Computer-implemented method and system for assisting in mapping cardiac rhythm abnormalities

Abstract

A computer-implemented method and system are described. The method comprises the steps of: identifying a region having an electrical activation sequence within a chamber of the heart from an electrograph, the electrical activation sequence characterizing the region as a potential drive of an abnormal heart rhythm; for each sensing location at or substantially around the region, determining an earliest activation electrode site from a primary activation; for each determined earliest activated electrode site: calculating a value for each of a plurality of modifiers associated with the electrode site, the modifiers determined from electrogram data of the site, tissue characteristics of the site, or anatomical characteristics of the site; determining a sorting factor calculated according to the plurality of modifiers; sorting each of the earliest activated electrode sites according to a sorting factor; and outputting data identifying the region, the data changing a significance of each of the determined earliest activated electrode sites depending on the ranking.

Description

[0001] This application is a divisional application of the Chinese national phase application of PCT International Application (entitled Computer Implemented Method and System for Aiding Mapping Heart Rhythm Abnormalities) having an international filing date of 29 October 2020 under the PCT and an application number PCT / GB2020 / 052734, which claims priority from a UK patent application having an application number GB2005938.8, filed on 29 October 2020. TECHNICAL FIELD

[0002] The present invention relates to a computer implemented method and system for aiding in mapping heart rhythm abnormalities, and in particular to a method and system for identifying cardiac regions that are statistically likely to drive abnormal heart rhythms. BACKGROUND

[0003] Atrial fibrillation (AF) is the most common persistent heart rhythm abnormality. The incidence of AF is increasing, in part due to an ageing population, and it is referred to as an increasingly serious epidemic. Atrial fibrillation causes the heart to contract irregularly, resulting in symptoms of palpitations and discomfort, and increasing the risk of stroke, heart failure (HF) and death. For patients with symptomatic AF, catheter ablation (CA) is a safe treatment option. Success rates for these procedures have improved over time due to better understanding of AF, development of new techniques and technologies, and more physician experience. However, success rates for these procedures still only remain between 50% to 70%. One major reason for the difficulty in finding a specific site responsible for the persistence and maintenance of AF (hereinafter referred to as an AF driver) is the irregular and chaotic nature of activation wavefronts in atrial fibrillation. The constant meandering and instability of individual rotations, reentries or focal activations makes interpretation of activation sequence very complex.

[0004] Recently, a number of computational and electroanatomical methods have been developed allowing for the recording of electrical data (i.e. electrograms) from within the atrium and presenting to the physician in a way that identifies specific “driver” regions. These drivers can also be more or less easily identifiable depending on the relationship between the frequency of the driver and the activation frequency of non-driver random and chaotic activity. Panoramic mapping techniques attempt to address this issue. Here, a multi-polar catheter is inserted into the chamber of interest whilst simultaneously acquiring signals across the chamber. Examples of this include non-contact mapping (Ensite, Abbott Medical; alternatively Acutus Medical), and specific 2D and 3D contact mapping methods (e.g. Cartofinder, Biosense Webster, J&J; Topera, Abbott Medical; Rhythmia, Boston Scientific).

[0005] Evidence is conflicting as to whether electrogram features can be used as surrogate markers for local drivers in human persistent AF. These can be referred to as atrial fibrillation drivers (AFD). It is understood that a single AFD can cause regular, non-fibrillatory arrhythmias, and thus AFDs can be considered as atrial fibrillation / arrhythmia drivers. While organizational features of electrograms can better identify sites that play a mechanical role in AF, the reliability of fast markers is low. Optical mapping studies in animals have shown that AF is maintained by sites that exhibit the fastest cycle length (CL) and highest dominant frequency (DF), however, in humans, this has proven to be a poor predictor of sites that support AF. The poor correlation can be because drivers in AF lack spatiotemporal stability, which can explain the apparent inconsistency of fast sites. Furthermore, optical mapping studies in animals have shown high-to-low frequency gradients within the atrium at rotor sites. While frequency gradients have also been confirmed in humans with AF, these gradients are limited to intra-atrial gradients.

[0006] STAR Mapping method

[0007] The STAR mapping method has been described in detail in the above patent applications and has been validated in vitro and in vivo by mapping atrial tachycardias (AT) prior to being used to map AF. In brief, the principle of the STAR mapping method is to use temporal comparisons of multiple electrograms to establish individual wavefront trajectories associated with AF. This is then used to identify atrial regions that most often activate prior to neighboring regions. By collecting data from many activations, a statistical model can be formed. This allows regions of the atrium to be ordered according to the amount of time they activate prior to neighboring regions. Monopolar activation timing is typically taken as the maximum negative deflection (peak negative dv / dt), but activation timing can be derived from other methods such as dipole density or peak bipolar or analytical full-pole voltage. By utilizing a predefined refractory period, the mapping method avoids assigning activations from individual wavefronts or fractional electrograms. The mapping method also excludes unreasonable electrode timing relationships due to conduction velocity limitations.

[0008] One form of STAR mapping shows a projection of electrode positions encoded by color onto a replica of the patient's atrial geometry created in a standard 3D mapping system. Each color represents the proportion of time an electrode leads other paired electrodes, as highlighted by the color bar to the right of the STAR map.

[0009] Problems with current methods

[0010] The global distribution of ordered sites of potential AF drivers when mapping the atrium using a basket can be produced using mapping systems that seek to identify the "leading" proportion of time of an electrode relative to its neighboring electrodes. However, for sequential high-density mapping techniques, no information is provided about the relative importance of early sites to each other. For example, many sites can all be considered leading.

[0011] Another problem with the analysis of sequentially acquired data relates to the complexity of interpretation of the leading edge designated as the edge of the acquired region. This is exemplified in Figures la-d Such data-based rule interpretation methods are advantageous in that they allow for automatic interpretation, and the removal or reduction of the prominence of the display of passive activation sites of the acquired edge reduces the area that can be highlighted as a potential driver site. SUMMARY

[0012] According to one aspect of the application, there is provided a computer-implemented method for identifying one or more regions of a heart responsible for supporting or initiating an abnormal heart rhythm, the computer-implemented method using electrogram data recorded from a plurality of electrodes on a multi-polar cardiac catheter, the electrogram data being obtained from a respective series of sensing locations on the heart over a recording time period, the method comprising the steps of:

[0013] identifying a region within a chamber of the heart having an electrical activation sequence according to the electrogram data, the electrical activation sequence characterizing the region as a potential driver of the abnormal heart rhythm,

[0014] determining, for each sensing location at or substantially surrounding the region, an earliest activation electrode site according to the primary activation;

[0015] for each determined earliest activation electrode site:

[0016] calculating a value for each of a plurality of modifiers associated with the electrode site, the modifiers being determined from the electrogram data of the site, tissue characteristics of the site, or anatomical characteristics of the site;

[0017] determining an ordering factor according to the plurality of modifiers;

[0018] ordering each of the earliest activation electrode sites according to the ordering factor; and

[0019] outputting data identifying the region, the data varying the prominence of each of the determined earliest activation electrode sites according to the ordering.

[0020] Data that changes the prominence of each determined earliest activated electrode site according to the ordering can suppress the display of sites, highlight them in reports / spreadsheets, change the visual prominence of sites in graphical displays (which can overlap on views of the heart, etc.), as triggers for the scanning system to consider / ignore that area, etc.

[0021] Modifiers determined from electrogram data can include:

[0022] a minimum cycle length of electrograms recorded at the electrode, an activation frequency gradient between electrograms recorded at the electrode or in the region and electrograms obtained within a predetermined geodesic distance, and an average voltage of local electrograms recorded in the region.

[0023] Modifiers regarding tissue characteristics can include a measure of the presence and density of scar tissue determined by imaging.

[0024] Modifiers regarding tissue characteristics can include a measure of the tissue impedance at the site.

[0025] Modifiers can be determined from the outcome of treatment at another earliest activated electrode site.

[0026] Modifiers regarding tissue characteristics can be computed by reference to the ablation outcome of similar sites in prior cases.

[0027] Modifiers regarding anatomical features can include predetermined weighting factors that depend on the location of the electrode site.

[0028] The output data can include the data displayed with a visual indication that changes the prominence of each of the determined earliest activated electrode sites depending on the ordering.

[0029] The method can further include:

[0030] determining a primary electrogram wavefront trajectory for each electrode site;

[0031] determining a vector of primary derived electrogram activation across multiple electrodes and activations;

[0032] classifying an electrode site as an earliest activated electrode site and a potential driver site if the direction of activation travels from the electrode site towards one or more electrode sites proximal to the potential driver site; and

[0033] the order and timing of activation is within a predetermined biologically reasonable manner.

