electrode and sensor pair. The systems and methods of the present disclosure do not require a reference pair. Instead, impedance measurements and magnetic measurements are used as inputs to an overall system model that estimates / predicts and updates catheter electrode positioning in the patient reference frame. Catheter and / or electrode positioning can be tracked using magnetic measurements and impedance measurements.

[0081] Figure 5AAn embodiment of an independent model for mathematically defining a catheter and / or electrode positioning system model is shown. That is, the independent models define a composite model 40 of the system (e.g., in a patient reference frame). The illustrated embodiment of the composite system model 40 includes five models: a catheter model 42 (e.g., a medical device model) that predicts the shape of a catheter having one or more electrodes and / or magnetic sensors in a catheter reference frame 8 (e.g., a catheter configuration); a catheter position and orientation model 44 that transforms the catheter model from the catheter reference frame 8 to the patient reference frame 6 based on a unique transformation specific to the catheter; a magnetic model 46 that predicts magnetic sensor measurements in the patient reference frame; an impedance model 48 that predicts electrode impedance measurements in the patient reference frame; and a breathing model 55 that predicts artifacts in the predicted impedance and / or magnetic measurements based on patient breathing. Each model mathematically describes a portion of the entire system. However, it will be appreciated that not all models are required for the composite model. That is, the composite model may use different combinations of some or all of the models. In an embodiment, the magnetic model further includes a patient reference sensor model 57 that tracks adjustments of the position of the PRS 26 relative to the patient reference frame.

[0082] Figure 5B The cooperation of one of the multiple models is further shown. Initially, the catheter model 42 predicts the catheter shape of the corresponding physical catheter 50 set in a three-dimensional space, such as a patient's body (e.g., heart 52), wherein the physical catheter 50 has a set of electrodes 331-334 and a magnetic sensor 282. In the illustrated embodiment, the catheter shape model 42 includes the model positions or locations of the model electrodes 301-304 and the model magnetic sensor 281 (that is, corresponding to the physical electrodes 331-334 and the magnetic sensor 282) in the catheter reference system 8. The position and orientation model 44 applies one or more transformations to the catheter model 42 to convert the model from the catheter reference system 8 to the patient reference system 6. When transforming, the positioning (e.g., predicted positioning) of the model electrodes 301-304 and / or the model magnetic sensor 281 in the patient reference system 6 is predicted (e.g., projected), as shown by the solid circles of the electrodes 301-304 shown in the patient's heart 52 and the vector for the magnetic sensor 281. The impedance model 48 predicts impedance responses or measurements 311-314 for the predicted electrode positioning of the model electrodes 301-304 in the patient reference frame, while the magnetic model 46 predicts responses or measurements for the predicted positioning of the model sensor 281 in the patient reference frame. Figure 5C, where the predicted electrode response (e.g., location) 311-314 for each predicted model electrode location is represented by solid dots and the predicted magnetic measurement 291 of the model sensor 281 is represented by a solid vector. The impedance-based medical positioning system measures the actual response 351-354 (e.g., observed measurements) of the physical electrodes 331-334 within the patient's body (e.g., within the patient's reference frame) to the applied electric potential field to determine the response (e.g., location) of the electrodes, as shown by the dashed circles. If utilized, the magnetic-based medical positioning system measures the response (e.g., location) 282 of the magnetic sensor within the patient's body, as shown by the dashed vector 292. As shown Figure 5C As shown in the enlarged portion of , the measured response of the physical electrode (e.g., 351) and / or sensor (not shown) and the predicted response of the electrode (e.g., 311) and / or sensor (not shown) each contain unknown errors or noise (e.g., uncertainties). In an embodiment, the predicted response includes breathing artifacts from the breathing model 55. In an embodiment, the uncertainties of the measured response and the predicted response can partially overlap. The predicted measurements and the observed measurements are then used to predict the actual position (e.g., updated position) or calculate the position of the electrodes 371-374, such as Figure 5C As shown by X' in Figure 5C As shown in the enlarged portion of , the calculated position 371 can be located in the overlap of the predicted response position and the measured response position. In any embodiment, the calculated position is generally more accurate than the position generated by the predicted response or the observed response. The calculated position can then be output to a display. See, for example, Figure 1 That is, an updated representation or rendering of the catheter or other medical device may be output to a display using the calculated positioning.

[0083] Catheter Model

[0084] A simplified catheter model is provided below (i.e., Fig. 6A ), which allows identification of the positioning of the magnetic sensors and electrodes within the catheter reference frame. Fig. 6A The model of involves a rigid catheter with a single magnetic sensor and four electrodes with known orientations relative to the magnetic sensor. However, it will be understood that other more complex catheter models are possible and will be relative to Figure 6B-8DSuch complex catheter models are further discussed. As described below, more complex catheter models can provide catheter deformations such that the model includes deformable portions (e.g., a small amount of curvature and torsion along a Frenet-Serret reference frame) to be used with rigid body transformations (e.g., unit quaternions and translations) to describe the catheter shape, and / or position and orientation in a patient reference frame. In an example, a catheter model for determining electrode positioning in a catheter reference frame is described in U.S. Provisional Application No. 62 / 756,915, entitled “Mechanical Models of Catheters for Sensor Fusion Processes,” filed on November 7, 2018, the entire contents of which are incorporated herein by reference.

[0085] Reference again Fig. 6A , depicts a side view of an exemplary medical device or catheter 24, wherein the catheter 24 has a single magnetic position sensor 28 and four electrodes 30-1, 30-2, 30-3, 30-4 (hereinafter 30, unless specifically referenced). In order to reduce the complexity of the model or reduce the dimensionality of the model (e.g., the number of model parameters), it may be desirable to determine the position and orientation of the electrode in the catheter reference system based on the position of the magnetic sensor. For example, using a specification associated with the catheter (e.g., a manufacturer specification detailing the position of the electrode 30 relative to the magnetic position sensor 28), the position of the electrode in the catheter reference system can be determined from the position of the magnetic sensor. For example, based on the position and orientation of the magnetic position sensor 28 (e.g., a five-degree-of-freedom sensor or a six-degree-of-freedom sensor), a vector for the magnetic position sensor can be determined. In some embodiments, the vector can be in a direction facing the distal end of the magnetic position sensor 28 (e.g., a magnetic coil) and can be coaxial with the magnetic position sensor 28. Because the magnetic position sensor 28 is disposed within the shaft of a rigid catheter, the position and orientation of the catheter shaft can be determined based on the vector associated with the magnetic position sensor. Specifications (e.g., manufacturer specifications) associated with the positioning of one or more electrodes 30 on the shaft relative to the magnetic position sensor 28 can be used to determine the model position of the electrode 30 in the catheter reference frame (i.e., along the vector ). Thus, a model equation (e.g., a state vector) can be determined that identifies the position of the sensor 28 and the electrode 30 in the catheter reference frame. That is, in combination with the sensor positioning and the electrode spacing, all electrode and sensor positions and orientations in the catheter reference frame are known.

[0086] Figure 6BAn embodiment of a deformable physical catheter 24 and a corresponding catheter shape model 124 is shown. The deformable catheter 24 includes a single catheter spline, a plurality of electrodes 30, and a magnetic sensor 28. In this embodiment, the catheter model divides the spline into two model segments, namely a proximal shaft segment 132 and a distal hoop segment 134. Each segment 132, 134 is described by a moving Frenet coordinate system of constant parameters that follows the arc of the corresponding segment of the physical catheter. According to the mechanical specification, the model electrode 146 is located on the distal hoop model segment 134. For example, the position of each electrode can be defined by its distance along the length λ of the Frenet coordinate system or the length l (e.g., starting from the origin of the coordinate system). In this embodiment, all electrodes are shown as being located on the distal hoop segment 134, however, it should be understood that each model segment may include an electrode, depending on the physical configuration of the physical catheter 24. In this embodiment, the proximal shaft model segment 132 includes a single model magnetic sensor 128. Again, it will be appreciated that each model segment may include one or more magnetic sensors and / or one or more electrodes. Therefore, the parameterization of the model segments completely describes the electrode positioning in the catheter reference frame 8 of the catheter model 124 .

[0087] Frenet's formula describes the geometric properties of continuously differentiable curves in three-dimensional space. More specifically, Frenet's formula describes the derivatives of the tangent "T", normal "N", and binormal "B" unit vectors in terms of each other at every point along a coordinate system length λ. See Figure 6B The tangent, normal, and binormal unit vectors are defined, or collectively referred to as the Frenet coordinate system, where T is the unit vector tangent to the curve, pointing in the direction of motion, N is the normal unit vector, the derivative of T with respect to the arc length parameter of the curve divided by its length, and B is the binormal unit vector, which is the cross product of T and N. The Frenet formula is:

[0088]

[0089]

[0090]

[0091] where d / ds is the derivative with respect to arc length, κ is the curvature (e.g., the inverse or radius of the curve), and τ is the twist of the curve. The two scalars κ and τ effectively define the curvature and twist of the curve. For each segment in the homogeneous coordinate system, the Frenet coordinate system (F F ) along the curve at a distance λ is defined as:

[0092]

[0093] The use of the Frenet coordinate system effectively allows each model segment to be defined using two parameters, curvature κ and torsion τ.

[0094] exist Figure 6B In an embodiment of the present invention, the catheter shape model includes two continuous curves (e.g., model segments) of constant curvature and torsion rotated 90 degrees from each other. The first curve represents the bend between the proximal shaft segment 132 and the distal hoop segment 134. The first curve is defined by κ1 and torsion τ1. The second curve represents the distal hoop segment 134. The second curve is defined by κ2 and torsion τ2. Therefore, the catheter shape model 124 is defined by four parameters: two curvatures and two torsion, which define all possible shapes that the catheter model can take. These parameters typically each have a predetermined or experimentally determined range of values ​​(e.g., from a corresponding physical catheter). In addition, the curve parameters typically form state variables in a random process that predicts the potential shape of the catheter model. For a given model, the positioning of the model electrodes and / or magnetic sensors can be derived from their known positioning along their corresponding coordinate systems.

[0095] In an embodiment, the state transition model (eg, matrix) that applies the effect of each curve parameter at time k-1 to the curve parameter at time k is as follows:

[0096]

[0097]

[0098] in:

[0099] i represents a curve segment (for example, in this embodiment, i=1 or 2);

[0100] Indicates the default curvature of each curve segment;

[0101] Indicates the default twist for each curve segment; and

[0102] A matrix representing the force factors that define each curve parameter.

[0103] When the transformation matrix is ​​applied, each of the state variables is changed to generate a plurality of possible catheter shapes. In an embodiment, this produces a state distribution of possible catheter shapes. See Fig. 7A In general, the mean of the state distribution represents the most likely catheter shape and corresponding set of catheter parameters.

[0104] In embodiments, the force factor may be derived from the mechanical parameters specific to the catheter. In embodiments, the force factor may represent the return force of the shape wires forming the splines of the catheter. In such embodiments, the force factor F applies a return force to the deformation associated with a given shape parameter, which represents the force applied by the shape wires in an attempt to return to an undeformed or nominal state from the current shape parameter. In embodiments of a single-segment catheter, which is nominally straight and untwisted, the torsional stiffness is the same as the rotational stiffness:

[0105]

[0106]

[0107]

[0108]

[0109] It will be appreciated that such force factors F are unique to a particular catheter. Introducing the force factors prevents the state transition model from being consistent with the previous state.

[0110] By applying the above mentioned transformation model, the state distribution of potential catheter shapes in the catheter reference frame can be estimated at time k. Fig. 7A An exemplary distribution is shown. Based on the most likely shape (eg, the mean of the distribution), the positioning of the model electrode in the catheter reference frame can be determined.

[0111] An observation model may be implemented to map state parameters into a physical domain (e.g., a catheter reference frame). In an embodiment, this is performed by evaluating a matrix index of a Frenet coordinate system. In an embodiment, the matrix index is an integro-differential matrix with constant terms (e.g., curvature and torsion) over the arc length for all electrodes, where the position l of the electrode varies over the arc length. In an embodiment, the matrix evaluation may be calculated using Givens rotations and trigonometric functions.

[0112] In an embodiment, the Givens rotation is initially calculated to eliminate two terms of the Frenet coordinate system:

[0113]

[0114] So that:

[0115]

[0116] After expanding the first few terms of the remaining matrix exponentials, the following trigonometric series identities can be identified:

[0117]

[0118] where Φ is the transformation from state space to the catheter reference frame. For a full Φ matrix, it is useful to preserve the Givens rotation in the solution. However, the last row is evaluated for each electrode on the hoop and contains the Cartesian coordinates for a given arc length along the curve:

[0119]

[0120] where P is the coordinate at position l in the Frenet coordinate system. The model electrodes and / or coils are then identified in the catheter reference frame by calculating the arc length along the specific curve for each electrode, calculating P as described above, and combining it with any Φ that may be closer.

[0121] for Figure 6B Model electrode on the distal cuff of the model (eg, subscript 2 in the current embodiment):

[0122]

[0123] in:

[0124] C i is the position of each electrode in the catheter reference frame;

[0125] λ2 is the length of the distal hoop curve;

[0126] Δ i′ is the intra-electrode distance specification (e.g., center-to-center);

[0127] Φ h is a transformation between the curves of the first and second Frenet coordinate systems to provide smoothness, and in an embodiment where the first and second coordinate systems have a 90 degree clockwise rotation, it is:

[0128]

[0129] Φ1 is the transformation of the distal hoop curve; and

[0130] λ1 is the length of the proximal shaft curve.