[0034] In one embodiment, if activation across at least 2 electrodes occurs from a vertex opposite the potential driver site to the 2 electrodes at an angle less than a predetermined angle, the site is classified as an earliest activation electrode site, and

[0035] At least one electrogram activation is determined to be later than the potential driver site within a defined arc of excited tissue.

[0036] The predetermined angle can be less than 180 degrees.

[0037] The method can further include outputting data to a display or medical scanning device to cause display or navigation of chamber geometry and to guide placement of a catheter or electrodes for subsequent electrogram data acquisition.

[0038] The method can further include identifying a site for subsequent placement of the catheter or electrodes to eliminate or confirm a previously determined first activation electrode, or to extend the electrogram data to an area not previously scanned or not fully scanned.

[0039] The method can further include identifying the first activation site from a location of the site that has been identified or from a gradient in activation vectors and signal lead scores.

[0040] The method can further include receiving electrogram data in substantially real time and providing an indication related to when sufficient acquisition timing has been performed on a site being scanned.

[0041] The step of determining whether sufficient acquisition timing has been performed can include determining that a pattern of activation of a site being scanned has reached a predetermined level of statistical certainty. For example, there is a 95% likelihood that the pattern is not random (this can be reached in a few seconds if the rhythm is very regular).

[0042] The step of determining whether sufficient acquisition timing has been performed can include one or more of:

[0043] Counting down a predetermined period of time (e.g., 30 seconds, with reference to the previous case), acquiring a predetermined number of activation cycles (e.g., 50 activation cycles) on at least a predetermined number of electrodes on a flow multipolar mapping catheter, and acquiring a predetermined time or a predetermined number of activation cycles, where the pattern of activation across multiple electrodes remains within a pattern-matched sequence.

[0044] The method further includes deriving a wavefront direction by determining a leading electrode in each electrode pair, and processing the wavefront direction to determine whether the earliest activation electrode is a true AFD or represents a passive activation site that was activated beyond a measurement range.

[0045] Advantageously, in embodiments of the application, additional information is utilized whereby the importance ranking of the earliest activated (leading) sites can be performed. In an irregularly complex rhythm, sites collected at different time points can be compared and used to verify whether they are truly leading. Other characteristics of the electrical activity of the atrial tissue region can also be used to help indicate the region that ablation will interrupt or slow AF. The combination of other measured factors improves the ability to indicate the region of the heart for which treatment will provide the best effect. Embodiments combine factors to modify, weight, or further rank the signal leading score resulting from such mapping to improve the selection of the ablation region and indicate the importance of the region for the persistence of arrhythmia. Furthermore, the combination of these factors can also be used to provide negative weightings indicating that these regions are less likely to contribute to the arrhythmia mechanism.

[0046] According to another aspect of the application, there is provided a computer implemented method for identifying one or more regions of a heart responsible for supporting or initiating an abnormal heart rhythm by analyzing electrograms collected from the heart, the computer implemented method using electrogram data recorded from a plurality of electrodes on a multi-polar cardiac catheter, the electrogram data obtained from a respective series of sensing locations on the heart over a recording time period, the method comprising the steps of:

[0047] defining a particular region within a chamber of the heart as a potential driver of the abnormal heart rhythm by analyzing the sequence of electrical activation,

[0048] weighting the classification of the region as a potential driver according to factors including:

[0049] the direction of wavefronts of activation during the same collection period travel from the potential driver site to one or more nearby electrodes compared to the driver site,

[0050] the sequence and timing of activation is within biologically reasonable means with respect to conduction velocity and path within reasonable activation sequence,

[0051] activation occurs across at least 2 electrodes from the potential driver site with an included angle to the vertex from the potential driver site to the 2 electrodes less than a predetermined angle, for example less than 180 degrees, and

[0052] at least one electrogram activation is determined to be later than the potential driver site within a defined excitable tissue arc, and

[0053] a further collection performed at an adjacent location opposite the missing arc of the first potential driver site does not confirm a potential driver site at a similar location; and

[0054] displaying the refined potential driver in a highlighted manner according to the weighting, the display in relation to a computer representation of the chamber of the heart.

[0055] According to another aspect of the application, there is provided a computer system for identifying one or more regions of the myocardium responsible for supporting or initiating an abnormal heart rhythm, the computer system using electrogram data recorded from a plurality of electrodes on a multi-polar cardiac catheter, the electrogram data being obtained from a respective series of sensing locations on the heart over a recording time period, the system comprising:

[0056] a processor;

[0057] a first memory for storing received electrogram data; and

[0058] a second memory storing program code which, when executed by the processor, causes the system to:

[0059] identify a region within a chamber of the heart having an electrical activation sequence from the electrograms, the electrical activation sequence characterising the region as a potential driver of the abnormal heart rhythm,

[0060] for each sensing location at or substantially surrounding the region, determine an earliest activation electrode site from a primary activation;

[0061] for each determined earliest activation electrode site:

[0062] calculate a value for each of a plurality of modifiers associated with the electrode site, the modifiers being determined from electrogram data for the site, tissue characteristics for the site or anatomical characteristics for the site;

[0063] determine a ranking factor from the plurality of modifiers;

[0064] rank each of the earliest activation electrode sites according to the ranking factor; and

[0065] output data identifying the region, the data varying the prominence of each of the determined earliest activation electrode sites according to the ranking.

[0066] The program code, when executed by the processor, can cause the system to determine modifiers from the electrogram data, including one or more of:

[0067] a minimum cycle length of the electrogram recorded at the electrode, an activation frequency gradient between the electrogram recorded at the electrode or within the region and an electrogram obtained within a predetermined geodesic distance, and an average voltage of a local electrogram recorded within the region.

[0068] The program code, when executed by the processor, can cause the system to determine a modifier in respect of tissue characteristics by obtaining a measure of the presence and density of scar tissue determined from imaging.

[0069] The program code, when executed by the processor, can cause the system to determine the modifier on the tissue characteristic by obtaining a measurement of tissue impedance at the site.

[0070] The program code, when executed by the processor, can cause the system to determine the modifier on the tissue characteristic by obtaining a treatment outcome at another earliest activated electrode site.

[0071] The program code, when executed by the processor, can cause the system to determine the modifier on the tissue characteristic by accessing data relating to ablation outcomes of similar sites in prior cases.

[0072] The program code, when executed by the processor, causes the system to determine the modifier by accessing data to obtain a predetermined weighting factor, the predetermined weighting factor depending on the location of the electrode site.

[0073] The program code, when executed by the processor, can cause the system to:

[0074] determine a primary electrogram wavefront trajectory for each electrode site;

[0075] determine a vector of primary derived electrogram activation across multiple electrodes and activations;

[0076] classify an electrode site as an earliest activated electrode site and a potential driver site if the direction of activation travels from the electrode site towards one or more electrode sites proximal to the potential driver site; and

[0077] the order and timing of activation is within a predetermined biologically reasonable manner.

[0078] The program code, when executed by the processor, can cause the system to output a visual indication, the visual indication varying in prominence of each of the determined earliest activated electrode sites depending on the ordering.

[0079] The program code, when executed by the processor, can cause the system to output data to a display or medical scanning device to cause display or navigation of chamber geometry and guide placement of a catheter or electrode for subsequent electrogram data acquisition.

[0080] Embodiments of the invention seek to provide a system and method that can be used to refine STAR methods and other similar mapping methods based on human body data, typically electrogram data and imaging data (e.g. tissue characteristics or anatomical data acquired from CT or MRI scans) are acquired in an electroanatomical mapping procedure, and to prioritize statistically identified potential AFDs in order of the following factors that can be included:

[0081] Period length measurement and comparison of the identified potential AFD, in particular the shortest cycle length, the lowest cycle length change compared to other identified AFD, the presence of a steep cycle length (CL) gradient around the AFD.

[0082] The steepest CL gradient or the presence of a CL gradient between the potential AFD and surrounding electrodes can be recorded in the same acquisition, but only in the case where the wavefront direction generated by the STAR mapping travels from the AFD to the paired electrode (to avoid high cycle length associated with wavefront collision.

[0083] If a wavefront propagates at at least two electrodes with an included angle of less than 108° (with the AFD opposite as the vertex), the wavefront is considered to be generated from the AFD site.

[0084] Other possible applicable criteria include:

[0085] If an AFD does not have a pairing of electrogram timing later than it within a 180° arc, it is assumed to be a peripheral site, whose activation can come from directions within the 180° arc rather than from the AFD, unless there is a scar within the 180° arc.

[0086] If an AFD has more than one electrode pair later than it, the AFD can only be labeled as such (and the spacing of these electrode pairs is large enough that they do not allow the AFD to create an empty arc of 180° arc in the case where the AFD is considered to be a peripheral device activation rather than a true AFD).