[0131] for Figure 6B Magnetic sensor or electrode on the proximal shaft of the two-segment catheter model (if present):

[0132]

[0133] Once the positions of the model electrodes and / or magnetic sensors in the catheter reference frame are known, they can be converted to the patient reference frame using any appropriate transformation. In an embodiment, a six-degree-of-freedom rigid transformation is used to position and orient the catheter model of the electrodes and magnetic sensors into the patient reference frame based on the position and orientation of the magnetic sensors relative to the position and orientation of the magnetic patient reference sensor. For each model electrode position in the patient reference frame, an impedance measurement can be predicted and an impedance measurement can be obtained (e.g., observed) from the physical electrode. The predicted measurements and observed measurements can be used to update the parameters of the catheter model to more closely approximate the physical configuration of the deformable catheter.

[0134] While the modeling of a relatively simple single spline conduit has been discussed previously, it will be appreciated that more complex conduits may be modeled based on a limited set of parameters. Figures 8A-8D A planar catheter 140 is shown having a substantially rigid shaft 142 having one or more magnetic sensors 148 and one or more electrodes (not shown); and a flexible paddle 144 including sixteen electrodes 146 arranged in a square matrix. 1-16 (hereinafter 146 unless otherwise noted). In an embodiment, the planar catheter 140 corresponds to an HD grid catheter commercially available from Abbott Laboratories, Lake Bluff, Illinois, USA. In the embodiment shown, the flexible paddle 148 is defined by four shaped wires, each of which supports four electrodes 146. In the non-deflected or relaxed state, the flexible paddle 144 is substantially planar in the XZ plane, with its origin at the end of the rigid shaft. Reference axis x Extending longitudinally from the origin, for example, axially aligned with a rigid shaft. In an embodiment, the catheter 140 is modeled as a curved plane with two model segments (proximal and distal). In addition to the curvature, the axis of curvature of the distal segment can be rotated to capture off-axis deformation, and the two segments can be laterally tilted. The three-dimensional positioning of the electrode can then be determined by the two-dimensional positioning on the curved plane. In addition, the positioning of the electrode and / or sensor can be defined along the length of multiple planes.

[0135] In an embodiment, the planar catheter is modeled by four parameters, which are base curvature, paddle curvature, tilt angle, and tube curvature. Such parameters relate to physical actions that may occur during the course of a clinical procedure that will cause the catheter 140 to take a particular shape (e.g., deform). For example, during a cardiac procedure, the catheter 140 is often pressed against the heart wall and / or pushed into a lumen (e.g., a blood vessel, an artery). Figure 8B As shown in the side view of FIG. 1 , pressure on the heart wall generally causes the base curvature and the paddle curvature to change from a relaxed state, wherein the paddle 144 is moved from the reference axis to the x In addition, Figure 8CAs shown in , pressing on the sidewall may result in a tilt of the paddle 144 relative to the reference axis x. Finally, moving the catheter into the lumen may result in a cylindrical curvature along the length of the paddle 144, as shown in Fig.8D as shown in .

[0136] Base curvature κ b and propeller curvature κ P exist Figure 8B As shown in the figure, the base curvature κ b corresponds to the curvature (eg, the inverse of the radius) of the proximal segment 150 of the paddle 144, and the paddle curvature κ P The curvature (e.g., the inverse or radius R1) of the distal segment 152 corresponding to the paddle 144. Through experimentation, it has been determined that for a corresponding physical catheter, vertical pressure on the paddle 144 (e.g., applied to the distal tip without tilting) causes the proximal segment 150 to bend in a consistent manner. Thus, the base curvature κ b κ may be expressed by a single curvature parameter having a range of values ​​(e.g., ±0.25) that may be established based on the desired curvature or determined experimentally. Typically, the proximal segment 150 of the paddle 144 is less rigid than the distal segment 152 of the paddle when the same pressure is applied. However, the curvatures of the two segments are related. In an embodiment, the paddle curvature κ P and the base curvature κ b The relationship between can be expressed as:

[0137] κ p ≈c·f(κ b )

[0138] The functional factor f(κ b ) is positive. This relationship can be determined experimentally, in which multiple paddle deformations are examined (e.g., in a bench test) to determine the shape or range of the base curvature. In an embodiment, the base curvature κ can be prepared b and propeller curvature κ P A graph of the relative values ​​of , such that, for example, a best fit curve can define the relationship of the parameters. In an embodiment, the relationship between these curvatures of a particular catheter is found to be:

[0139] κ p ≈c·f(κ b )=c·c1arctan(c2 κ b )

[0140] Among them, c1 and c2 are constants determined experimentally.

[0141] Figure 8CThe tilt angle applied to the paddle is shown. When vertical pressure is applied to the device, it can be imagined to wrap around a generalized cylinder with a transverse axis (i.e., perpendicular to the direction of the rigid shaft). An inclined pressure also produces a generalized cylindrical surface, but its axis is not transverse, but at a certain angle α. The tilt angle can have an established range (e.g., ±45°). Fig.8D shows the curvature or tube curvature κ along the length of the paddle b For example, when the catheter is pushed into the lumen, the paddle acquires a tubular shape having a curvature that is generally transverse to the longitudinal or reference axis of the catheter. x The catheter model utilizes the four noted parameters to define all possible shapes (e.g., states) that the planar catheter 140 can take. Again, these parameters can define state variables in a stochastic process that predicts the model's potential shapes. Thus, based on the known spacing of the electrodes, their locations can be determined for possible states in the catheter reference frame in a manner similar to that described above.

[0142] As part of the stochastic process, a catheter model can be implemented to estimate the shape or state of the catheter. In this type of arrangement, the catheter model can be used to predict or estimate the current shape of the catheter, and thus predict or estimate the location of the electrode in the catheter reference system based on the previously known shape of the catheter. In this type of arrangement, a shaping function can be applied to adjust each group of model parameters (e.g., previous curvature, torsion, tilt angle, etc.) to estimate a new potential catheter shape. In this type of arrangement, the model parameters can form hidden state variables, and in an embodiment, an extended Kalman filter or other estimator can be used to estimate these hidden state variables to predict the catheter shape. In this type of arrangement, the state distribution of all possible catheter shapes can be generated, and it is transformed from the catheter reference system to the patient reference system to predict the electrode location in the patient reference system. The predicted electrode measurement associated with the predicted location of the electrode in the patient reference system (e.g., from an impedance model) can be used together with the actual electrode measurement in the patient reference system to update a set of shape parameters associated with the updated shape estimation. This can allow the real catheter shape and electrode location in the patient reference system to be identified.

[0143] When estimating or predicting the shape of a catheter described by a small number of parameters, it has been recognized that the shape estimate is overdetermined. That is, the distribution of states for the predicted catheter shape based on the shape parameters may include shapes that are possible but unlikely. For example, a catheter may be straightened by setting all curvatures to zero. Figure 6B When the tube curvature is set to a larger value, Figures 8A-8DThe planar catheter is rolled into a tight loop. Both scenarios are unlikely. Furthermore, the electrode measurements all contain some error such that no combination of shape parameters exists that accurately reproduces the predicted or observed electrode measurements. Therefore, it would be beneficial to eliminate unlikely states from the estimate to improve the overall accuracy of the process.

[0144] In an embodiment, the present disclosure describes a technique for pruning parameter space to physically achievable states by determining the likelihood of a set of shape parameters. More specifically, a likelihood function is applied to the estimated state distribution of the catheter shape model to exclude unlikely states from the estimated state distribution. This results in biasing the estimator toward more likely parameters.

[0145] In an embodiment, the likelihood function can be experimentally determined by deforming a physical catheter associated with the catheter model under a constant force and calculating the energy associated with a set of shape parameters for each shape of the catheter. The stored energy can then be used in proportion to the negative log-likelihood of a set of associated shape parameters for a particular catheter shape. Along these lines, it has been recognized that many catheters have one or more shape memory wires or splines that attempt to return to a nominal or original configuration when deformed. For example, the planar catheter discussed above can return to a planar configuration once the deforming force is removed from the catheter. Therefore, it can be assumed that the energy stored in the catheter when bent is proportional to the likelihood of the deformation. When the catheter is deformed by pushing the catheter against a structure, it can be assumed that the catheter will adopt the configuration with the lowest energy. For example, for deformation in response to an obstacle, if a lower energy configuration can produce the same measurement, the catheter will be in a lower energy configuration. Therefore, it is concluded that the energy of the deformation is proportional to the likelihood of the corresponding set of shape parameters.

[0146] Fig. 9 A test system 200 is shown for experimentally determining the deformation energy of a catheter 140 during a bending procedure. As shown, the system has a support or clamp 202 that receives and holds the shaft 142 of the catheter 140 in a known orientation, a movable slide 210 and an actuator 212 that displaces the distal end of the catheter, a force sensor 220, one or more cameras 230, and a controller 232. The system is used to apply a deformation of known magnitude and / or displacement and record the positioning of the catheter electrode 146 in 3D space.

[0147] The chuck 202 holds the catheter shaft at a desired roll angle α. In an embodiment, the chuck is configured to rotate around an axis substantially coaxial with the longitudinal axis of the catheter shaft. Therefore, the chuck can be rotated to any desired roll angle α. During each bending procedure, the catheter shaft can be maintained at a predetermined fixed roll angle. The movable slide 210 moves at an angle θ (e.g., a contact angle) relative to the longitudinal axis of the catheter 140 to contact the distal end of the catheter. The slide 210 is attached to the actuator 212 via a force sensor 220. Then, the slide 210 is controllably displaced by the actuator 212, which is configured to maintain a force setting value (e.g., via PID control). The slide 210 moves toward the catheter until it contacts the distal end of the catheter at a known angle θ and reaches the force setting value. Once the initial force setting value is reached, the slide can be additionally displaced for the additional force setting value. For each advancement or displacement of the slide and the force setting value, the catheter will bend or deform. By recording the displacement and force, they can be integrated to calculate the energy stored in the catheter. A similar procedure may be provided in which the distal end of a catheter is displaced (eg, pushed) into a lumen of known size and orientation.

[0148] The bending procedure is performed in a calibrated multi-camera system. That is, the camera 230 identifies the position of each electrode 146 so that the low error 3D coordinates of each electrode are determined for each deformation. For each deformation, the controller 232 uses the coordinates from the camera, the angle information from the chuck and the slide, the force and displacement to determine the corresponding shape parameters (e.g., curvature, tilt angle, etc.). In an embodiment, the nonlinear least squares minimization of shape, position and orientation parameters is used to find the shape parameters associated with a specific deformation. By iterating a constant force displacement on the arrangement of α, θ and force, a sample of shape / energy profile is obtained. Then the curve describing the energy as a function of the shape parameters can be fitted to the sample. Fig.10 Examples are shown of the curvature of the proximal and distal planar segments of catheter 140. Many curves may be generated for any given catheter.

[0149] In an embodiment, the experimentally determined curve is the basis of the likelihood function r(x). The likelihood function is used to regularize the estimated state distribution. Typically, the likelihood function describes the rationality of a state (e.g., catheter shape). In an embodiment, a negative log likelihood is utilized. In such embodiments, an impossible state has an infinite negative log likelihood, while the most likely state has the smallest negative log likelihood. In order to apply this regularization, in an embodiment, a probability density function (regularized PDF) is calculated by negating, exponentiating, and normalizing the negative logarithmic function. The estimated state distribution is then multiplied by the regularized PDF and renormalized to create a regularized state distribution that ignores unlikely states (i.e., states outside the combination of the state distribution and the regularized PDF).

[0150] In an embodiment, the log-likelihood function may be approximately applied by a second-order Taylor series expansion of the negative log-likelihood function at the mean of the estimated state distribution to create a probability density function. In an embodiment, the approximation of the negative log-likelihood function may be performed via the following equation:

[0151]

[0152] The Hessian of the second-order expansion Viewed as the inverse of the covariance, the Gaussian mean is given by the Jacobian of the second-order expansion multiplied by the inverse of the Hessian. This approximation is equivalent to a Gaussian PDF, which can be multiplied by the state distribution by well-known methods.

[0153] The regularization of the state distribution estimate is Figures 7A-7C Specifically, Fig. 7A A state distribution 100 of possible catheter shapes predicted by the catheter model is shown. Figure 7B The regularized PDF 104 applied to the state distribution is shown. Figure 7C A regularized state distribution 106 is shown, which is generally surrounded by a dashed circle for illustrative purposes. It should be understood that the regularized state distribution excludes unlikely states from the initial state distribution estimate. This results in a new or updated state distribution (e.g., a regularized state distribution) with an updated mean and an updated covariance. In other words, the regularization process results in a tighter state distribution that more accurately predicts the true state of the system.

[0154] Catheter Position and Orientation Model

[0155] For any catheter model with magnetic sensors, the magnetic sensors will typically define vectors with six degrees of freedom. Three degrees of freedom for position (i.e., x, y, z), which can define the origin model in the catheter reference frame defined by the model's magnetic sensor or sensors, and three degrees of freedom for orientation (i.e., yaw, pitch, roll). The three degrees of freedom for orientation can define a 3D bivector (b yz , b zx and b xy), the 3D double vector is the logarithm of the quaternion. The catheter shape model can be transformed into the patient reference system using a transformation (e.g., a catheter transformation) that retains the shape and size of the catheter model. That is, the catheter position and orientation model can be represented by a rigid body transformation (e.g., a six-degree-of-freedom rigid body transformation) that converts the vector (e.g., state vector) of the shape model into the patent reference system. For example, such a transformation can align the origin and orientation (e.g., vector in an embodiment) of the catheter model relative to the origin of the patient reference system (e.g., as determined by the patient reference sensor). At this point, the positions of the magnetic sensor and the electrode are known or estimated in the patient reference system. It is noteworthy that the origin of the patient reference system and the origin of the catheter reference system may be offset due to the patient's movement (e.g., breathing, the patient's body movement, etc.). Therefore, the transformation and registration between the patent reference system and the catheter reference system can be updated.