[0087] The example of 180 degrees is given because the activation site can be represented as coming from a direction within a defined arc, rather than from a potential driving site.

[0088] In a preferred embodiment, potential AFDs can further refine their relative importance by reference to modified characteristics from electrogram features or tissue features of the underlying tissue.

[0089] One embodiment relates to a system for analyzing electrograms collected from a heart to

[0090] define specific regions within a chamber of the heart having an electrical activation sequence that identifies the specific region as a potential driver of an abnormal heart rhythm, and

[0091] classify the importance of each defined region by electrogram features through the region, rather than activation sequence, anatomical or imaging characteristics, surface electrograms, and / or patient characteristics, which can include:

[0092] the shortest or minimum cycle length compared to all other early activation sites identified, and / or

[0093] If the direction of the activation sequence calculated is from the potential driver to the electrode compared to, further refinement can be made by considering only the gradient between the potential driver and the nearby electrodes recorded in the same set, and / or

[0094] The lowest cycle length variability compared to all other early activation sites identified, and / or

[0095] The steepest cycle length gradient or presence of cycle length gradient between one early activation site and one or more electrodes in the same set within a defined geodesic distance of the potential site of interest, e.g. < 3 cm for irregular rhythm; and / or

[0096] Evaluation of the geodesic distance of the site of interest in the unipolar and / or bipolar electrogram voltage abnormality region, and / or

[0097] Evaluation of the frequency analysis or frequency gradient analysis, e.g. dominant frequency of the site of interest and other sites within a defined geodesic distance, and / or

[0098] Evaluation of anatomical and structural features identified by imaging methods (e.g. intracardiac echocardiography, cardiac magnetic resonance imaging, cardiac computed tomography), e.g.:

[0099] Near pulmonary vein or atrial appendage tissue

[0100] Interface of appendage and vein

[0101] Marshall vein

[0102] Mitral valve annulus

[0103] Modulating chordae

[0104] Aneurysm

[0105] Identified scar

[0106] Change in tissue thickness

[0107] And / or

[0108] Evaluation of tissue electrical impedance, and / or

[0109] Evaluation of tissue motion or thickening directly by imaging or reference to motion of a cardiac catheter in contact with cardiac tissue, and / or

[0110] Evaluation of correlation between the identified site of interest and sites identified as important by other cardiac rhythm mapping systems, e.g. giving higher importance to a region of interest identified by two or more methods, which region can be on the same cardiac geometry or within the same anatomical region, which anatomical region is on two or more geometries created by different cardiac mapping systems; and

[0111] application weights further classify the likelihood of a beneficial effect of intervention at these sites, which can be determined by reference to previously acquired data and response to ablation,

[0112] wherein machine learning techniques can be used to calculate and optimize the weighting factors to be used, and,

[0113] displaying the classification data of the computed importance of these sites with a visual indication.

[0114] In one embodiment, a system for analyzing electrograms acquired from a heart is disclosed. The system can define a specific region whose electrode activation generally precedes electrode activation within a specific geodesic distance and compute a time proportion or signal lead score for each identified location, and

[0115] modifying the lead score by applying an adjustment factor calculated from electrogram, anatomical or imaging characteristics, surface electrograms, and / or other characteristics of the patient.

[0116] The characteristics of the mean wavefront direction or vector (from determining the leading electrode in each pair, the STAR map) can be used to confirm whether an electrode identified as a potential AFD is indeed a true AFD. Electrodes on the edge of a region that are mapped and identified as potential AFDs in one acquisition can represent passive activation sites that themselves activate from outside the measured boundary. If this potential AFD site can be passively activated, its significance can be reduced.

[0117] Embodiments can be used to record data (highlighting potential problematic or confirmed AFD sites). Alternatively, embodiments can operate in substantially real time to highlight or instruct the user to move the electrode mapping catheter to regions most likely to cover AFDs, or instruct them to move away from regions less likely to cover AFDs (i.e. introduction of a system that uses these characteristics to guide mapping catheter movement).

[0118] In one embodiment, a system for recording and analyzing electrograms acquired from a heart is disclosed. The system can define a specific region within a chamber of a heart having an electrical activation sequence that identifies the specific region as a potential driver of an abnormal heart rhythm, where apparent potential heart rhythm abnormality driver sites acquired by continuous multipolar mapping can be further classified by separate rules to determine whether a driver site identified in a single mapping acquisition is a true source of a heart rhythm disorder:

[0119] wherein an electrode site is classified as a potential driver only when the direction of a wavefront produced by an activation travels from the potential driver to one or more electrodes compared to a driver site during the same acquisition; and

[0120] An electrode site is classified as a potential driver only if the activation sequence and timing is biologically reasonable, the reference conduction velocity and pathway is within a reasonable activation sequence, and at least 2 electrode angles are less than a defined angle (e.g. the angle of the potential driver site as the apex is less than 180 degrees);

[0121] And where, if an electrode site has no electrogram timing pairings later than it within a defined arc of excitable tissue, e.g. 180 degrees (as such an activation site can be represented as a direction from within that defined arc rather than from a potential driver site), the electrode site can be excluded as a potential driver, i.e. classified as a passive activation site, and further acquisition performed at an adjacent location to the absent arc pair of the first potential driver site fails to confirm a potential driver site at a similar location, and where the identified scar tissue is classified as non-excitable and considered to have no contribution to the activation arc.

[0122] Further modifications and modifications of the weighting factor can depend on electrogram characteristics of the region (in addition to the activation sequence), anatomical characteristics of the potential driver site, imaging characteristics around the potential driver site, characteristics of the surface electrogram, and / or patient characteristics. Typical electrical characteristics considered as the weighting factor can include:

[0123] If the direction of the calculated activation sequence travels from the potential driver to the electrode being compared, the shortest or minimum cycle length compared to all other early activation sites identified can be further refined by considering only the gradient between the potential driver and nearby electrodes recorded in the same acquisition.

[0124] In addition, other electrical characteristics of the electrogram recorded at the potential driver site compared to all other early activation sites identified can be considered relevant and their quantification used as an input to the modification factor (including the minimum cycle length variability), and / or

[0125] The steepest cycle length gradient or presence of a cycle length gradient between one early activation site and one or more electrodes in the same acquisition within a defined geodesic distance of the potential site of interest; and / or

[0126] Evaluation of the geodesic distance to the site of interest in a region of abnormal monopolar and / or bipolar electrogram voltage, and / or

[0127] Evaluation of a frequency analysis or frequency gradient analysis, e.g. dominant frequency of the site of interest and other sites within a defined geodesic distance, and / or

[0128] Measurement of tissue electrical impedance at the driver site.

[0129] Any or combination of the above can be used to further determine the likelihood that a particular site contributes to the occurrence of arrhythmia, and direct quantification of such sites can be used as a modifier.

[0130] Furthermore, assessment of anatomical and structural features can be performed using imaging methods such as intracardiac echocardiography, cardiac magnetic resonance imaging or cardiac computed tomography, and one or more resulting quantitative metrics used as inputs for the modifier.

[0131] Anatomical location itself can also be used with factors based on proximity to anatomical structures, such as computed geodesic distance to any one or more defined anatomical points, such as:

[0132] Proximity to pulmonary vein or atrial appendage tissue

[0133] Interface of appendage and vein

[0134] Marshall vein

[0135] Mitral annulus

[0136] Modulating chordae

[0137] Aneurysm

[0138] Recognized scar

[0139] Changes in tissue thickness

[0140] Dynamic assessment of tissue motion or thickening, whether directly with imaging or by reference to motion of a cardiac catheter in contact with cardiac tissue, can be used as a modifier to further refine the importance of a site.

[0141] In one embodiment, a computer-implemented method for analyzing electrograms acquired from a heart is operated to identify, from an electrogram region within a chamber of the heart, regions having an electrical activation sequence that characterizes the region as a potential driver of an abnormal heart rhythm, and

[0142] ordering the importance of each of the defined regions by applying an ordering factor to each identified region, wherein the ordering factor is a product of a plurality of normalized factors and is influenced by a predetermined weighting factor, including

[0143] a minimum cycle length of the electrogram recorded by the electrode, once extreme outliers are excluded, and

[0144] a gradient of activation frequency between the electrogram recorded at the electrode or the region and an electrogram obtained within a predetermined geodesic distance, and

[0145] an average voltage of the local electrogram recorded within the region, and

[0146] Data showing a visual indication of the computed importance classification of these sites.

[0147] In another embodiment, a system for analyzing electrograms acquired from a human or animal heart comprises a computer processor configured to execute computer program code to:

[0148] identify an activation sequence of electrical activity from the electrograms to define a specific region within a chamber of the heart having an electrical activation sequence that characterizes the region as a potential driver of abnormal heart rhythm, and

[0149] rank the importance of each of the defined regions by applying a ranking factor to each identified region,

[0150] wherein the ranking factor is a product of a plurality of normalization factors and is influenced by a predetermined weighting factor, wherein the ranking factor is computed from factors selected from a group of factors comprising:

[0151] the minimum cycle length of the electrogram recorded by the electrode, and

[0152] an activation frequency gradient between the electrogram recorded at the electrode or the region and an electrogram obtained within a predetermined geodesic distance, and

[0153] an average voltage of the local electrogram recorded within the region, and

[0154] and data displayed with a visual indication of the classification of these sites representing chambers of the heart.

[0155] Multiple cardiac mapping systems can be used at one time, and the correlation between the identified sites of interest and sites identified as important by other cardiac rhythm mapping systems can be used as a modification factor to give higher importance to regions of interest identified by two or more methods, which can be on the same cardiac geometry or in the same anatomical region on two or more geometries created by different cardiac mapping systems.

[0156] Each defined factor can be influenced by a weighting factor designed to provide a measure of the likelihood of a beneficial effect of an intervention at these sites. These weighting factors can be determined by reference to previously acquired data, ablation responses of previous patients, and / or by reference to ablation responses of the same patient in a study. Machine learning techniques can be used to compute and optimize the weighting factors to use and the final sites, which are displayed with a visual indication of the final computed importance classification of each site, such as by a color scale, percentage likelihood of arrhythmia termination, or beneficial effect or importance ranking.

[0157] Embodiments of the present invention provide systems and methods for improving CL analysis, exhibiting faster rates and organization of sites can improve the sensitivity and specificity of identification of driving sites in AF, such as by identifying cardiac regions that activate prior to adjacent regions. In one embodiment, the STAR mapping method is employed, but there are other methods that can identify potential driving sites. Potential AFDs can be identified by statistical methods or other commercial methods (e.g. Cartofinder [Biosense Webster, Haifa, Israel], Acutus, ECGi [e.g. Cardioinsight, Medtronic, Ireland], Ablacon, Topera mapping [Abbott Ltd., Mn, USA]) that identify the local region that activates earliest. It will be appreciated that the electrogram features applicable to the atrium (e.g. AFDs) can also be applicable to other heart rhythms and chambers, such as ventricular fibrillation or tachycardia.

[0158] In experimental trials, the STAR method has been used to identify potential AFDs, and confirm that the AFDs are the potential AFDs that respond to ablation of those sites either with a slowing of the AF cycle length by >30ms or AF termination. However, it will be appreciated that other factors can also be used to more accurately identify AFDs.

[0159] Furthermore, testing has shown that the driving sites identified using the STAR method exhibit faster rates and organization in maintaining AF, which is mechanically more important, with the likelihood of ablation terminating AF being greater.

[0160] There are several novel methods that can be used as weighting factors that can be used to modify the signal lead score (ranking factor) that can be used in embodiments of the present invention. These can be subdivided into modifiers related to individual electrodes, modifiers related to the relationship between individual electrodes and surrounding electrodes, modifiers related to specific locations and individual patient specific data that are not derived from electroanatomical or physiological, location specific modifiers of outcome characteristics from patient groups and general modifiers, which can include demographic data.

[0161] Embodiments of the present invention include an apparatus and related computer implemented method that seeks to improve the effectiveness of identifying potential arrhythmia driving sites from electroanatomical mapping, allowing them to be better classified and determining their importance. The resulting data can be displayed to a physician or otherwise output or communicated to other systems. It can be used during a cardiac catheter procedure, such as a catheter ablation procedure (immediately or a period of time later), to highlight where ablation will produce the most beneficial effect. Such data can be used to enable targeting of non-invasive therapies, such as radiotherapy, gamma knife, proton beam therapy.

[0162] One potential use of embodiments of the present application is to treat patients who have been diagnosed with persistent atrial fibrillation and whose ablation treatment consisting of pulmonary vein isolation has not resulted in complete cessation of their arrhythmia. In these patients, other mapping procedures such as STAR mapping can be used to better target further ablation and this procedure will be improved by embodiments of the present application. BRIEF DESCRIPTION OF DRAWINGS

[0163] Embodiments of the present application will now be described, by way of example only, with reference to the accompanying drawings in which:

[0164] Figures la-d is a schematic diagram illustrating aspects of determining a driving site in embodiments of the present application;

[0165] Figures 2a-e is a schematic diagram illustrating aspects of further determining a driving site in embodiments of the present application;

[0166] Figures 3a-f is a diagram illustrating determination and use of a maximum spread angle in embodiments of the present application;

[0167] Figures 4a-b is a schematic diagram illustrating operation of aspects of embodiments of the present application;

[0168] Figure 5 is a cycle length (CL) histogram obtained at basket catheter electrodes;

[0169] Figure 6 is a flow diagram illustrating a study to identify potential AFDs;

[0170] Figure 7 A-F of Figure 1 are images and electrograms obtained from a subject;

[0171] Figure 8 A-C of Figure 2 are images and electrograms obtained from a subject; and

[0172] Figure 9 A-Cii of Figure 3 are images and electrograms obtained from a subject. DETAILED DESCRIPTION

[0173] In the embodiments described below, the system can be used in conjunction with a full chamber basket catheter (Constellation catheter, Boston Scientific, ltd, US and FIRMap catheter, Abbott, US) to allow for simultaneous panoramic left atrial (LA) mapping. However, other suitable catheters / electrodes can be used to obtain electrogram data without necessarily collecting data from the entire area of interest at the same time. For example, the area of interest can be divided into smaller areas, the electrogram data collected sequentially before analysis. The results of the analysis can then be combined and displayed in a single STAR map.

[0174] The system for acquiring and processing electrogram data typically comprises one or more multipolar catheters inserted into the heart chambers of a patient (e.g. the basket catheters mentioned above, as well as other intracardiac catheters, such as a decapolar catheter placed in the coronary sinus), an amplifier and analog-to-digital (AD) converter, a control console comprising a signal analyzer, a processor and a GPU, a display unit, a control computer unit, a system for determining and integrating 3D position information of the electrodes.

[0175] The system can use known hardware and software, such as the Carto™ system (Biosense Webster, J&J), NavX Precision™ (Abbott Medical) or Rhythmia™ (Boston Scientific) for catheters, 3D electroanatomical integration and processing units. Catheters such as "basket" catheters, circular or multi-strip mapping catheters (e.g. Lasso catheter, Biosense Webster, J&J, HD-mapping catheter, Abbott Medical SJM, Pentarray Biosense Webster, J&J), decapolar catheters or catheters can be used. These systems and catheters are used to collect electrogram signals and corresponding position and time data relating to electrical activity at different locations within the heart chambers. These data are passed to a processing unit which performs algorithmic calculations on these data and aims to transform these data to provide the physician with location information of the areas within the heart most likely to cause abnormal heart rhythms to persist and continue. Alternatively, the system can use custom catheters, tracking systems, signal amplifiers, control units, computing systems and displays.

[0176] In the above patent application, a mapping system is described, hereinafter referred to as the "Stochastic Trajectory Analysis of Ranked Signals (STAR) mapping" system. Its aim is to identify the driving site of arrhythmias, which can be displayed, for example, in the form of a 3D map. The mapping created using the STAR mapping system is referred to as "STAR map".

[0177] When performing STAR mapping (and collecting data for embodiments of the present invention), the physician can: place the multi-polar panoramic mapping catheter in the lateral left atrium, collect data for a period of time (e.g., 5 seconds to 5 minutes), then reposition the catheter to ensure tight adherence to the septal or anterior wall of the left atrium and perform another recording. It will be appreciated that this can be performed in advance and the pre-recorded data processed.

[0178] The proportion of "leading" electrodes will be correlated between mappings, so the data and proportions will be able to be displayed without issue on the same mapping. In this way, multiple correlated histograms can be built in sequence by moving the catheter within the chamber and taking further recordings.

[0179] The following are the main steps in the process of identifying "leading" signals (which indicate regions / sites that are statistically likely to drive abnormal rhythms in the heart) after collecting electrogram data (and corresponding spatial and temporal data).

[0180] First, interference and far-field signal components are removed from the input electrogram signals. In one embodiment, the system decomposes the signals into relevant components, for example, by spectral analysis, far-field signal blanking, far-field signal subtraction, filtering, or by another method known in the art. Signal components from within the chamber of interest in the heart (e.g., atrial signals) are identified. The relative timing of the atrial signals can be established in a deterministic, stochastic, or probabilistic manner. In some embodiments, the phase of each signal can be determined and the relative timing established from the relative phase shift between different electrodes.