[0156] PRS model and magnetic model

[0157] In an embodiment, a magnetic patient reference sensor model or PRS model (e.g., PRStoPat) is used to describe the displacement of the patient reference sensor relative to the patient reference frame 6. As will be appreciated, magnetic measurements are generated in a static magnetic-based coordinate system or magnetic reference system 4, which typically has an origin based on a magnetic field generator. In order to eliminate patient motion relative to the static magnetic reference frame, the patient's motion can be monitored by a magnetic position patient reference sensor (PRS) 26. See Figure 5A . Using this method, the magnetic measurements from the medical device will be transformed into the patient reference frame. This method is prone to changing the displacement of the position reference sensor relative to the patient's body. These displacements may be caused by patient skin movement due to breathing or cardiac movement, body stretching, patient talking, etc. The uncompensated displacement of PRS26 leads to offsets in magnetic measurements and increased inaccuracy in the positioning of the medical device. In an embodiment, the present disclosure describes a PRS model that takes into account the changes in magnetic positioning due to the displacement of PRS 26. The magnetic PRS model can be used to describe the displacement of the PRS with the help of a hidden state vector. That is, the model is defined as a random process that determines the true state of the model (i.e., the hidden state or potential state).

[0158] The PRS model includes the position and orientation of one or more position reference sensors (e.g., patient reference sensors) defined in a patient reference frame, wherein each patient reference sensor further has 6 degrees of freedom (e.g., 3D position and 3D orientation), which are also expressed by state variables of the sensor fusion algorithm. The model allows tracking of the motion of the patient reference sensors relative to the patient. The motion can be modeled as having small magnitude errors. The consistent placement of the patient reference sensors on the patient allows at least one of them to be considered to be at a specific initial translation offset from the origin of the patient coordinate system. For example, the patient origin can be placed at or within a typical heart position. Alternatively, the patient origin can be defined by the patient reference sensors.

[0159] The magnetic model (PatToMag) defines a rigid transformation from a patient coordinate system (e.g., a patient reference system) to a magnetic field generator coordinate system, which can change over time. In an embodiment, the magnetic model defines an initial transformation between a location (e.g., an origin) in a patient reference system and an origin of a magnetic reference system. In an embodiment, the magnetic model allows the magnetic values ​​of the location in the patient reference system to be predicted. The magnetic model has 6 degrees of freedom expressed by the state variables of a sensor fusion process or algorithm. The magnetic model allows the patient's movement to be tracked substantially in real time. At the beginning of the procedure, the patient typically lies on an operating table with a consistent orientation that is at or near a known orientation relative to the magnetic field generator. Therefore, for the initial state, a fixed orientation of the magnetic field generator (e.g., a magnetic reference system 4) relative to the patient reference system can be assumed.

[0160] The purpose of the model is to determine at any time k (i.e. PatToMag k ) is a time-varying three-dimensional patient-to-magnetic model / transformation. At any given time, the transformation provides a rigid transformation from the patient reference frame to the magnetic reference frame (e.g., the field generator coordinate system). The time-varying transformation allows locations in the patient reference frame to be transformed into the magnetic reference frame and these locations to be updated periodically or continuously in response to the patient's movement. Broadly speaking, the model allows one reference frame to be represented in another reference frame so that a point in one frame (e.g., the patient reference frame) at a given time can be identified in another coordinate system (the magnetic reference frame) at that time. In this regard, by applying a transformation matrix M, a first reference frame can be represented in terms of a second reference frame. That is, a set of resulting coordinates (e.g., position and orientation) in one reference frame is a function of the coordinates (e.g., position and orientation) in another reference frame. In the generalized equation:

[0161]

[0162] where the column vector [x, y, z] is the position in the patient reference frame and the matrix M represents PatToMag kThe matrix, column vector [x', y', z'] is the predicted measurement of positioning in the magnetic reference frame. This relationship can be realized as a random process, so that PatToMag k The transformation / model can be used to predict magnetic measurements of the predicted location of the magnetic sensor in the patient reference frame (e.g., from a medical device model). Such predicted measurements can be updated or corrected based on subsequent observations. For example, actual magnetic measurements of the magnetic sensor at or near the predicted location can be obtained from the medical positioning system 22. The actual measurements can be used together with the predicted measurements to determine the true position of the magnetic sensor in the patient reference frame. In addition, the actual measurements and the predicted measurements can be used to update the matrix M, which predicts the value of the location in the patient reference frame.

[0163] The model uses the patient reference sensor to establish a reference position in the patient reference frame. When the patient reference sensor moves with the patient, PatToMag k The model must evolve over time to account for this type of patient motion. For example, the transformation can be performed in the following ways:

[0164]

[0165] Where the position value is further represented in the last entry of the column vector using 1 or 0 to represent the point / position or direction respectively. In embodiments where the process is a stochastic process, the variables of the matrix (eg, ai) are allowed to evolve over time.

[0166] In this example, the model has the following relationship:

[0167] Pat k =PRSToPat i,k ref k

[0168] as well as:

[0169] mag k =PatToMag k Pat k

[0170] where ref k is the coordinate 100 in the patient's body (eg, reference sensor space) at time k (eg, before alignment with the patient reference frame to account for patient motion), pat k is the coordinate value in the patient reference frame at time k (e.g., the patient coordinate system coordinate), and mag k is the time k (i.e., pat after transformation from the patient reference frame to the magnetic reference frame) k ) in the magnetic reference system. Fig.11AIn an embodiment, the transformation of the patient reference sensor to the patient can be expressed as PRSToPat i,k , where the subscript i refers to a specific patient reference sensor (e.g., i=1, i=2, i=n, ​​etc.). In the case of a single patient reference sensor, the subscript may be omitted. These relationships provide a method for predicting magnetic measurements or values ​​for any coordinate or location in the patient reference frame.

[0171] PRSToPat i,k The transformation will be in ref k The patient reference sensor 26 at PRSToPat is aligned (eg, rotated) with the patient reference system 6. That is, PRSToPat i,k The transformation represents a patient reference sensor transformation between the patient reference sensor and the patient reference system. In an embodiment, the patient reference sensor may define the origin of the patient reference system. In another embodiment, the patient reference sensor may be offset from the origin of the patient reference system. In either embodiment, the initial orientation of the patient is known. For example, at the beginning of a procedure, the patient typically lies on an operating table in a consistent orientation so that the orientation of the patient reference system is known. The patient reference system 6 may extend, for example, from the head to the feet of the patient initially aligned with the support / operating table, from left to right, and vertically up and down. The patient reference sensor 26 has six degrees of freedom (e.g., x, y, z, roll, pitch, and yaw). When the patient reference sensor is applied to the patient externally, the initial orientation 5 (e.g., roll, pitch, yaw) of the sensor 26 may not be aligned with the patient reference system 6. However, the orientation of the patient reference sensor 26 may be determined (e.g., using a medical positioning system) and the transformation applied to the patient to align the patient reference sensor 26 with the patient reference system 6 that defines PRSToPat i,k The patient reference system 6 is aligned with the transformed patient reference system 6. Because both the patient reference sensor 26 and the origin (and other coordinates within the patient) move with patient motion, it is assumed that the relationship between the patient reference sensor 26 and the origin or any coordinate within the patient remains substantially constant. Therefore, the nominal patient reference sensor to patient transformation (e.g., NomPRStoPat) is a definable and fixed transformation.

[0172] PatToMag k The transformation aligns the patient reference frame with the magnetic reference frame4. In other words, PatToMag k The transformation PatToMag defines a transformation between a magnetic reference frame having an origin associated with a magnetic field generator of a magnetic-based medical positioning system and coordinates or locations in a patient reference frame. kIt may initially be based on a fixed orientation of the magnetic field generator (eg, magnetic reference frame 4) relative to a known initial orientation of the patient reference frame. In other words, the nominal patient-to-magnetic transformation (eg, NomPatToMag) is a definable and fixed transformation.

[0173] Based on these relationships and transformations, PatToMag k Transformation in a magnetic reference (such as mag k ) system to determine any coordinate in the patient reference system (e.g., pat k )’s position and orientation. In this regard, the relationship between mag k =PatToMag k Pat k Forming part of the observation model, as discussed further herein. In an embodiment, the position of the patient reference sensor can be monitored to identify displacements caused by patient motion. That is, the medical positioning system can identify displacements of the patient reference sensor, and these displacements can be used to update the above relationship.

[0174] To implement the model, it is necessary to define the transformation between coordinate systems (e.g., patient coordinate system and magnetic coordinate system). The relationship from the patient reference sensor space to the magnetic field generator (e.g., magnetic coordinate system) at any time k is:

[0175] PRStoMag i,k =PatToMag k PRSToPat i,k

[0176] That is, the transformation from the patient reference sensor to the magnetic field generator is PatToMag k and PRStoPat i,k The product of the two transformations. Since the patient reference sensor is a magnetic sensor, its position in the reference magnetic field can also be directly observed. However, due to the patient's movement, PatToMag k and PRStoPat i,k Both may change or evolve over time, although the two transformations should remain in their nominal relationship NomPRSToPat i (also defined as PRStoPAT i,nom ) and NomPatToMag (also defined as PatToMag nom ). In an embodiment, the nominal relationship is the initial relationship (eg, when k=0). PatToMag k and PRStoPat i,kis defined as tracking deviations from their nominal relationship and capturing these changes so that the transformation between these reference frames is updated in conjunction with the patient's motion (e.g., motion of the patient reference sensor).

[0177] In an embodiment, PRStoPat i,k It can be defined as tracking deviation as follows:

[0178] PRStoPat i,k =NomPRSToPat i PRSToNomPRS i,k

[0179] This is Fig. 11B . As shown, the patient reference sensor 26 is moved from a nominal (e.g., initial, in an embodiment) offset position 26A as may be recognized by a medical positioning system. Similarly, other coordinates (e.g., 100) are moved from the initial position to an offset position 100A. In the embodiment shown, the fixed transformation between the nominal (e.g., initial) patient reference sensor position and orientation and the patient reference system 6 is labeled as the nominal patient reference transformation NomPRSToPat i In addition, the time-varying transformation PRSToNomPRS i,k The movement between the initial sensor position 26 and the subsequent / current sensor position 26A is defined. Fig. 11B That is, PRSToNomPRS i,k is the patient reference sensor displacement transform. The displacement can be determined based at least in part on monitoring the movement of the patient reference sensor using a magnetic field generator. In this embodiment, PRStoPat i,k is the product of the nominal transformation and the patient reference sensor displacement transformation.

[0180] In an embodiment, PatToMag k It can be defined as tracking deviation as follows:

[0181] PatToMag k =NomPatToMag PatToNomPat k

[0182] This is Fig. 11C As shown, the coordinates 100 move from a nominal or initial position to a displaced position 100 in response to patient motion. In the illustrated embodiment, the transformation between the patient reference system and the magnetic reference system is labeled the initial or nominal magnetic transformation NomPatToMag and is a fixed transformation. Additionally, the time-varying transformation PatToNomPat between the initial coordinate position 100 and the subsequent / current position 100A is kdefines the displacement of the coordinates in the patient reference frame. That is, PatToNomPat k is a coordinate displacement transform. The displacement can be determined at least in part based on monitoring the movement of the patient reference sensor and thus the patient reference frame. In this embodiment, PatToMag k is the product of the nominal transformation and the coordinate shift transformation. In an embodiment, PRSToNomPRS i,k and PatToNomPat k is controlled by state variables that can be determined during the sensor fusion process. In an embodiment, these variables are determined in an estimation system such as a recursive Bayesian estimator (e.g., an extended Kalman filter or a particle filter) to adapt the magnetic and catheter state variables for magnetic and other measurements. In summary, these transforms PRSToNomPRS i,k and PatToNomPat k is not directly observable and is subject to constant variation due to, for example, patient breathing and other patient motion, and is allowed to evolve over time. However, in combination with observable parameters (e.g., PRSToMag i,k ), these transformations can be estimated.

[0183] As previously mentioned, the patient typically lies on the operating table in a consistent orientation at the start of the procedure. Therefore, for the initial state, a fixed orientation of the magnetic field generator (e.g., magnetic reference system 4) relative to the patient reference system can be assumed. Therefore, in an embodiment, the nominal relationship (e.g., transformation, sensor positioning, etc.) is assumed to be substantially equal to the initial relationship at the start of the procedure. In an embodiment:

[0184] PRSToNomPRS i,nom =I≈PRStoNomPRS i,0

[0185] PatToNomPat nom =I≈PatToNomPat0

[0186] That is, the desired nominal configuration is close to the initial configuration I, which is substantially equal to the nominal configuration at time 0. Therefore, the initial patient-reference sensor positioning is for PRSToMag 0,nom are appropriate values ​​because they are expected to be close to PRSToMag 0,0 .

[0187] In an embodiment, a patient reference sensor may be measured (that is, observed) in a magnetic field:

[0188]

[0189] Where P nomis the observed orientation of the patient-reference sensor at a nominal time (e.g., time 0 or a combination of multiple observations near time 0), and p nom is the position of the patient reference sensor at the nominal time. Therefore, PRSToMag can be determined from direct measurements 0,nom Transformation. Based on PRSToMag 0,nom Transformation, the following transformation can be derived:

[0190]

[0191]

[0192] in:

[0193] A nom is the fixed nominal rotation between the patient coordinate system and the magnetic coordinate system, which is assumed to be fixed;

[0194] a nom is the patient's position relative to the field generator (e.g., position on a bed for a magnetic field-based positioning system), which can be estimated as follows;

[0195] B nom is the orientation of the patient-reference sensor on the patient, which can be estimated as follows; and

[0196] b nom is the position of the patient reference sensor on the patient, which is known.

[0197] In an embodiment, it is desirable to set a fixed position (e.g., 0 or origin) of the patient reference sensor in the patient coordinate system so that the patient origin is generally inside the heart. In an exemplary embodiment, for the posterior patient reference sensor, the offset between the patient reference sensor and the origin inside the heart is selected to be b nom =[0 175 0] T , which is a column vector denoted by superscript T. Other offsets are possible and generally depend on the system in which they are implemented and / or the placement of the patient reference sensor on the patient (eg, chest, back, etc.).

[0198] When the patient is placed on the table in a characteristic orientation, the orientation between the field generator, the table, and thus the patient reference frame may be initially known. In an embodiment, a fixed nominal rotation may be established between the patient reference frame and the magnetic reference frame. For example:

[0199]

[0200] This fixed nominal rotation is usually determined by the configuration of the specific magnetic field based positioning system. The remaining degrees of freedom are a nom and Bnom . Given the previous relationship:

[0201] PRStoMag k =PatToMag k PRSToPat k

[0202] but:

[0203]

[0204] P nom =A nom B nom

[0205] p nom =A nom b nom +a nom

[0206] therefore:

[0207] B nom =A T nom P nom

[0208] a nom =p nom -A nom b nom .

[0209] Therefore, A can be measured and / or derived nom 、a nom , B nom and b nom Each of these allows estimation of PatToMag nom and PRSToPat 0,nom In other words, there are enough observable parameters to allow estimation of the variable transformation.

[0210] The model defines a time-varying rigid transformation from the patient coordinate system to the magnetic field generator coordinate system to allow determination of magnetic values ​​located in the patient reference system. In an embodiment, the PRS model is used in conjunction with the catheter model and the position and orientation model. Fig. 12A As shown in FIG. 1 , the catheter model 42 (e.g., Fig. 6A The model 42 is used to transform the catheter 24 of the patient into the patient reference frame 6. That is, the model 42 is used to predict the positioning of the sensor and / or electrode in the patient's heart 52, as shown in the figure. Fig. 12AAlso shown is a physical catheter 50 disposed within the heart. The catheter model 42 corresponds to the physical catheter 50 disposed within the patient's heart 52. Initially, the location of the model magnetic sensor 281 of the catheter model is predicted in the patient reference system 6 using the position and orientation model. That is, the catheter model is transformed from the catheter reference system to the patient reference system to place the model magnetic sensor 281 at a coordinate within the heart, for example. The position may initially represent the coordinate pat k At this point, as mentioned above, you can use PatToMag k The transformation determines (e.g., predicts) magnetic measurements / values ​​of the model magnetic sensor 281 in the magnetic reference frame. Accordingly, magnetic measurements / values ​​can be measured in the magnetic reference frame for the corresponding physical magnetic sensor 282 of the physical catheter 50. For example, the medical positioning system can obtain actual measurements of the magnetic sensor 282 for the physical catheter (e.g., using a certain amount of system noise). Predicted measurements and actual measurements can be used to update multiple models in a random process. That is, predicted measurements and actual measurements (or if there are multiple sensors, multiple measurements can be performed) can be used to adjust the position and orientation of the catheter model 42 in the patient reference frame and / or generate the positioning of the physical magnetic sensor. In an embodiment, the pat k After multiple iterations, the position and orientation of the catheter model 42 projected into the patient reference frame may be updated to more closely approximate the position and orientation of the physical catheter 50 disposed within the patient's heart. Fig. 12B In an embodiment, the PRS model allows the positioning of the catheter model within the patient reference frame to be nearly continuously updated to account for patient motion. That is, rather than aligning the catheter model based on a static transformation, the development of the PRStoPat and PatToMag transformations provides for continuous updating of the model magnetic sensor positioning of the catheter model, even in the presence of patient motion, such that its predicted positioning more closely represents the physical positioning of the physical magnetic sensor.

[0211] In an embodiment, an extended Kalman filter is used to infer hidden state variables corresponding to the transformed hidden state variables of the model. Based on the hidden state variables, hidden state measurements (e.g., magnetic values ​​in the patient reference frame) can be predicted at any time, and the extended Kalman filter (or other estimator) framework can be used to update the estimates of the state variables in a manner that allows those portions of the accessible hidden state variables to be updated. Therefore, at any moment, although there may not be enough information to determine portions of the state variables, by using the extended Kalman filter framework, predictions associated with the appropriate portions of the state variables (associated with the transformation from the impedance-based domain to the patient domain) can be made.

[0212] A difference can be made between a prediction and actual measurement of an appropriate portion of the state variable associated with the model, and the appropriate portion of the state variable can be updated based on the difference between the prediction and actual measurement. In this way, the state variable can be modified over a given period of time rather than at a given moment in time. For example, a previous prediction of an appropriate portion of the state variable can be corrected based on a measurement at the current point in time.

[0213] As further discussed herein, the magnetic and / or PRS models may form part of an overall or composite system model. During implementation, the model is queried to predict the positioning of the sensor in the patient reference frame. These predictions are then used together with the sensor positioning measurements to further improve the estimated positioning of the sensor in the patient reference frame and update the magnetic model and / or PRS model. It will be understood that additional magnetic models are possible and are considered to be within the scope of the present disclosure. In an example, a magnetic model for transforming between patient relative coordinates and magnetic coordinates is described in U.S. Provisional Application No. 62 / 756,936, entitled “Patient Reference Sensor Model for Medical Device Localization based on Magnetic and Impedance Sensor Measurements,” filed on November 7, 2018, the entire contents of which are incorporated herein by reference.

[0214] Impedance Model

[0215] Within the context of the sensor fusion process, the usefulness of impedance measurements to locate the physical catheter and its electrodes in three-dimensional space (e.g., a patient reference frame) depends on having an effective model for any catheter configuration to predict the impedance measurements within the potential field or impedance potential field. That is, based on the predicted positioning (e.g., model positioning) of the catheter and / or catheter electrodes (e.g., model electrodes) in the patient reference frame (e.g., three-dimensional space), it is desirable to predict the impedance measurements of the modeled catheter electrodes to improve the positioning of the physical catheter and / or its electrodes and / or update the impedance model. It has further been recognized that, in some cases, previous efforts to model impedance for electrode positioning have lacked accuracy due to a failure to account for noise.

[0216] In an embodiment, the impedance model transforms between a patient coordinate system and an impedance measurement (e.g., PatToImp). In addition, the impedance model can incorporate noise and / or distance-related modeling errors between individual electrodes to improve estimates of impedance measurements for determining electrode locations within a patient. In an embodiment, the impedance model is a stochastic process in which the true state of the model is a determined hidden or latent state. The model generates estimates of electrode impedance measurements in three-dimensional space (e.g., location-impedance values). Such estimates and / or models can be improved based on actual impedance measurements of electrodes located in three-dimensional space. In an embodiment, such impedance models implement a variety of separate methods that can be used in combination.

[0217] In an embodiment, a first methodology involves modeling the positioning-impedance values ​​as a linear combination of mapped harmonic basis functions (such as regular solid harmonics). However, it will be understood that additional harmonic basis functions are possible and are considered to be within the scope of the present disclosure. However, it is believed that regular solid harmonic basis functions can provide appropriate descriptiveness with reduced degrees of freedom, thereby simplifying the model. In addition, since the electrodes are located in a common blood pool, they are generally subject to uniform conditions so that the Laplace operator of the harmonic basis function should be zero, thereby providing constraints for the model. Further, the linear combination of the harmonic basis functions can be constrained to obey Kirchhoff's voltage law. In general, this helps to solve the spatial nonlinearity problem of impedance measurement. In an embodiment, a second methodology involves modeling the measurement noise characteristics of the impedance system, including the covariance between measurements from different electrodes. In an embodiment, a third methodology involves introducing an artificial measurement noise covariance between different electrodes, which decreases with the distance between those electrodes. This noise term represents the amount of error that is not modeled in other cases, and helps to solve the spatial nonlinearity of impedance measurement and / or breathing-related artifacts.

[0218] The hardware for impedance-based localization measurements consists of a set of electrical patches (e.g., 6 patches: neck, leg, chest, back, right side, left side) fixed to the patient / patient reference frame. Figures 3A-3D and 13A. AC voltage is applied to multiple groups of patch pairs (e.g., back->left, left->chest, right->back, chest->right, neck->back, leg->back), and the potential (e.g., impedance) on each catheter electrode 30 of the catheter 24 placed in the resulting impedance field is measured while driving each patch pair. See Fig.13AThe measured potential depends on the relative impedance between the electrode and each driven patch. That is, each driven patch pair induces a potential field across the patient reference space measured by the electrodes. Therefore, the purpose of the impedance model is to model the potential field and its measurement characteristics so that the impedance measurement can be estimated for any location within the potential field. For example, after establishing such an impedance model of a three-dimensional space (e.g., the patient reference space), the impedance value can be estimated or predicted for any location in that space at a given time.

[0219] Fig.13A A mathematical graph of impedance patches is shown, where each patch forms a vertex of the graph, and graph edges (e.g., solid connecting lines) extend between the vertices of each driven patch pair. In a mathematical graph, a cycle is a path of edges and vertices where a vertex is reachable from itself. That is, the cycle forms a closed loop. In this embodiment, a set of driven patch pairs defines a single cycle: back->left->chest->right->back. Each cycle (i.e., one in this embodiment) of the graph that forms a closed loop or circuit (e.g., a cycle) is subject to Kirchhoff's voltage law, which means that the potential difference around the cycle must sum to zero. That is, Z B-L +Z L-C +Z C-R +Z R-B = 0. See Fig.14 Accordingly, the sum of the driving potentials on any electrode in this loop must be zero, because the potential drop around the circuit back->electrode->left->electrode->chest->electrode->right->electrode->back must also be zero.

[0220] Based on these constraints, the number of independent impedance potential fields is the number of driven patch pairs (i.e., six in this embodiment) minus the number of loops in the diagram (i.e., one in this embodiment). In this embodiment, there are five independent impedance potential fields. In addition, there is a linear mapping from these independent impedance potential fields to the larger set of potentials driven by the patch pairs. For the set of driven patch pairs shown in this embodiment, the mapping is as follows:

[0221]

[0222] Where z is a vector of measured or estimated potentials and y is a vector of independent impedance potentials, which in embodiments are hidden state variables of the random process. The digital matrix denoted below by M maps the five independent impedance potential fields y to six potentials z. In embodiments, the matrix M forms an observation model for the random process. The independent impedance potential fields represent virtual potential fields that, in the proposed embodiment, are never excited despite being present in the system. That is, there are independent impedance potential fields between non-existent patch pairs. Two of these independent impedance potential fields y左侧-右侧 and 背部-胸部 exist Fig.13A For example, y 左侧-右侧 =z 左侧-胸部 +z 胸部-右侧 and -y 左侧-右侧 =z 右侧-背部 +z 背部-左侧 .y xy The independent impedance potential field is Fig. 13B The potential field is based on four patches located in a roughly common plane XY, namely the back, left side, chest, and right side, as shown in FIG. Fig.13A As shown in . A similar diagram can be provided for the YZ plane of the patient to describe the remaining independent impedance potential fields. Therefore, the independent impedance potential fields can be calculated or estimated as algebraic functions of the measured impedance values. In an embodiment, the independent impedance potential fields define the state variables of the random process and are described using harmonic basis functions as described below.

[0223] For each time point, and for each independent driven patch pair or "impedance module (mod)", the model impedance measurement for each independent impedance potential field i and electrode j is as follows:

[0224]

[0225] as well as:

[0226]

[0227] in is a series of harmonic bases Y l The calculated potential of electrode j in the independent impedance potential field i, ε ij is a modeling error term that covaries between a pair of electrodes according to their distance, and ν k is the measurement noise term.

[0228] For each independent driver patch pair or impedance module i', is the electrode position x j (e.g., in the patient reference frame) and a function of the state variables of the impedance model (e.g., the patient-to-impedance transformation). In an embodiment, the impedance transformation may be a global dynamic non-rigid transformation that maps the patient reference frame to the impedance reference frame. i'is the vector of all distance-dependent modeled error terms for impedance module i' and is modeled as a multivariate normal random variable whose terms have covariances that depend on the distance between electrode pairs. Finally, ν (the vector of electrical noise terms for all electrodes) is a multivariate normal random variable that reflects the electrical noise characteristics of the measurement system. This impedance measurement behavior model is used as part of a sensor fusion process or algorithm, such as a recursive Bayesian estimator (e.g., an extended Kalman filter or a particle filter) to adapt the impedance and catheter state variables for impedance measurements and other measurements.

[0229] In an embodiment, is a linear combination of basis functions. Each basis function at a point in space maps the electrical value of the modeled potential field to an electrical value (e.g., voltage, impedance, etc.). If the electrode is at position x j ,but:

[0230]

[0231] where Y l is a scalar-valued function of the l'th solid harmonic basis function evaluating the electrode positioning, and b i,l is the weight of the basis function (e.g., the l'th basis function) that relates the patient reference frame to the impedance potential field. l should be harmonics. That is, the Laplace operator should be zero everywhere. In the embodiment, Y l It can be regular solid harmonics up to a predetermined order. For example, Y l can be regular solid harmonics up to the fourth order. Using harmonics up to the fourth order results in 25 basis functions per electrode. It will be appreciated that limiting the harmonic basis functions to the fourth order truncates information in higher order harmonics that may provide additional description of the potential field. Exclusion of this information is accounted for in the modeling error term ε. In an embodiment, the weight b i,l These weights b may be set to predetermined or default values, which may be based on an experimentally determined baseline. During operation of the impedance model, these weights b i,l To make the impedance and catheter state variables suitable for impedance measurements and other measurements.