[0181] Second, the signal timing from adjacent electrodes is paired. Signals are paired with each other only if the electrodes are within a specified geodesic distance of each other, i.e., only electrodes that are close to each other are paired with each other. This can be further improved by pairing only electrodes that are on the same side of the chamber wall; i.e., adjacent electrodes on the posterior wall of the heart will be considered adjacent, but not if the electrodes are on discontinuous opposite sides, e.g., the pulmonary veins, even if the absolute distance between these electrodes can be small. The relative timing of activation at paired electrodes is thus established, and a value assigned to "leading" electrodes. This pairing can be performed for discrete analysis time periods, typically lasting between 10 milliseconds and 200 milliseconds. The length of the analysis time period need not remain constant throughout all of the data being analyzed. The goal is to compare the timing between paired electrode activations caused by the same activation sequence, and the analysis time period can be determined accordingly. For example, each analysis time period can be selected to contain electrode activations likely to result from the same activation sequence. Thus, the analysis time periods can overlap each other.

[0182] Third, this process is repeated multiple times over a given time period (i.e., for multiple many analysis time periods, for each pairing). Advantageously, the analysis time periods overlap, and are offset relative to the initial analysis time period, e.g., 10 to 120 seconds. As noted above, the analysis time periods can overlap within a given time period. For example, if the analysis time period is 200 ms, the analysis time periods can overlap by 100 ms, i.e., 50%. In other words, the leading electrode is determined for the first 200 ms time period, then for the second 200 ms time period, the second time period begins 100 ms after the first time period begins, and so on, for the given time period. As with the analysis time periods themselves, the degree of overlap can vary from data set to data set. Through this repeated analysis, activation sequences that are less frequently repeated, or not repeated at all, can be discarded, and activation sequences that are more frequently repeated can be ranked with higher importance and priority.

[0183] In atrial fibrillation, the activation pattern appears chaotic with frequent changes in the wavefront propagation. The relative proportion of the "time" each site precedes each of its neighbors in the activation map is calculated, thus creating a proportional map of the more frequently "leading" electrode sites.

[0184] Fourth, the proportion of "time" each recorded region "leading" activation takes is calculated. This calculation can be based on determining the actual duration of time each electrode leads its paired electrode. Alternatively, it can be the proportion of the total analysis time period that a site leads in the mapped activation, whether those time periods are of the same duration as each other or not. Thus, in some examples, the relative proportion is actually determined by looking at the total number of atrial activation signals seen by a given electrode, and determining the proportion of those activation signals that the electrode leads relative to a number of other electrodes that are paired with each other. Although mapping only adjacent electrodes can lead to errors, the system will map every electrode against every other electrode for each activation cycle to establish activation directions within the mapping field. The order is analyzed to identify the primary activation sequence and the sites that caused those activations (i.e., the point from which the activation emanated) during the recording period. The activation sequence with a trajectory representing a local source and AFD is considered to be mechanically important. The STAR mapping system calculates the proportion of activation sequences to a given vector to establish a dominant vector, if any, and calculates the proportion of activations that follow that vector. For all sites in the mapping field, the proportion of mapped activations from that site is calculated to determine its relative importance. These proportions can be referred to as "leading signal scores," and allow for further modification, comparison, and calculation.

[0185] The leading signal scores are normalized so that the proportions can be compared across the entire heart. Many embodiments of statistical processes, normalization, and subsequent display of this data can be envisioned.

[0186] Electrodes that are completely overlaid on a passive region of cardiac activation will tend to have regions that emit little activation, if any. Only activation sites that are frequently "leading" in the activation sequence will be assigned a value that indicates a high likelihood of being a source of activation. Similarly, when considering overlapping electrode sampling locations A and B, and the earliest repeated activation is at the edge of sample A, it can be seen that activation travels from B to A, so the early activation site of A can be considered passive, and the electrode from sample B is considered leading, and thus the leading electrode from sample B is more emphasized.

[0187] This process can be repeated over multiple recording time periods and locations to further refine and define repeating and discarding non-repeating activation patterns, to build a more extensive mapping area than that resulting from single activations recorded from a multi-polar electrode catheter.

[0188] The STAR mapping method can be used in embodiments to identify, for a region of a chamber of the heart having an electrical activation sequence that characterizes the region as a potential driver of an abnormal heart rhythm, the primarily earliest activating electrode sites. Alternatively, methods such as the CARTOFINDER™ module running on the Carto™ system (Biosense Webster, J&J) can be used to indicate the proportion of time that an electrode location activates before its neighbors. This information can be normalized to form a signal lead fraction that can be used in the described methods to provide improved information about the importance of these sites.

[0189] In one embodiment, a computer-implemented method is used to identify one or more regions responsible for supporting or initiating an abnormal heart rhythm in a heart. The computer-implemented method uses electrogram data recorded from a plurality of electrodes of a multi-electrode array on a multi-polar cardiac catheter from a respective series of sensing locations on the heart over a recording time period. The method comprises the steps of:

[0190] identifying, from the electrograms, a region of a chamber of the heart having an electrical activation sequence that characterizes the region as a potential driver of an abnormal heart rhythm,

[0191] for each sensing location at or approximately surrounding the region,

[0192] for each determined earliest activating electrode site

[0193] for each determined earliest activating electrode site:

[0194] calculating a value for each of a plurality of modifiers associated with the electrode site, the modifiers being determined from electrogram data of the site, tissue characteristics of the site, or anatomical characteristics of the site;

[0195] Determining a ranking factor computed by a plurality of modifiers

[0196] Ranking each of the earliest activated electrode sites according to the ranking factor; and

[0197] Outputting data identifying the region, the data varying the prominence of each of the determined earliest activated electrode sites depending on the ranking. This is discussed in the examples below.

[0198] Figures la-d is a schematic diagram showing aspects of the determination of a drive site (the earliest activated electrode site corresponding to a potential driver of an abnormal rhythm).

[0199] Figure la is a schematic diagram showing a human heart atrium. The diagram shows 3 activation sites driving an atrial fibrillation, represented by circles A, B and C. These can be identified as drive sites by a variety of methods, such as stochastic trajectory analysis of the rank signal (STAR mapping). In sequential mapping of arrhythmic wavefronts, the electrogram wavefront trajectory is determined by using a multi-electrode array, represented by the large circle MEA, which encompasses a number of individual electrode sites including electrode el. The derived electrogram activation vector is derived across the MEA, represented here by the arrowed line. In this example, the earliest activated site is determined to be el. When the MEA is moved to a new position, as shown in the picture of Figure lc , the electrode site covering the original site el is no longer the leading site representative of the drive site characteristics, electrode site el' is in fact generally leading in activation.

[0200] In embodiments of the invention, when an electrode site on the edge of the electrode array appears leading in one acquisition but not leading in another acquisition overlapping the leading site in the first acquisition, the site is classified as a driver.

[0201] Now consider site el', which is represented as leading in the picture of Figure lc , and in fact covers the true drive site (site A in Figure la ). Here, in the picture of Figure ld , the performance of a third acquisition is represented, but in this case, the electrode covering site el' (e2') continues to have the characteristics of a leading electrode's signal, while the edge electrode site el" now moves to the following region. The system confirms that the site at el' (and e2') is now definitively a drive site.

[0202] Performing successive acquisitions of electrograms around the heart chamber in this way can be used to distinguish the true drive sites from apparent drive sites that are passively activated due to conduction from a remote drive site.

[0203] Those skilled in the art will appreciate that once all sites have been mapped, it is possible to perform this sequential mapping and determination of driving sites dynamically during mapping or any intermediate point. For example, a map can be generated from the data collected, the locations of the determined potential driving sites, and potential driving sites that need to be classified as potentially driving (or not) in the region can be highlighted in the map presented to the user or physician by further mapping or electrogram collection. It is envisaged that such a system is employed to guide a robotically controlled catheter towards and around potential driving sites for mapping.

[0204] Figure 2a is a diagram of a heart chamber (here, the left atrium) with two independent driving sites denoted by stars Al and A2. The general vector of the heart activation wavefront is denoted by the shaded arrow.

[0205] In embodiments of the invention, as shown in Figure 2b a multi-polar mapping catheter with electrodes el to e10 is placed over the AF driver Al. Electrode e2 is activated first, and the activation vector is generally away from this site. The resulting vectors are all shown pointing towards the e2 electrode with the highest leading signal score (for clarity, only arrows from e2, e3, e5 and e9 are shown). As a result, e2 is highlighted to the user.

[0206] As shown in Figure 2c the multi-polar mapping catheter is then moved to a site away from both activation sources. The "leading signal score" will now be highest on the peripheral electrode e9'. Activation is generally in one direction, and generally the peripheral electrode e9' leads all other electrodes.