[0232] The distribution of the modeling error term ε:

[0233] The modeled error terms are intended to represent sources of unmodeled signals in the impedance measurement, such as unmodeled higher-order terms of the harmonic basis or perturbations caused by the patient's breathing. Explicitly incorporating the expected magnitude of the unmodeled phenomena or errors in the impedance basis model makes the system robust to such differences.

[0234] For each independent impedance module i', let ε j =n j +r j , where nj It means that due to not being g k The error caused by the nonlinearity of the modeled impedance is a multivariate normal random variable, and r j is a multivariate normal random variable representing the error due to, for example, unmodeled respiratory artifacts.

[0235] In the embodiment, n j :

[0236]

[0237] Where N is a normal random distribution, 0 represents the mean of the distribution, and the matrix h represents the covariance of each electrode based on the absolute distance d (e.g., vector) between each electrode (i.e., |x#-x#|). See, e.g. Fig. 6A . This choice of covariance causes the error to vary smoothly across the space. Parameter a represents the size of the modeling error. The larger the parameter a, the larger the expected size of the nonlinear derivative error. Parameter b represents the width or radius over which the modeling error decays. The larger the parameter b, the smaller its expected distance scaling.

[0238] In the embodiment, r j :

[0239]

[0240] At each time i, respiration applies a shared transformation to all points of each independent impedance potential field. Parameter c controls the magnitude of the respiration error. In an example, a system and method for modeling respiration errors or artifacts is described in U.S. Provisional Application No. 62 / 756,926, entitled “Respiration Model for Device Localization Based on Impedance Sensor Measurements,” filed on November 7, 2018, the entire contents of which are incorporated herein by reference.

[0241] Electrical noise term ν k Distribution:

[0242] For each driver patch pair j:

[0243]

[0244] In this embodiment, each electrode has a noise variance component e that is independent of other electrodes and a noise variance component d that is shared with other electrodes.

[0245] In use, the impedance model mathematically defines an impedance field of the patient reference frame. In an embodiment, the impedance field is defined so that the impedance potential of a set of driven patch electrodes at any location in the impedance field is a function of a set of independent impedance potential fields mapped to the impedance potential. A variety of methods can be embedded to generate the impedance model. In an embodiment, the impedance model is dynamically generated during the procedure. That is, the catheter 24 can be placed within the impedance field at the beginning of the medical procedure. See Fig.13A . During this process, as the catheter 24 moves through the impedance field, electrode impedance measurements are obtained, and independent impedance potential fields are mapped to the impedance measurements. In another embodiment, a mapping catheter having a large number of electrodes can be used to generate an initial impedance model. In such embodiments, the mapping catheter can be moved (e.g., scanned) around an internal patient cavity (e.g., a ventricle) to obtain impedance measurements for a large number of locations. This can produce an impedance model with a large number of locations and corresponding impedance measurements and independent impedance potential fields for mapping. In such embodiments, the mapping catheter can be removed after the initial impedance model is generated and replaced with a catheter (e.g., an ablation catheter) for performing a medical procedure. An exemplary mapping catheter is described in U.S. Patent No. 8,744,599, entitled "High Density Mapping Catheter", which is incorporated herein by reference in its entirety. In another embodiment, a default impedance map can be provided that assumes an initial state of the impedance model. In this embodiment, subsequent impedance measurements are used to update the model more quickly for the current impedance field or patient reference space. In any embodiment, positioning, impedance, and independent impedance potential fields can be recorded. The impedance model can then utilize this information, for example, to interpolate impedance measurements for locations between known locations in the modeled impedance field.

[0246] In an embodiment, the independent impedance potential fields and their parameters are state variables of a random process, so that they can evolve over time. Initially, the impedance model estimates the j A set of impedance measurements Z of electrode j at i,est (e.g., back->left, left->chest, right->back, chest->right, neck->back, legs->back). See e.g. Figure 5B Predicted electrode locations 301-304 in the heart 52 and Figure 5C In an embodiment, the predicted positioning of these electrodes can be estimated in the patient reference frame using a catheter model. ... i,l When the impedance model is implemented, a set of impedance measurements (eg, actual measurements) Z for corresponding physical electrodes in the patient reference frame may be obtained. i,act . See e.g. Figure 5C The predicted measurements Z i,est Compared with the actual measurement Z i,act (e.g., observable parameters / electrode impedance measurements, calculated independent impedance fields) are used together to determine, for example, a correction or gain in a random process. This correction / gain can then be used to adjust the hidden state variables of the impedance model. For example, the weight b can be adjusted i,l and other hidden state variables of the impedance model (e.g., basis functions of the independent impedance potential fields) to better fit or match the actual impedance measurement. In application, the impedance model is adjusted (e.g., recursively) to more closely approximate the actual impedance measurement. Thus, the updated impedance model can then be used to predict impedance measurements at locations in the modeled potential field (e.g., as predicted by the catheter model). In addition, corrections / gains can be used to estimate updated electrode locations in the patient reference frame. See, e.g., Figure 5C The true positioning in 371-374.

[0247] In an embodiment, an extended Kalman filter is used to infer hidden state variables corresponding to the hidden state variables of the model. From the hidden state variables, hidden state measurements (e.g., impedance values ​​at spatial locations) can be predicted at any time, and the extended Kalman filter framework can be used to update estimates of the state variables in a manner that allows updates to those portions of the hidden state variables that are accessible. Thus, at any moment in time, while there may not be enough information to determine portions of the state variables, by using the extended Kalman filter framework, predictions associated with the appropriate portions of the state variables (associated with the transformation from the location in the patient reference frame to the impedance measurement) can be made.

[0248] A difference can be made between a prediction and actual measurement of an appropriate portion of the state variable associated with the model, and the appropriate portion of the state variable can be updated based on the difference between the prediction and actual measurement. In this way, the state variable can be modified over a given period of time rather than at a given moment in time. For example, a previous prediction of an appropriate portion of the state variable can be corrected based on a measurement at the current point in time.

[0249] As further discussed herein, the impedance model forms part of an overall or composite system model. During implementation, the impedance model is queried for use with the predicted electrode positioning in the patient reference frame estimated by the catheter model. More specifically, the impedance model is used to predict measurements of each electrode positioning in the patient reference frame. These predictions are then used with the actual electrode measurements to further improve the estimated positioning of the electrodes in the patient reference frame, as well as to update the impedance model. In an example, an impedance model for determining impedance measurements or values ​​for positioning in a patient reference frame is described in U.S. Provisional Application No. 62 / 756,931, entitled “Impedance Transformation Model for Estimating Catheter Locations,” filed on November 7, 2018, the entire contents of which are incorporated herein by reference.

[0250] Breathing Model

[0251] Respiration artifacts can contribute significantly to impedance measurement errors, thereby increasing errors in the determination of the location of a medical device. In an embodiment, the present disclosure describes a respiration model that accounts for impedance changes due to respiration / breathing. In an embodiment, the respiration model describes catheter electrode impedance artifacts, which allows for increased accuracy of impedance measurements and / or predictions, thereby providing improved accuracy in determining the location of a medical device and / or its electrodes.

[0252] During cardiac medical procedures, in vivo impedance measurement errors covary significantly due to respiration. That is, respiration causes time-varying artifacts in the impedance measurement relative to spatial variations within the patient reference frame (e.g., on or within the patient's chest). Time-varying artifacts occur during each breathing cycle due to the increase and decrease in volume changes in the patient's chest. More specifically, the changes in volume change the patient's physiological state and thereby change the impedance measurement of the impedance potential field within the patient's reference frame. Therefore, it is desirable to monitor and resolve such artifacts. Because respiration varies from cycle to cycle, monitoring artifacts caused by respiration is very complex. Although respiration does move in a periodic phase space from the end of exhalation to inhalation and back to exhalation, respiration is not regularly periodic. The amplitude and duration of each breath are unique, and the time between breaths also varies. This is particularly true in the context of the patient's breathing. Fig.15, which shows a waveform 120 for a series of breaths. The waveform 120 increases during inspiration of the patient and decreases during exhalation. This waveform is not directly monitored in the proposed process and is shown by way of example. The phase angle θ of the respiratory waveform has a periodic domain that varies from π to -π. More specifically, the phase angle θ increases from zero to π during inspiration and decreases from -π to zero during exhalation. As shown, it does not have to remain constant during inspiration or exhalation. For example, the inspiration or exhalation rate can change during a single inspiration or exhalation cycle, as shown by the nonlinear slopes of the various phase angle waveforms. Expressed as ω The rate of change of the respiratory waveform is the derivative of the phase angle θ. As shown in the figure, the amplitude and duration of adjacent breaths will change. In other words, each breath is a quasi-periodic process, rather than a regular periodic process. Since breathing is quasi-periodic, the purpose of the present disclosure is to model the artifacts caused by breathing as quasi-periodic functions. In an embodiment, the artifact is modeled as a function of the current location and position in a given breathing cycle.

[0253] While each respiratory effort / breath is unique, there are some relatively stable or constant respiratory parameters that can be used to model breathing as a quasi-periodic function. For example, the functional residual capacity (FRC), which is the amount of air present in the lungs at the end of a passive exhalation, is relatively constant for a given individual. In contrast, the tidal volume (TV) is the lung volume that represents the normal amount of air expelled between a normal inspiration and exhalation. While the FRC is relatively constant, the TV (e.g., amplitude) varies more from breath to breath. As a result, the phase between the end of exhalation and the beginning of inspiration represents a relatively stable phase θs, with each breath representing a perturbation of that stable phase.

[0254] To model breathing-related artifacts in the impedance measurements, each electrode impedance measurement Z i Modeled as a quasi-periodic function g j (θ) (which in the embodiment is a time-dependent function) and the non-periodic function f i (which in the embodiment is a spatial correlation function). That is, f i The impedance measurement for a location in the patient reference frame and / or in the impedance field can be represented. The quasi-periodic function depends on the phase angle θ of the respiratory cycle. That is, the quasi-periodic function depends on the location between the beginning of inspiration and the end of expiration in the respiratory cycle, which is represented as the phase angle θ in this analysis. Since there is a common respiratory process that affects all impedance measurements in the patient reference frame, it is assumed that at time sample k, a single phase angle θ and amplitude γ determine all quasi-periodic functions.

[0255] Fig.13AA medical device or catheter 24 is shown disposed within an impedance field defined by six external or surface patch electrodes. In the non-limiting embodiment shown, the patch electrodes include a left patch electrode 56, a right patch electrode 58, a chest patch electrode 60, a back patch electrode 62, a neck patch electrode 64, and a leg patch electrode 66. For any electrode impedance measurement i of an electrode 30 on a catheter 24 that is within the impedance field by a driven patch, a quasi-periodic function g is applied. j In an embodiment, it is assumed that the same quasi-periodic function is applied to all electrode impedance measurements measured in response to driving a single pair of surface patch electrodes (e.g., impedance modules) (e.g., impedance measurements of different catheter electrodes if multiple catheter electrodes are present). Fig.13A In an embodiment where there is a single catheter electrode 30 (e.g., electrode i) in the impedance field and six surface patch electrodes defining the impedance field, driving the chest to the right patch pair (e.g., impedance module) will produce a single quasi-periodic function g j . Where j represents a quasi-periodic function. In this embodiment, this produces one quasi-periodic function (e.g., g1). That is, if current is driven between the chest and right patch electrodes, the first quasi-periodic function will control electrode 30. In an embodiment, each electrode impedance is a combined measurement of six driven patch pairs (e.g., back->left, left->chest, right->back, chest->right, neck->back, legs->back). Therefore, one quasi-periodic function is used for each driven patch pair, which results in six quasi-periodic functions. Adding a second electrode does not result in a new quasi-periodic function.

[0256] In an embodiment, the relationship between the impedance measurement and the quasi-periodic function (eg, the measurement vector) may be written as:

[0257] zi+(j-1)*numElec(x k ,θ k , γ k ) = f ij (x k )+γ k g j (θ k )

[0258] The electrode impedance measurement value Z of the electrode at time k is i+j*numEtec is a function of: the impedance of the electrode location given by the non-periodic function f(x), which can be provided by the above impedance model; and the impedance of the electrode location given by the quasi-periodic function g jThe artifact of the positioning is determined, and the quasi-periodic function is modified by the amplitude γ of the respiratory cycle at time k and the positioning θ in the respiratory cycle at time k. In an embodiment, the non-periodic function f(x) is a constant. Although each quasi-periodic function is controlled by the same phase angle, each function may lead or lag relative to other functions. Similarly, it is expected that each quasi-periodic function has a larger or smaller, positive or negative amplitude relative to other functions.

[0259] For each driven patch pair, four patches are not driven, and four additional quasi-periodic functions control the non-driven patch pairs. Fig.13A In the embodiment of FIG. 1 , driving the chest to the right patch pair (e.g., impedance module) will result in four additional quasi-periodic functions (e.g., neck->Ref patch; leg->Ref patch; back->Ref patch; and left->Ref patch, represented by g7, g8, g9, and g 10 Description). In an embodiment, each electrode impedance is a combined measurement of six driven patch pairs (e.g., back->left, left->chest, right->back, chest->right, neck->back, leg->back). Therefore, for each driven patch pair, four quasi-periodic functions are employed, which results in twenty-four additional quasi-periodic functions. In an embodiment, a non-driven patch measurement can be added to the end of the measurement vector, where the relationship between the additional non-driven patch impedance measurement and the quasi-periodic function can be written as:

[0260] z 6*numElec+j′ (x k ,θ k , γ k ) = f′ j′ (x k )+γ k g6+j′(θ k )

[0261] As mentioned earlier, when the phase angle is zero (e.g., between breaths), the breath can be considered to have a relatively stable period θ s During this stable period, the quasi-periodic function should have no effect on the measurement. Therefore, it is expected that when the phase angle is zero, all quasi-periodic functions are equal to zero. Therefore, in an embodiment, each quasi-periodic function can be formulated as:

[0262] g j (θ k )=α j (1-cos(θ k ))+β j sin(θ k )

[0263] Where α and β represent weights for adjusting the phase of each quasi-periodic function. In an embodiment, α and β are constants. In an embodiment, each quasi-periodic function and the control phase angle and amplitude vary according to a random process, and there are some normally distributed errors between samples. That is, phase and amplitude are hidden state variables that can be predicted based on previous values ​​and observable parameters (e.g., impedance measurements).

[0264] In this embodiment, it is assumed that once a breathing cycle begins, it will continue at a relatively steady rate. ω forward, except when the phase is close to zero (given by the stable phase angle θ s When it is close enough to zero, the predicted phase angle is stable and tends to zero. The phase angle has a periodic domain from π to -π. In an embodiment, the evolution of θ from the previous time step (e.g., k-1) to the current time step (e.g., k) is at a steady rate ω That is, the model assumes a predictable progression of θ between time steps. ω and α j and β j Each of is represented as a constant, but these parameters can also vary. That is, these parameters can form the hidden state variables of the random process, although they vary on a slower time scale.

[0265] As noted, phase angle and amplitude are hidden state variables that can be predicted from previous values ​​and current measurements. In this regard, it is known that θ k-1 and γ k-1 and use it to predict the current phase θ k and amplitude γ k , used to predict the electrode impedance measurement Z at location (x) at time k i (x k ,θ k , γ k). For example, in conjunction with a catheter model that predicts the location of an electrode in an impedance field (e.g., a patient reference frame) at location (x), the impedance model can predict the impedance value of the location, while the breathing model can predict the artifact of the electrode. As a result, impedance measurements including breathing artifacts can be predicted for the model electrode location in the patent reference frame. Accordingly, actual impedance measurements of physical electrodes (e.g., physical catheters) corresponding to the model electrodes can be obtained / measured. For example, a medical positioning system can obtain actual impedance measurements (e.g., with a certain amount of system noise) of corresponding electrodes of a physical catheter disposed in a patient reference frame. Predicted measurements and actual measurements can be used to update a variety of models. That is, predicted measurements and actual measurements can be used to determine the true location of the physical electrode in the patient reference frame. In addition, phase θ and amplitude γ can be updated based on predicted measurements and actual measurements. It can be understood that although these parameters have never been directly observed, the model allows for estimation of phase and amplitude. In this regard, a stochastic process infers phase and amplitude, and an iterative process is used to adjust the breathing model so that it substantially matches the actual phase and amplitude of the breathing cycle.

[0266] In an embodiment, an extended Kalman filter is used to infer hidden state variables of a quasi-periodic function. Hidden state measurements can be predicted from the hidden state variables at any time, and estimates of the state variables can be updated using the extended Kalman filter framework in a manner that allows updates to those portions of the hidden state variables that are accessible. Thus, at any moment in time, while there may not be enough information to determine portions of the state variables, by using the extended Kalman filter framework, predictions associated with the appropriate portions of the state variables (associated with the quasi-periodic function) can be made.

[0267] A difference can be made between a prediction and an actual measurement of an appropriate portion of a state variable associated with a quasi-periodic function, and the appropriate portion of the state variable can be updated based on the difference between the prediction and the actual measurement. In this way, the state variable can be modified over a given period of time rather than at a given moment in time. For example, a previous prediction of an appropriate portion of the state variable can be corrected based on a measurement at a current point in time.

[0268] In general, the model completely describes the motion of the medical device in the absence of noise. In other words, the model describes the possible states of the system and represents the individual state variables of the system. Typically, knowledge of the state variables at an initial time and at least partial knowledge of the system inputs and / or outputs allows the current state and / or subsequent states of the system to be estimated as a point or distribution in a state space (e.g., a geometric manifold). In such an arrangement, the state variables are set on coordinate axes of a state space (e.g., an N-dimensional space). In order to abstract from the number of inputs, outputs, and states, the state variables (e.g., the model) are represented as vectors that are combined to form the state vector of the system. The state of the system can be represented as a distribution 100 of possible states within the state space (i.e., as points in the state space). See Fig.16 . Each point in the state distribution includes information from all models (e.g., electrode locations, sensor locations, model variables, etc.). Typically, the mean of the distribution is considered to represent the most likely or true state of the system. Therefore, the mean can be used as an optimal estimate of all state variables. Typically, it is desirable to reduce the number of state variables (e.g., state vector components) to reduce the computational complexity of the system. However, it should be understood that in addition to the above-mentioned models, additional variables (e.g., models) can also be incorporated into the system model.

[0269] The state vector describes the movement of the medical device in the absence of noise. However, the actual measurements of the electrodes and magnetic sensors are noisy. That is, the measurements of these parameters each include an error or noise of unknown size. Actual systems and many system components (e.g., individual models) are random processes. In order to provide improved system modeling and estimation, noise should be included in the system model. Therefore, the present disclosure provides an observation model that defines the relationship between the state vector, the noise measurement, and the control vector (e.g., a known input). The observation model estimates the new state of the system using a new measurement of the previous state of the system (e.g., with a certain amount of noise).

[0270] One advantage of the disclosed system is that it provides a continuous update or estimate of the state of the system at each time step. That is, the observation model provides a continuous correction as opposed to a static correction factor. For example, the disclosed system can predict and update the state of the system approximately 100 times per second. Further beneficially, the observation model only requires information about the previous state of the system (e.g., at time k-1) and the current system measurement (e.g., at time k) to generate an update / estimate of the system state, where the state estimate is provided by the estimator.

[0271] The overall random process estimates a new position of the medical device and / or a new position of the electrodes of the medical device, as well as estimates of various state variables for each model (e.g., impedance model, magnetic model, catheter model, etc.). In an embodiment, the process assumes that the state of the system at time k evolves from the previous state at k-1 according to the following equation:

[0272] x k =F k (x k-1 )+B k (u k )+w k

[0273] in:

[0274] x k is the state vector containing the parameters of interest to the system (e.g., parameters of the model). This equation is used to predict the subsequent state with error.

[0275] F k is a state transition matrix that applies the effect of each system state parameter at time k-1 to the system state at time k. In other words, the transition matrix defines the relationship between the previous state vector and the current state vector.

[0276] u k is a vector containing any control inputs (e.g., robotic controls for a medical device).

[0277] B k is the control input matrix, which transforms the vector u k The effect of each control input parameter in is applied to the state vector. It is worth noting that this embodiment does not utilize any control input and input matrix, and B k and u k However, it should be understood that if control inputs are incorporated into the system, both a control vector and a control input matrix may be included.

[0278] w k is a vector containing the process noise term for each parameter (e.g., model) in the state vector. In an embodiment, it is assumed that the process noise is derived from a matrix with a covariance matrix Q k The covariance defined is drawn from a multivariate distribution.

[0279] A systematic errored measurement is also performed at each time step according to the following model:

[0280] z k =h k (x k )+v k

[0281] in:

[0282] z k is a measurement vector; a set of variables measured by a sensor (e.g., impedance measurement, magnetic sensor measurement, etc.).

[0283] h k is the observation model (i.e., transformation matrix) that maps the state vector parameters to the measurement domain. In other words, the observation model defines the relationship between the state vector and the noisy measurement; and

[0284] v k is a vector containing the measurement noise terms for each measurement variable in the measurement vector. In an embodiment, it is assumed that the process noise is derived from a covariance matrix R k The covariance defined is drawn from a multivariate distribution.

[0285] A random process is utilized to determine the true state of the system, which is a hidden state or potential state. The purpose of this process is to generate an estimate of the system state (e.g., electrode positioning, hidden variables of the model, etc.), and to determine the true state (e.g., a more accurate state) based on these estimates. In an embodiment, an estimator is implemented in an extended Kalman filter, which is suitable for use with a nonlinear system model or a linearized system model and / or a model with a non-Gaussian noise distribution. However, it will be understood that other estimators such as unscented Kalman filters, Markov models, and / or particle filters can be used to implant changes, which can each be applied to a nonlinear system and / or a system with a non-Gaussian noise distribution.

[0286] Each estimate of the estimator is a mean (i.e., the center of the distribution of state estimates) and a covariance that describes the probability about that mean. In applications, estimates include a priori estimates (predictions) before the measurements are combined and a posteriori estimates (updates) after the measurements are combined. The prior estimate uses the state estimates from the previous time step to produce the potential state at the current time step (mean). x k|k-l and covariance P k|k-l ) of:

[0287] x k|k-1 =F k x k|k-1 +B kUk

[0288] P k|k-1 =F k P k|k-1 +F k T +Qk

[0289] That is, the prior estimate is an estimate from a transformation matrix that produces the estimated distribution and covariance from the previous state (i.e., k-1). The transformation matrix takes each point in the original distribution and moves it into a new predictive distribution, which may have an expanded covariance (e.g., shifting Q k Added to the covariance matrix P) to account for unknown system noise. In the posterior estimation, the current prior prediction is combined with the observation model to improve the state estimate. More specifically, the observation model estimates (such as the mean x k|k-1 and covariance P k|k-1 ) is mapped to the measurement domain to predict the measurement:

[0290] z k =h k x k|k-1

[0291] The predicted measurements can be z k The actual measurement of the observable parameter z k (e.g. electrode measurement and sensor measurement of the system) for comparison:

[0292] y k =z k - z k

[0293] This allows the gain K of the system to be determined, where K minimizes x k|k - x k The expected sum of squared errors between Fig.17 It is shown graphically in Fig.17 is a 1-D representation of the state distribution combined with an observation model that produces a predicted measurement with a first predicted mean μ0 and a first predicted covariance σ0. The actual observed measurement is represented by a second distribution with a second mean μ1 and a second covariance σ1. The overlap of these distributions defines a system gain (e.g., a Kalman gain) that is used to correct the estimated state and the estimated covariance. In other words, the two distributions are fused to generate an updated distribution with a fused mean μ' and a fused covariance σ' (e.g., two Gaussian distributions are multiplied together to generate a Gaussian distribution of the overlapping portion of the two distributions). The gain K can be combined with the estimated state distribution and the estimated covariance to generate an updated state distribution (e.g., an updated state mean and an updated covariance):

[0294] x k|k = x k|k-1 +K k y k

[0295] P k|k =(IK k H k ) P k|k-1 (IK k H k ) T +K k R k K T .

[0296] The updated average state can be used to determine an updated or true position (eg, calculated position) of the electrode and / or magnetic sensor. In addition, the state can be used to update various state variables of various models.

[0297] constraint

[0298] Although the above process allows the current state of the system to be predicted, it is further recognized that the state vector and the corresponding estimate of the state can be subject to a variety of constraints. Such constraints can be used to limit or otherwise improve the state distribution and thereby improve the overall accuracy of the system. Given a state vector x, the model constraint can be expressed in functional form as g(x)=0. In this form, any true state must satisfy this equation. For example, impedance measurements are made by driving current on surface patch electrodes to excite the electrodes. As previously described, the electrode excitation process occurs quickly and sequentially because different groups of patch electrodes are selected and one or more unexcited (in an embodiment) surface electrodes are used to measure voltage. During the delivery of the excitation signal (e.g., a current pulse), the remaining (unexcited) patch electrodes can be referenced to a reference or abdominal patch while measuring the voltage applied to these remaining electrodes. In addition to driving a specific surface electrode patch pair, the potential across each surface patch electrode can be obtained for all samples. In Fig.14 In the two-dimensional representation shown in , the back (B), left (L), chest (C), and right (R) surface patch electrodes define a current loop in the patient. Kirchhoff's voltage law imposes linear constraints on this voltage loop. Specifically, the sum of the driving potentials (i.e., impedances) from the loop across all pairs of patch electrodes must be zero. That is:

[0299] Z B-L +Z L-C +Z C-R +Z R-B =0

[0300] Accordingly, the sum of the driving potentials on any electrode of the cycle must be zero. Therefore, this constraint can be applied to the portion of the state vector related to impedance measurement (e.g., impedance model). Another constraint can be that the magnetic model may be constrained to correspond to the change of rigid body transformation without scaling. That is, all identified objects must have the same relative orientation before and after the transformation. Other constraints can be applied to composite models or independent models. In applications, one or more such constraints can be applied to limit or improve the state distribution.

[0301] Fig.18 A constraint g(x)=0 on the state distribution estimate is shown. As shown, the constraint forms a feasibility manifold or constraint manifold in the state space that satisfies the constraint. That is, the state of g(x)=0. As shown in Figure 11, the initial state distribution estimate 100 is not located on the constraint manifold. Therefore, in order to enforce the constraints on the state distribution estimate, the state distribution estimate must be moved to the constraint manifold. The constraint application is performed by generating a delta function that satisfies the constraint and multiplying it by the state distribution estimate to produce a constrained state distribution estimate.

[0302] The constraint application can be approximated using a first order Taylor series expansion that generates a linear representation or tangent line 102 about the mean of the unconstrained state distribution estimate 100. This produces a first order approximation of the unconstrained mean of the state distribution estimate. The tangent line can be projected onto the surface of the constraint. More specifically, the first order approximation can be orthogonally projected onto the null space of the Jacobian of the constraint:

[0303]

[0304] Project the distribution into the null space of G. By continuously projecting through G, the estimated state distribution will track the constraint manifold even if the constraints are not exactly linear. The result is that the state distribution estimate is constrained by that constraint.