[0207] Figure 2d A simplified concatenated mapping is shown, as in STAR, with the electrode with the highest signal leading score in each successive acquisition highlighted. The two sites representing AFDs are now shown as outlines rather than solid shapes. In this example, two electrodes e2 and e9' are highlighted. It is clear that although electrode e9' is attributed to a high leading signal score, this does not indicate that this electrode is indicative of the true location of the AFD, unlike e2. The described invention modifies the display of this site by identifying the site according to the composite vector towards which the sites of the periphery are acquired, where the activation order of the adjacent acquisitions is consistent (i.e. in approximately the same direction / away from site e9'). Experimental observations also indicate that these sites exhibit differences in other metrics compared to the true AFD, such as activation frequency, electrogram voltage and dominant frequency.

[0208] In Figure 2eThe diagram shows the mapping obtained from the embodiment. Although the signal lead score is significantly high, e9' has been identified as unlikely to be an AFD site. Therefore, the visualization of the electrode is modified by reducing the projection size, reducing color highlights, or similar methods. Site e2 can be further highlighted by increasing color intensity, size, or other visual modifications.

[0209] In a preferred embodiment, a graphical representation of the mapping of the driving site is displayed on the display unit.

[0210] In one embodiment, the relative importance of identified potential AFDs can be further refined by referencing specific modified features from other criteria (e.g., anatomy) derived from electrogram features, tissue features, or underlying tissue.

[0211] In this embodiment, electrocardiogram acquisition at different electrode sites throughout the cardiac chamber is performed in groups (g 1- Wait until g x The process is performed sequentially. For example, the electrodes on the pentagonal mapping catheter (e...) 1 to e 20 It can be moved to different sites in the left atrium of a patient with atrial fibrillation and perform different sequential electrocardiogram acquisitions at each site.

[0212] Based on the electrogram dataset acquired in each sequence, the signal lead fraction for each electrode within the group is calculated. This can be indicated, for example, by the STAR method or another statistical method, by the proportion of time that activation seen at that electrode precedes activation detected at other electrode sites acquired simultaneously.

[0213] Alternatively, other methods can provide the activation rate prior to activation at other sites, or the rate at which that particular electrode is considered the leading electrode during acquisition, or a combination of indicators.

[0214] Consider the case where there are 5 electrodes (e1-e5) on a single acquisition (g1):

[0215]

[0216] For each of these sites, one or more other factors can be calculated, such as the change in cycle length at each electrode or the minimum average cycle length.

[0217] The cycle length index is calculated by initially identifying all activations on each electrode during acquisition.

[0218] To calculate the minimum stable cycle length (Min-CL), the initial activation-to-activation coupling intervals are filtered to remove any activations that are faster or slower than the setpoint around the average cycle length. For example, activations that are faster or slower than 30% of the average CL are ignored. The shortest of the remaining cycle lengths can represent unstable activations, so the 10% shortest activations are excluded. The assignment of Min-CL is the minimum of the remaining values, equivalent to the shortest cycle length in the 9th decile of the sorted activation intervals, where the longest cycle length is in the first decile, and so on.

[0219] For example, each electrode site of each consecutive acquisition (here g 1 -g 3 has a calculated Min-CL value (in ms).

[0220]

[0221] Min-max feature scaling normalization can be performed on all sites, allowing min-CL to be compared between individual acquisitions. This normalization is performed on the range of values of the acquisition in the case of modifying the variable Min-CL. In this example, the absolute value of Min-CL is normalized to the change in that value from the maximum of all Min-CL from all electrodes and acquisitions.

[0222] For example for g2e2,

[0223] abs(187ms - 201ms) = 14ms;

[0224] and

[0225] Change in CL = 36ms;

[0226] So

[0227] Normalized min-CL for g2e2 = 14 / 36 = 0.39

[0228] This value can be multiplied by the original signal lead fraction for that electrode (here 0.2) to give the modified signal lead fraction, i.e. 0.2 * 0.39 = 0.05

[0229] At this point, all modified signal lead fractions can be normalized again.

[0230]

[0231] In this example, sites g1e3 and g3e5 will be the two sites with the highest fraction. In contrast, the lead fraction for site g2e5 is reduced because the min-CL value here is much higher than for the other sites.

[0232] The modified variable can be weighted prior to standardization.

[0233] It can be appreciated that this method of providing weighting to the electrode lead signal score derived from STAR mapping or other electroanatomical mapping methods can be applied to other electrophysiological phenomena, such as cycle length variation, bipolar or unipolar voltage signals at the site, electrogram duration, or stability of activation of other electrodes (i.e. low CL variability on other electrodes).

[0234] Further modified variables can be applied from look-up tables, for example in relation to anatomical sites which can themselves be manually or automatically tagged. Alternatively, they can be calculated from anatomical or tissue measurements derived from MRI or CT scans.

[0235] Preferably, the geometric site corresponding to the location of the highest ranked, normalized modified signal lead score site is highlighted, for example by increasing or changing the size, opacity, color or marker of the region or site on the surface of the cardiac chamber presented in the data output.

[0236] In a preferred embodiment which can be used in addition to or as an alternative to the above, it is determined whether a potential AFD is truly a true AFD by reference to the activation vector determined by, for example, STAR mapping.

[0237] As previously described, the electrograms are preferably acquired sequentially in groups (g 1 etc. to g x ) of different electrode sites throughout the cardiac chamber. For each simultaneous group acquisition of electrograms, an approximate vector of activation is derived by reference to the signal lead score (unmodified) or according to other methods (e.g. wavefront direction calculation). Electrodes with high signal lead scores surrounded by electrodes with lower signal lead scores are likely to represent sites of true AFD. However, electrodes on the periphery of the acquisition geometry and with high signal lead scores can represent true AFD or simply the closest passive activation site to the AFD in that particular acquisition. Advantageously, it is indicated automatically whether an acquisition is likely to be a passive activation site.

[0238] To determine whether a lead electrode site (e.g. el) in an acquisition is likely to be a true AFD, the electrode with the highest signal lead score in the group acquisition is identified. A tangent plane to the geometric surface of the chamber is created at the midpoint of the electrogram acquisition sites. It is advantageous to use a highly smooth representation of the chamber geometry to prevent in-plane warping. The in-plane angular distribution from the arbitrary bisector of el is then calculated and the in-plane angular distribution of electrodes around the point (spread angle) is determined. No angular separation between surrounding electrodes greater than a predetermined angle (e.g. 120°) indicates that el is likely to be a true AFD.

[0239] Thus, a modification factor, either determined from a look-up table or as a preselected value, can be applied to the leading electrode site identified as the AFD. Alternatively, the maximum angular separation between electrodes at the time of acquisition can be inversely rescaled (1 - (min-max normalisation)) and used as a variable for the modification.

[0240] For example, consider the case where e2 is surrounded by 4 electrodes at the time of acquisition g1, and the consecutive angular separations of these electrodes are 93°, 65°, 119° and 83°. Thus, the maximum angle for this example acquisition is 119°. The maximum spread angle taken is >170°, meaning that the potential AFD site lies at the edge of its electrode group.

[0241] In this example, the other three acquisition sites (g2 to g4) provide a maximum angular separation of 174°, 120° and 240° for each acquisition.

[0242] These angles can be normalised to provide a modification factor which can be applied to each most likely AFD seen at the time of each acquisition by weighting. In this example, 1, 0.55, 0.99 and 0 can come from a standard max-min normalisation. Using the maximum possible separation (360°) rather than the measured maximum angular separation to provide a non-zero normalisation modification factor, in this example: 1, 0.77, 0.99 and 0.49.

[0243] In the case where the maximum spread angle of a potential AFD site is greater than 170°, then it must be determined whether the site is passively activated. This can be performed by reference to other acquisitions which occur within a predetermined geodesic distance of the potential AFD and within the maximum spread angle arc.

[0244] Determining passive activation

[0245] By reference to the activation order or average wavefront vector of the individual acquisitions close to the site covered within the maximum spread angle arc, the electrodes of a potential AFD site lying on the periphery of a simultaneous acquisition can be classified as either a possible true AFD or a passively activated site. Preferably, these are within a certain geodesic distance of the AFD site in question, typically within a pre-set distance, for example 3cm.

[0246] For example, an electrode collection g1 e1 is located at the periphery of the collection and is identified as a potential AFD. Further collection g2 is performed within 3 cm and within the maximum spread angle arc and the calculation is applied. If g1 e1 is passively activated, then the expected signal lead fraction on the nearest electrode signal on g2 will be low. This can define a signal lead fraction on one (or more) nearest electrodes of g2 (e.g. g2 e1) that is 50% lower than all signal lead fractions of g2. Furthermore, the activation vector of these electrodes will be directed away from g1 e1, i.e. lie on the tangent to the plane, e.g. between 90º and 270º, with the vector from g2 e1 to g1 e1 taken as 0º.