[0305] An unlikely state

[0306] The parameterization of a system (e.g., within a system model) can describe possible system states that cannot be realized or well determined by measurements of the system. Therefore, the present disclosure may further include penalizing such unlikely states. More specifically, the present disclosure provides a method for regularizing an estimate so that a more likely estimate / state is generated for the system. In order to regularize the estimated state distribution, a regularization function can be defined that expresses a quantity that is proportional to the likelihood of the state. Typically, the likelihood function describes the rationality of a given observed state and is the product of a probability distribution function and a state distribution. In an embodiment, a negative log likelihood is utilized. In such an embodiment, an impossible state has an infinite negative log likelihood, while the most likely state has the minimum negative log likelihood. In order to apply the regularization, in an embodiment, a probability density function (regularized PDF) is calculated by negating, exponentiating, and normalizing the negative logarithmic function. The estimated state distribution is then multiplied by the regularized PDF and renormalized to create a regularized state distribution that ignores unlikely states (e.g., states outside the state distribution and the regularized PDF combination).

[0307] In an embodiment, the general negative log-likelihood function can be approximately applied by a second-order Taylor series expansion of the negative log-likelihood function at the mean of the estimated state distribution to create a probability density function. In an embodiment, the approximation of the negative log-likelihood function can be performed via the following equation:

[0308]

[0309] The second-order Hessian Viewed as the inverse of the covariance, the Gaussian mean is given by the Jacobian of the second-order expansion multiplied by the inverse of the Hessian. This approximation is equivalent to a Gaussian PDF that can be multiplied by the state distribution by well-known methods.

[0310] The regularization of the state distribution estimate is Figures 7A-7C Specifically, Fig. 7A A state distribution 100 is shown. Figure 7B The regularized PDF 104 applied to the state distribution is shown. Figure 7C A regularized state distribution 106 is shown, which is generally surrounded by a dashed circle for illustrative purposes. It should be understood that the regularized state distribution excludes unlikely states from the initial state distribution estimate. This results in a new state distribution (e.g., a regularized state distribution) with a different mean and a smaller covariance. In other words, the regularization process results in a tighter state distribution that more accurately predicts the true state of the system.

[0311] Fig.19A block diagram depicts an example of a computer-readable medium in communication with a processing resource of a computing device according to an embodiment of the present disclosure. Figure 1 As discussed, the main controller 12 may utilize software, hardware, firmware, and / or logic to perform many functions. The main controller 12 may include multiple remote computing devices.

[0312] The main controller 12 can be a combination of hardware and program instructions configured to perform multiple functions. The hardware can include, for example, one or more processing resources 160, computer readable media (CRM) 162, etc. Program instructions (e.g., computer readable instructions (CRI) 164) can include instructions stored on the CRM 162 and can be executed by the processing resource 160 to implement the desired functions (e.g., using an observation model to determine an updated positioning of an electrode on an impedance-based medical device, etc.). CRI 164 can also be stored in a remote memory managed by a server and represents an installation package that can be downloaded, installed, and executed. The main controller 12 can include a memory resource 166, and the processing resource 160 can be coupled to the memory resource 166.

[0313] The processing resource 160 may execute the CRI 164, which may be stored internally or externally on a non-transitory CRM 162. The processing resource 160 may execute the CRI 164 to perform a variety of functions, including those described above.

[0314] The plurality of modules 168, 170, 172, 174, 176 may be submodules or other modules. For example, the estimation module 172 and the estimator module 174 may be submodules and / or included in a single module. In addition, the plurality of modules 168, 170, 172, 174, 176 may include individual modules that are separate and distinct from each other.

[0315] The navigation module 168 may include the CRI 164 and may be executed by the processing resource 160 to obtain measurements from the medical device 24 and render output on the display 16. The measurements may include impedance measurements of the electrodes 30 disposed on the catheter and / or impedance surface patch measurements. The measurements may also include magnetic positioning of the magnetic position sensor 28 disposed on the catheter and / or magnetic measurements of the patient reference sensor 26. The navigation module 168 may call the locator module 170 to obtain updated positions of the electrodes and / or sensors of the medical device 24.

[0316] The locator module 170 may include the CRI 164 and may be executed by the processing resource 160 to coordinate the operations of the estimation module 172, the model module 174, and the estimator module 176. In an example, the locator module may receive raw measurements from the navigator module in conjunction with an update request. The locator module 170 may call the estimation system module 172 to pre-process the raw measurements. Once the pre-processed measurements are obtained from the estimation module, the locator module 172 may provide the pre-processed measurements to the estimator 176 to request an update of the current state of the system.

[0317] The estimation system module 172 may include the CRI 164 and may be executed by the processing resource 160. In an embodiment, the estimation system module 172 defines the stochastic process of the entire system, including state transitions and observation models. In an embodiment, the estimation system may be a Kalman system implementing Kalman filtering techniques. In an embodiment, the estimation system module 172 calls the model module 174 to the estimator module 176 to obtain an updated state estimate.

[0318] The model module 174 may include the CRI 164 and may be executed by the processing resource 160. The model module may include multiple separate models. These separate models may include one or more catheter models. In an embodiment, the medical device / catheter may be represented as one or more models. In addition, the catheter model may include models of different medical devices used when there is more than one catheter in the patient reference system. The separate models may also include a magnetic model (e.g., a magnetic transformation model) that transforms positioning from the patient reference system to the magnetic reference system. The separate models may also include an impedance model or an impedance transformation model that predicts the impedance of the positioning in the patient reference system.

[0319] The estimator module 176 may include the CRI 164 and may be executed by the processing resource 160. The estimator module may receive update requests and inputs from the estimation system 172 and in response provide updated state estimates and / or predicted measurements. In an embodiment, the estimator module may be implemented as an extended Kalman filter.

[0320] Fig. 20A flowchart 300 is depicted that is associated with an overall process (e.g., a sensor fusion process) to update an estimated electrode positioning within a three-dimensional space according to an embodiment of the present disclosure. Initially, the flowchart includes processing raw measurements at box 302. The raw measurements may include raw patch impedance measurements from surface patch electrodes and patch continuity data. The patch continuity data may provide an indication of the contact of each surface patch and therefore its reliability. Raw electrode impedance measurements of electrodes of a medical device / catheter (hereinafter referred to as a catheter) may also be received. Raw magnetic data of a magnetic sensor and a patient reference sensor for the catheter are also received. Processing the raw measurements may include processing the raw measurements to detect any measurements (e.g., with non-Gaussian errors) outside a predetermined statistical range for the measurements. Any such irrelevant measurements may be excluded from subsequent processing. In the case of raw electrode impedance measurements, in some embodiments, the impedance measurements may be filtered to remove noise from the impedance signal. In an embodiment, bioimpedance scaling may be performed to help address drift in impedance measurements (e.g., position values) of electrodes in some embodiments. Such bioimpedance scaling can compensate for systematic changes in conductivity, under the assumption that the scalar multiplier accounts for changes in impedance measurements over time. Such scaling can measure the average impedance on the non-driven patch and scale all impedance measurements by the ratio between the current average and the historical value. In another embodiment, patch center subtraction can be applied. Patch center subtraction is a drift compensation algorithm complementary to bioimpedance scaling that compensates for changes in the system reference potential. In some cases, the system reference can be located at a certain distance from the heart. The patch center subtraction algorithm calculates a virtual reference coordinate from the non-driven patch and subtracts the coordinate from the impedance coordinate measured by the electrode after bioimpedance scaling. Typically, such patch center subtraction re-references the impedance coordinate to a location closer to the center of the heart. Other processing of the raw signal is possible and is considered to be within the scope of the present disclosure.

[0321] At box 304, the flowchart includes calculating one or more state constraints to limit or otherwise improve the state distribution and thereby improve the overall accuracy of the system. Once such constraints are calculated, the previous state can be projected to the constraints in box 306. Notably, this can include expanding the covariance matrix for the previous state to account for additional uncertainty or noise in the system for the upcoming prediction. Once this additional process noise is included in the previous state, the previous state may no longer be on the constraint manifold. Therefore, as described above, the previous state can be moved to the constraint manifold.

[0322] Once the previous state is projected onto the constraints, the next state of the system is predicted at block 308 of the flowchart. That is, using the state transition matrix F tA new state distribution (e.g., mean and covariance) is generated that applies the effect of each system state parameter at time k-1 to the system state at time k. In other words, the current state is predicted. Once the new state distribution (e.g., mean and covariance) is generated, one or more constraints may be calculated for the current state at block 310. At block 312, the current state may be projected to the constraints.

[0323] In an embodiment, the unlikely state in the current state is penalized to reduce the distribution of the current state. In an embodiment, the negative log likelihood is calculated at block 314 of the flow chart. In an embodiment, a probability density function is generated. The function can be applied to the current state distribution. That is, the current state distribution can be regularized at block 316 of the flow chart.

[0324] The predicted measurements can be generated at box 318 of the flowchart. That is, given the current predicted state, the observation model can be used to predict the measurements (e.g., electrode and sensor position measurements) to produce a distribution of predicted measurements with a mean and covariance. Once the measurements are predicted, they can be compared with the actual measurements. At box 320 of the flowchart, the difference between the predicted measurements and the actual measurements can be used to correct the current predicted state. In an embodiment, once the current state is corrected, outliers can be identified at box 322 of the flowchart and removed from the current state. At this point, a new state distribution (e.g., time step) will be generated for the current update. Based on the new or updated state distribution, electrode positioning can be calculated at box 324 of the flowchart. In addition, all state variables of a variety of models can be calculated based on the updated state distribution.

[0325] Fig.21A , 21B 22 shows a process call graph, which describes the process call graph in Fig.19 An embodiment of the interaction between modules 168, 170, 172, 174, 176 described in. Fig.21A, depicting an update process call graph 340. Initially, the navigation module 168 requests a state update 342 from the locator module 170. The navigation module 168 provides new measurements (e.g., raw measurements) for the current time step (e.g., time t) to the locator module 170 in conjunction with the update request. The locator module 170 requests 344 the estimation system module 172 to process the raw measurements. The estimation system 172 communicates with one or more catheter models of the model module 174. More specifically, the estimation system requests 346 to pre-process the raw measurements relative to a particular catheter model. It is noteworthy that in the case of multiple catheters within the patient reference frame, the process can be performed for multiple catheters using multiple catheter models. The model module 174 returns 342 the pre-processed measurements to the estimation system module 172, which returns these measurements to the locator module 170. The locator module 170 provides the pre-processed measurements to the estimator module 176 via a request 352 to update the state distribution (e.g., state mean and state covariance) of the system for the previous time step (k-1). Request 352 also includes a measurement of the current time step.

[0326] The estimator module 176 works with the estimation system module 172 that describes the random process of the system to generate an updated state mean and state covariance for the previous time step. The estimator module 176 requests the estimation system module 172 to calculate the constraints of the system (e.g., state vector), and the estimator system module 172 provides the constraints to the estimator module 176 in the request and response loop 354. The estimator 176 projects 356 the state (e.g., k-1) to the constraints. Once constrained, the estimator module 176 requests the estimation system module 172 to predict the next state of the system, and the estimation system 172 returns the next state distribution estimate (e.g., mean and covariance) of the system to the estimator 176 in the request and response loop 358. The estimator module 176 and the estimation system module 172 calculate the updated constraints for the next state distribution estimate (e.g., at time t) in the request and response loop 360. The estimator projects 362 the next state distribution estimate to the updated constraints. The estimator module 176 and the estimation system module 172 calculate a likelihood function for the state distribution estimate in the request and response loop 364. The estimator module 176 generates 366 a regularized state distribution estimate using the likelihood function.

[0327] The estimator module 176 and the estimation system module 172 then predict the measurements (e.g., electrode and sensor measurements) through the application of an observation model that maps the regularized state distribution estimate to the measurement space in the request and response loop 368. Fig. 22This loop 368 is discussed further. Based on the predicted measurements, the estimator module 176 determines the correspondence 370 of the predicted measurements to the actual measurements. In an embodiment where the estimator module 176 is an extended Kalman filter, this is a determination of the optimal Kalman gain. The correspondence between the predicted measurements and the actual measurements is used to correct the state distribution estimate to generate an updated state mean and covariance (e.g., an updated state). That is, the regularized state distribution estimate is corrected to generate an updated state mean and covariance at time k, which is provided 372 to the locator module 170. In an embodiment, the locator module requests 374 the estimation system module to identify 374 suspicious outliers with the updated state mean and state covariance. The locator module estimation system module 172 returns 376 the state to the navigation module 170.

[0328] If the updated state mean and state covariance are acceptable, the call graph is in Fig.21B Continued above. In an embodiment, the navigation module requests the positioning of each catheter, electrode and / or sensor in the patient coordinate system based on the updated state mean and covariance. That is, the navigation module 168 requests 378 the locator module 170 to obtain the positioning (e.g., electrode positioning) in the patient reference system. The locator module 170 requests 380 electrode positioning from the estimation system based on the updated state. The estimation system 170 requests a transformation (e.g., patient to impedance transformation) from the model module 174. The transformation is applied to the updated electrode positioning to predict the true electrode positioning in the patient reference system, which is provided to the navigation module 170. The navigation model 170 can then render or otherwise process 386 the electrode positioning for the imaging system and output the electrode positioning to the display 16.

[0329] As described above, the estimator module 176 and the estimation system module 172 predict measurements (e.g., electrode and sensor measurements) by applying an observation model that maps the regularized state distribution estimate to Fig.21A The measurement space in the request and response loop 368. Fig. 22The request and response loop is further shown. As shown, the estimator module 176 initially requests 402 the estimation system module 172 to predict the measurement. At this point, the estimation system module 172 interacts with the model module 174. More specifically, the estimation system 172 provides the magnetic portion of the state distribution estimate to the magnetic transformation model 174C, and provides the impedance portion of the state distribution estimate to the impedance transformation model 174C. For example, in the request and response loop 404, the estimation system requests the transformation of the patient reference frame to the magnetic reference frame from the magnetic transformation model 174B. In such a request, for each state in the estimate, the estimation system module 172 can provide six variables (e.g., in the case of a 6-degree-of-freedom sensor) to the magnetic transformation model 174B, which provides the transformation matrix as a response. In the request and response loop 406, the estimation system requests the transformation of the patient reference frame to the impedance coordinate system of the impedance from the impedance transformation model 174C. In such a request, the estimation system module 172 can provide impedance parameters for each state to the impedance transformation model 174C, which provides a transformation matrix in response. In loop 408, the estimation system module 172 can obtain the predicted coordinates (e.g., positions) of the sensors and electrodes in the patient reference frame from the catheter model 174A. In a second loop 410, the estimation system module 172 can apply the obtained transformation to each predicted position and catheter model 174A. In an embodiment, a Jacobian is calculated 412 at the mean for each state. The Jacobian describes the local difference of each state due to the state space transformation. The predicted measurement and Jacobian are returned 414 to the estimator module 176, which compares the predicted measurement with the actual measurement to calculate the gain (e.g., correction) of the estimated state distribution, and thereby generates an updated state and an updated covariance for time k.

[0330] Confidence and Fault Detection

[0331] As about Fig. 20 As noted, the sensor fusion process may include determining the reliability of the raw measurements (e.g., block 302) and identifying outliers in the estimated state distribution (e.g., block 322). Along these lines, the measured location and / or estimated location of a catheter feature (such as an electrode, magnetic sensor) or the catheter itself (e.g., a catheter spline) has an indication of whether the location is trustworthy. In an embodiment, such an indication may be based on statistical confidence intervals and convergence times calculated from the estimated state distribution. When the reliability of the location is poor, the collection of historical data points will be inhibited.

[0332] For each set of measurements from the electrodes (e.g., 6 impedance queries (hereinafter referred to as "impedance modules"), measurements from the sensors, each patch, all patch impedance data combined, all magnetic data combined, each single impedance module, all impedance data combined, each single catheter), the reliability value indicates whether there are multiple measurement problems, such as an electrode disconnected or improperly connected, or measurements that cannot be explained by the state transition and observation model. When these measurement problems are detected, for some types of failures, an indication of the failure can be displayed to the user and / or data collection from the corresponding medical device can be inhibited. The detection of some measurement problems will exclude the corresponding measurement from the sensor fusion process. In general, it is desired to: 1) determine whether the corresponding estimated catheter positioning is reliable; 2) detect measurement failures that require the user to take action to correct and / or detect erroneous measurements to be excluded from the sensor fusion algorithm; and / or 3) determine statistical outliers.

[0333] reliability

[0334] In an embodiment, after the sensor fusion process begins, all location estimates can be marked as unreliable for a specified time interval. This unreliable period can be realized in the estimation of magnetic transformations, impedance transformations, catheter shape, and other state variables may take time to stabilize. Therefore, during a specified time interval (such as 120 seconds) after the sensor fusion process begins, all location estimates can be marked as unreliable. Location estimates on the catheter can also be marked as unreliable during the time after the catheter is introduced, when the catheter is in the sheath, during the time after the catheter leaves the sheath, or when the catheter is disconnected.

[0335] In an embodiment, a boundary may be placed on a variety of distance errors and / or the positioning of the participating catheter features may be marked as unreliable. In such embodiments, the boundary may include determining: the error distance between the actual positioning of the catheter feature and the estimated positioning; and the distance error between the catheter feature and the anatomical positioning (such as an adjacent surface, an adjacent lesion, or an adjacent mapping point); and / or the distance error between two independent catheter features on the same or different catheters. For such distances, a tolerance is set. The tolerance can be set as a physical distance (e.g., 2 mm) or as a percentage of the estimated value (e.g., 10%). In application, the estimator generates a probability distribution on the state variables of the system. For example, a one-dimensional state distribution (e.g., a bell curve). For each possible state, the positioning of all catheter features can be calculated. The system calculates the probability that each distance of the catheter feature relative to the mean of the state variable exceeds its tolerance. If the probability exceeds a threshold (e.g., two standard deviations in a one-dimensional distribution), the positioning of the catheter features participating in the distance is marked as unreliable.

[0336] In an embodiment, a numerical method for calculating the probability that a distance exceeds its tolerance includes Monte Carlo sampling. In such an embodiment, states are randomly sampled from a probability distribution of states. The distance of each sample is calculated. The percentage of distances that exceed their tolerance is determined. If the percentage of distances that exceed the tolerance exceeds a threshold, the distance is considered unreliable.

[0337] In an embodiment, the numerical method for calculating the probability that a distance exceeds its tolerance includes an analytical heuristic. In this embodiment, for a Gaussian state distribution, a local linearization of the estimated catheter feature offsets is calculated. Each offset will have a Jacobian determinant. If the state distribution has a mean x and a covariance S, then the distribution of the offsets can be well approximated by:

[0338] N(0,JSJ T )

[0339] If the error is close to isotropic, then trace(JSJ T The square root of ) approximates the standard deviation of the squared distance error. When this value exceeds the upper limit, the distance is considered unreliable, and therefore the participating positioning estimates are considered unreliable.

[0340] Wrong measurement

[0341] Several heuristic fault detection algorithms can be implemented to detect problems with the impedance data. The following conditions detect electrodes that are considered disconnected or physically faulty. If these conditions are detected, the measurement is excluded from the estimator. In an embodiment, electrodes whose impedance values ​​are greater than a threshold are considered disconnected or faulty. In another embodiment, electrodes with impedance values ​​that are different from the average of all electrodes are not considered faulty.

[0342] The following embodiments detect electrodes that are considered to be disconnected, physically faulty, or have interchanged connections. These measurements are excluded from the estimator. In an embodiment, electrodes whose positioning estimates based on the sensor fusion process differ from their estimates based on the impedance main algorithm by more than a threshold are excluded. In another embodiment, a broken line is suitable for electrode positioning. If a broken line passing through a series of adjacent electrodes has an acute angle at one of the electrodes, the electrode is considered to be a candidate for misconnection detection. Alternatively, if the average bending angle exceeds a threshold, all electrodes on the catheter spline are considered to be candidates for misconnection detection. Then, misconnection candidates that are not actually misconnected are removed from consideration as follows: If there is a single electrode marked as a potential misconnection, and no other electrodes have a bad measurement state or are marked as disconnected, the single electrode is removed. In an embodiment, if the sensor fusion process estimate of electrode positioning is closer to the impedance positioning of the electrode than the impedance positioning of any other misconnection candidate, the electrode is removed.

[0343] Statistical outliers.

[0344] The state estimator calculates a probability distribution for all measurements. For each measurement group, the measurement can be compared to its expected distribution to calculate a test statistic. In an embodiment with a general Gaussian distribution calculated by the estimator, in an embodiment, the Mahalanobis distance from the mean to the measurement can form a test statistic. The Mahalanobis distance is a measure of the distance between a point P and the mean of the distribution D containing the point. If the distance exceeds a predetermined threshold, the measurement group is marked as a statistical outlier.

[0345] Embodiments of various devices, systems and / or methods are described herein. Many specific details are set forth to provide a thorough understanding of the overall structure, function, manufacture and use of the embodiments described in the specification and shown in the drawings. However, it will be appreciated by those skilled in the art that these embodiments can be practiced without such specific details. In other cases, well-known operations, parts and elements are not described in detail to avoid making the embodiments described in the specification unclear. It will be appreciated by those of ordinary skill in the art that the embodiments described and shown herein are non-limiting examples, and it will therefore be appreciated that the specific structural and functional details disclosed herein may be representative and do not necessarily limit the scope of the embodiments, which scope is defined only by the appended claims.

[0346] References throughout the specification to "various embodiments," "some embodiments," "one embodiment," or "an embodiment," etc., mean that a particular feature, structure, or characteristic described in conjunction with that embodiment is included in at least one embodiment. Thus, phrases such as "in various embodiments," "in some embodiments," "in one embodiment," or "in an embodiment," etc., that appear throughout the specification do not necessarily all refer to the same embodiment. Furthermore, in one or more embodiments, particular features, structures, or characteristics may be combined in any suitable manner. Thus, particular features, structures, or characteristics shown or described in conjunction with one embodiment may be combined in whole or in part with features, structures, or characteristics of one or more other embodiments without limitation, provided that such combination is not illogical or non-functional.

[0347] It should be understood that the terms "proximal" and "distal" may be used throughout the specification with reference to the end of an instrument that a clinician manipulates to treat a patient. The term "proximal" refers to the portion of the instrument that is closest to the clinician, and the term "distal" refers to the portion farthest from the clinician. It will be further understood that for simplicity and clarity, spatial terms such as "vertical," "horizontal," "upward," and "downward" may be used herein with respect to the embodiments shown. However, surgical instruments may be used in many orientations and positions, and these terms are not intended to be restrictive and absolute.

[0348] Although at least one embodiment for estimating the positioning of an electrode based on utilizing a system model has been described above with a certain degree of particularity, those skilled in the art may make various changes to the disclosed embodiments without departing from the spirit or scope of the present disclosure. All directional references (e.g., up, down, upward, downward, left, right, leftward, rightward, top, bottom, above, below, vertical, horizontal, clockwise and counterclockwise) are only used for identification purposes to help readers understand the present disclosure, rather than limiting the position, orientation or use of the device in particular. References (e.g., fixing, attaching, connecting, connecting, etc.) should be interpreted broadly, and may include intermediate members between the connections of elements and relative motion between elements. In this way, references may not necessarily infer that two elements are directly connected and have a fixed relationship with each other. It is intended that all contents contained in the above description or shown in the accompanying drawings be interpreted as being exemplary only, rather than restrictive. Changes in details or structure may be made without departing from the spirit of the present disclosure as defined in the appended claims.

[0349] Any patent, publication, or other public material described as being incorporated herein by reference is included herein in whole or in part only to the extent that the included material does not conflict with existing definitions, statements, or other public materials set forth in this disclosure. Therefore, and to the extent necessary, the disclosure explicitly set forth herein supersedes any conflicting material incorporated herein by reference. Any material or portion thereof described as being incorporated herein by reference but conflicting with existing definitions, statements, or other public materials set forth herein is included only to the extent that the included material does not conflict with existing public materials.

Claims

1. A method for identifying electrode locations and outputting the electrode locations on a display, comprising: defining relative positions of physical electrodes in the catheter shape model, wherein the relative positions correspond to the spacing of the physical electrodes of the physical catheter; predicting the positioning of model electrodes in a patient reference frame by utilizing a catheter shape model, the catheter shape model including the positioning of model electrodes and model magnetic sensors in the catheter reference frame, wherein the model electrodes and the model magnetic sensors correspond to a set of physical electrodes and magnetic sensors of the physical catheter disposed in the patient reference frame; generating predicted impedance responses of the model electrode locations by applying an impedance model of the applied potential field to the model electrode locations, wherein the impedance model transforms each model electrode location into a predicted impedance response; measuring an impedance response of the physical electrodes of the physical conduit in response to the applied electric potential field; and generating a calculated position of the physical electrode based at least on the predicted impedance response and the measured impedance response; and The calculated positions of the physical electrodes are output to a display.

2. The method according to claim 1, further comprising: predicting a positioning of the model magnetic sensor in a patient reference frame by utilizing the catheter shape model; generating predicted magnetic responses of model magnetic sensor locations; measuring a magnetic response of the magnetic sensor in a magnetic reference frame in response to an applied magnetic field; Wherein the calculated positioning is further based on the predicted magnetic response and the measured magnetic response.

3. The method according to claim 1, further comprising: A catheter transformation is applied to the catheter shape model to transform the positions and orientations of model electrodes of the catheter shape model from a catheter reference system to a patient reference system.

4. The method according to claim 3, wherein: Applying the catheter transformation to the catheter shape model includes applying a rigid body six degrees of freedom transformation to the catheter shape model.

5. The method according to claim 3, wherein: Applying the catheter transformation to the catheter shape model further comprises: A catheter transformation is applied to the model magnetic sensor to transform the position and orientation of the model magnetic sensor from a catheter reference system to a patient reference system.

6. The method according to claim 1, wherein: Generating a predicted impedance response further includes: The impedance model is updated based on the predicted impedance response and the impedance response of the physical electrode.

7. The method according to claim 1, wherein: The catheter shape model and the impedance model are state variables of a composite model that models the physical catheter in the patient reference frame.

8. The method according to claim 7, wherein: An extended Kalman filter is used to infer the state variables.

9. The method according to claim 7, further comprising: The composite model is used to generate an estimated state distribution of potential electrode locations, wherein the calculated location is generated using the state distribution.

10. The method according to claim 9, further comprising: At least a first constraint is applied to the estimated state distribution, wherein the first constraint constrains at least one of the state variables, wherein the first constraint limits the estimated state distribution.

11. The method according to claim 9, further comprising: A function is applied to the estimated state distribution to remove unlikely states from the estimated state distribution.

12. The method according to claim 9, further comprising: comparing the predicted impedance response to the impedance response; as well as A correction is generated based on the comparison.

13. The method according to claim 12, further comprising: applying the correction to the estimated state distribution to generate an updated state distribution, wherein the calculated position is generated using the updated state distribution, and An outlier state in the updated state distribution is identified, wherein the outlier state is removed from the updated state distribution.

14. A non-transitory computer readable medium storing instructions for identifying electrode positioning, the instructions being executable to perform the steps of the method according to any one of claims 1 to 13.

15. A system for identifying electrode positioning, comprising: a physical catheter having physical electrodes disposed in three-dimensional space; a medical positioning system for measuring an impedance response of the physical electrode in response to an applied electric potential field; a processor and a memory for storing non-transitory computer-readable instructions; and a display operatively connected to the processor and the memory, The processor is adapted to execute instructions to perform the steps of the method according to any one of claims 1 to 13.

Images