[0247] If both conditions are met, g1 e1 can be defined as an absolute passive activation site. A modification factor can be applied to reduce its signal lead fraction, thereby modifying by reducing the prominence of g1 e1 in the calculation. This modification factor can be arbitrary according to a look-up table or previous investigations, and can be applied to a single electrode site (g1 e1) or to all electrodes across sequential collections (e.g. g1).

[0248] An iterative process can then be performed in which the signal lead fractions are recalculated after redistribution and normalisation of the signal lead fractions.

[0249] However, if g2 e1 and / or its neighbouring signal lead fractions are high (e.g. within the top 75% of the lead signal fractions of the activation g2), and the activation vector is directed towards the site g1 e1 (e.g. between -90º and +90º on the tangent-to-plane electrode vector), the true AFD driver can lie at or very close to both electrode sites. A positive modification factor can be applied, and the area between g2 e1 and g1 e1 can be highlighted visually.

[0250] There can be a situation where an electrode with a high signal lead fraction is located at the edge of the collection and is not collected within the maximum spread angle arc and the defined geodesic distance. In this case, the area can be highlighted on the display screen to indicate that further collection should be performed in this area. This can be by highlighting with a different colour or by indicating with arrows, pointers or other animations the area where further collection should be performed.

[0251] During such a collection, an indicator can be displayed that indicates whether sufficient electrogram data has been collected. Most simply, this can be a simple timer, but other indicators can count the number of electrogram activations on each electrode and preset a target number by reference to a predetermined value.

[0252] Figures 3a-e is a diagram showing the determination and use of the maximum spread angle in an embodiment of the invention.

[0253] Figure 3a A multi-polar mapping catheter is shown with 6 electrodes labeled el to e6. In this example, electrode e6 has a high signal lead fraction than the other electrodes. The activation vectors between this electrode and the surrounding electrodes are drawn by solid arrows, which project on a tangent to the smoothed heart chamber surface representation at e6. There is an angle between these vectors, for example the two vectors from e6 to el and from e6 to e5, the angle of which is given by 0i-5here. In this example, e6 is surrounded by electrodes, and the largest angle 0s-6, which can be called the maximum spread angle, is less than a preset maximum, for example 120°. This confirms e6 as an AFD, and a modification factor can be applied.

[0254] Figure 3b The same electrode arrangement is now shown moved to another position. Here the potential AFD, the electrode with the highest signal lead fraction, is el. However, in this example, the largest inter-electrode angle is now 0 2-5, which is greater than a set amount (e.g. 170°), confirming this as an activated peripheral point.

[0255] Figure 3c An arc set to a diameter (e.g. 3 cm) and centered at the maximum spread angle (0 2-5 in this example) can be considered as shown by the shaded area. This can cover the angle 0 2-5 completely or can be smaller and centered at it. Then the closest two or more electrode activations from separate group acquisitions located within this arc are considered.

[0256] As Figure 3d shown, then the closest two or more electrode activations from separate group acquisitions (g2el-g2e6) located within this arc are considered. In this example, the two closest acquisitions g2e4 and g2e3 both have low signal lead fraction, and are not potential AFDs. Furthermore, the average activation vector, represented by the thick solid arrow, is directed towards el (rather than away from el, which can be expected if el were a true AFD). This confirms that el in this example is not an AFD, and a modification factor can be applied to it to reduce its visual prominence.

[0257] Figure 3e Another situation is shown, where one of the closest electrodes in a further acquisition (g3e4) has a high signal lead fraction, and the vector from this electrode is directed opposite to the vector from el. This indicates that there is indeed an AFD in the vicinity of these two electrodes, and the area between them can be highlighted (e.g. here with a striped circle).

[0258] Figure 4a and Figure 4bIt is shown how data from other electrogram analysis or from imaging techniques can be used in conjunction with the areas identified by electroanatomical mapping to better classify and rank the importance of the identified areas. In this example, areas A, B and C are first identified using a mapping technique, e.g. STAR mapping; a data set that can indicate the likelihood of an area being important to maintain arrhythmia is further combined and used to rank the identified driving sites. One way to do this is to use the areas to provide a normalized modification factor that is then applied to the corresponding driving site. Several modification factors can be applied to each identified driving site to give a final likely ranking or display of the respective importance of all driving sites identified by the original mapping method. Data such as from historical patient data can be further applied and machine learning algorithms can be used to determine the exact modification factors to use and the ranking in which to apply them.

[0259] As Figure 4a shown, data gathered from electroanatomical mapping can be used to determine potential driving sites by activation sequence mapping, e.g. STAR mapping, but also to analyze other electrical characteristics of these sites, i.e. to analyze the electrogram features of these areas. If the mapping has been sequentially done, another rule-based approach as detailed in Figures la-d can be used to remove driving sites that are virtually passive activated. These resulting data can be further combined with other information being researched about the patient’s heart, e.g. imaging derived indicators, by which imaging data a modification factor will be produced that can be normalized, weighted and used to modify the importance attributed to each identified driving site. Such data can be cardiac MRI data containing e.g. percentage scar, wall thickness or tissue edema measurements, which can themselves consist of an enhancement scale with contrast agent, e.g. late gadolinium enhancement. Another source of modification information can refer to historical data that can be coupled with machine learning techniques to further derive a modification factor for each identified potential driving site.

[0260] Figure 4b Data flow through the proposed device is shown. Electrogram data is collected by a multi-polar mapping catheter connected to a cardiac mapping system that can be incorporated into or detached from a 3D electroanatomical mapping system. It is understood that instead of using an intracardiac catheter to collect data about the cardiac electrogram, surface electrograms can be used by inverse solution to obtain a “virtual” electrogram on a representation of the heart cavity surface. The data is then transmitted to an analysis module that can exist as a separate computer or be integrated into the electroanatomical mapping system. Here, the data integration and processing as described in Figure 4a is performed before being integrated and displayed to the user, e.g. physician, with the 3D anatomical representation of the heart.

[0261] Figure 5A histogram of cycle lengths (CL) obtained at one of the basket catheter electrodes is shown, where the percentage is made up of each CL on the y-axis and the CL on the x-axis. Each bar represents a defined CL. For illustrative purposes, CLs with a frequency lower than the mean were excluded. Ai shows the Min-CL at 131 ms, which also represents the dominant CL. According to a part of the method, the CLs at 102 ms and 105 ms were excluded because they represent more than 30% of the mean CL recorded at the electrode and represent only less than 10% of the total recorded CLs. Aii shows the Min-CL at 135 ms, while the dominant CL is 142 ms. Again, according to a part of the method described above, the CLs at 105 ms and 109 ms were excluded.

[0262] Figure 6 is a flowchart of the study showing the identified potential AFDs and AFDs related to ablation response, and how many of them are co-localized with the site of Min-CL, lowest CLV, regional DF gradient, and LVZ.

[0263] Figure 7 A-F of are images and electrograms obtained from a subject (patient ID 6). Figure 7 A) of shows the STAR mapping of LA in the titled top view showing a potential AFD mapped to the middle of the roof. Figure 7 B) of shows the mapping CL map in the titled top view demonstrating the overall fastest Min-CL at the mapped potential AFD. Figure 7 C) of shows the CLV mapping in the titled top view demonstrating the lowest CLV at the mapped potential AFD. Figure 7 D) of shows the potential AFD (highlighted by the asterisk) designated by the STAR mapping method and the electrogram obtained at the adjacent electrode. The electrogram obtained at the potential AFD demonstrates that the site is leading and has a faster CL compared to its adjacent counterpart site. Figure 7 E) of shows the CARTO mapping of LA in the top view demonstrating the ablation at the potential AFD guided by the STAR mapping. Figure 8 F) of shows the electrogram demonstrating that the AF terminated to sinus rhythm upon ablation at the potential AFD.

[0264] LUPV - left upper pulmonary vein

[0265] RUPV - right upper pulmonary vein

[0266] LAA - left atrial appendage

[0267] MVA - mitral valve annulus

[0268] Figure 8A-C are images from a different subject (patient 16). In Figure 8 A) of FIG. 1, STAR mapping of LA in the side view with title shows a potential AFD. Figure 8 B) of FIG. 1, CARTO LA mapping of the side view with title shows that ablation at this site resulted in AF to AT termination. Figure 9 C) of FIG. 1, the potential AFD is co-located with the site of regional fastest CL, which is reduced from the CL obtained from the adjacent electrodes.

[0269] LUPV - Left Upper Pulmonary Vein

[0270] LAA - Left Atrial Appendage

[0271] MVA - Mitral Valve Annulus

[0272] Figure 9 A-Cii are images from yet another subject (patient ID 18). In Figure 9 A) of FIG. 2, STAR mapping of LA in the side view shows a potential AFD. Figure 9 B) of FIG. 2, the potential AFD is co-located with the site of regional highest DF, which is reduced from the DF obtained from the adjacent electrodes. Experimental results Ci-Cii) of FIG. 2, ablation at the potential AFD resulted in slowing of CL from 164 ms to 229 ms as shown by the electrogram obtained from the BARD.

[0273] LUPV - Left Upper Pulmonary Vein

[0274] LAA - Left Atrial Appendage

[0275] MVA - Mitral Valve Annulus

[0276] Parameter development method

[0277] The inventors performed experiments as described below to further refine and develop STAR mapping and established a series of parameters that can be used as modification factors to determine other factors by which potential driver sites can be classified according to their importance or as a likelihood of being a true driver site.

[0278] Spectral analysis and AFD

[0279] All patients had high-density bipolar voltage mapping created using a PentaRay Nav catheter with 2-6-2 mm electrode spacing. Points > 3 mm from the geometry surface were filtered due to lack of contact with myocardium, while simultaneously acquired points were breath-gated to optimize accuracy of anatomical localization.

[0280] Examination of parameters

[0281] Bipolar voltage and AFD

[0282] At least 800 bipolar voltage points were acquired per patient to ensure adequate atrial coverage. The interpolation threshold for surface color projection was set to 5 mm, and points were collected to achieve complete LA coverage (i.e., regions no more than 5 mm from a data point).

[0283] Regions with bipolar voltage < 0.5 mV were defined as low voltage zones (LVZ). The voltage map was divided into non-low voltage zones and low voltage zones. The identified potential AF drivers were then categorized as present in non-LVZ or LVZ. The relationship between the potential AFDs mapped to LVZ and the achievement of AF termination at ablation was evaluated. The relationship between the proportion of LVZ and the number of identified AFDs was also evaluated.

[0284] Figures la-d

[0285] CL was determined at each electrode pole contacted in a 5-minute recording per patient using the unipolar signal recorded from the basket catheter. CL was measured as the time difference between two consecutive atrial signals using an automatically tailored Matlab measurement script. A custom algorithm was used to model the reasonable biological behavior based on well-described refractory periods. The algorithm was used to avoid double counting of fragmented electrograms. Unipolar activation timing was taken as the maximum negative deflection (peak negative dv / dt).

[0286] A histogram of all the CLs identified on the x-axis (rounded to the nearest whole millisecond) and the percentage of the recording formed by each CL on the y-axis was plotted for each 5-minute recording per individual electrode per patient. A previous method has been used to identify sites of rapid activity from atrial CL measurements. One method uses the histogram of CLs and takes the center of the narrowest range of CLs in the histogram containing 50% of the cycles and defines it as the "dominant CL." The dominant CL of each electrode is then projected onto a replica of the anatomic geometry to identify the sites of the fastest dominant CLs as defined as the values in the front decile and assess their spatial relationship with AFDs.

[0287] For the new method of determining the site of the fastest CL defined as the minimum CL (Min-CL), the initial extreme CL outliers were ignored, which were defined as CLs that were 30% slower or faster than the average CL at the electrode and those containing < 10% of the cycles. From the remaining CLs, the Min-CL of each electrode was identified as the CL that was the slowest of the remaining CLs. Regional CL and frequency gradient at potential AFD). Thus, after removing outliers, Min-CL is the shortest CL that constitutes >10% of the cycles. Min-CLs are compared across all electrodes to identify the site of the overall fastest Min-CL in the LA. Two definitions of the overall fastest Min-CL are tested: i) Min-CLs within the top decile of all electrodes and / or ii) the shortest one of the overall Min-CLs or "absolute Min-CL" and any other points with a Min-CL within 5% of the overall fastest Min-CL. The location of the electrode with the fastest overall Min-CL is then identified on the same STAR map that displays the ESA to ensure accurate anatomic correlation.

[0288] CL variability (CLV) is used as a marker of organization. CLV is determined for each electrode by taking the standard deviation (SD) of the CLs. Thus, a smaller CLV indicates less variation in CLs. CLVs are compared across all electrodes to identify the site of the overall lowest CLV in the LA. The site of the lowest CLV is defined as i) the CLV that is in the lowest decile and / or ii) the CLV that is within 5% of the overall lowest CLV, referred to as the absolute lowest CLV. The location of the electrode with the lowest CLV is then identified on the STAR map and again correlated to the AFD.

[0289] Cut-off response of potential AFD

[0290] Min-CL at potential AFDs determined using the STAR mapping method is compared to the Min-CLs obtained at adjacent electrodes to yield whether there is a CL gradient from the potential AFD to the surrounding area. This is repeated for all potential AFDs identified in each STAR map in each patient. Adjacent electrodes are defined as electrodes within 3 centimeters of the potential AFD.

[0291] To determine DF, a Butterworth second order filter is applied to the unipolar signal after filtering the far field ventricular signal. After applying a low pass filter to the signal, rectification is performed to obtain the absolute value of the signal. A Hamming window is then applied for Fourier transformation. DF is then determined for each four second window. The median of all these values taken within 5 minutes of recording the DF at the electrode is determined. DF at the electrode identified as a potential AFD is compared to the DFs obtained at adjacent electrodes to determine whether there is a high-low frequency gradient from the potential AFD to the surrounding area. This is repeated for all potential AFDs identified in each STAR map for each patient. The relationship between the LA site with the overall highest DF (from those values in the top decile) and the potential AFD is also assessed.

[0292] Summary of Results

[0293] including 32 patients, ablation was performed for 83 potential AFDs identified by STAR mapping. Ablation response was observed in the case of 73 sites (24 AF termination and 49 CL slowing >30 ms). In 73 sites, 54 (74.0%) were located at the same site as the fastest CL and 55 (75.3%) were located at the same site as the lowest CLV. However, when using conventional markers, sites with the fastest dominant CL and highest DF were less commonly co-located with potential AFDs with ablation response (39, 53.4% CL and 41, 56.2% DF). At potential AFDs, PVI did not affect CL (131.0 ± 12.1 ms pre-PVI vs. 131.2 ± 15.5 ms post-PVI; p = 0.96) or CLV (10.3 ± 3.9 ms pre-PVI vs. 11.0 ± 5.4 ms post-PVI; p = 0.80). These potential AFDs also frequently exhibited regional CL (61 / 73, 83.6%) and frequency gradients (58 / 73, 80.8%). Potential AFDs were more common in the LVZ, with the proportion of LVZ correlating with the number of potential AFDs identified (rs= 0.91; p < 0.001). Using these new markers of rapidity and organization in conjunction with the STAR map allowed for high sensitivity and specificity in predicting which potential AFDs would produce an ablation response. Potential AFDs co-located with the lowest CLV (24 / 24 (100%) vs. 31 / 49 (63.2%); p < 0.001) and fastest CL (24 / 24 (100%) vs. 30 / 49 (61.2%); p < 0.001) were consistently associated with AF termination, rather than CL slowing, upon ablation.

[0294] Detailed Results

[0295] 32 patients underwent STAR mapping-guided ablation. Mean AF duration was 15.4 ± 4.3 months, and 21 of 32 patients (65.6%) underwent antiarrhythmic drug pre-ablation.

[0296] Figures 3a-f

[0297] In summary, in 32 patients, 92 potential AFDs were identified on the STAR map post-PVI (2.8 ± 0.8 per patient), of which 83 (90.2%, 2.6 ± 0.7 per patient) were ablated ( Bipolar voltage and potential AFD ). The 9 potential AFDs that were not ablated occurred in patients in which ablation of a prior site resulted in AF termination.

[0298] 73 potential AFDs (2.3 ± 0.6 per patient) achieved ablation response, which included all 32 patients having at least one response. Based on each potential AFD, AF termination was achieved by ablation at 24 sites (18 tissue to AT and 6 terminated to sinus rhythm), and CL slowing of >30 ms was achieved at 49 sites. Based on each patient, 24 patients achieved AF termination, and CL slowing of >30 ms was achieved in the remaining 8 patients.

[0299] Spectral analysis and potential AFD

[0300] On a per-patient basis, an average of 3.4 ± 0.7 well-defined LVZs were identified. Most of the identified potential AFDs were mapped to LVZs (62 / 92, 67.4%). Of the 83 potential AFDs that were ablated, 60 were mapped to LVZs (72.3%), of which 59 (98.3%) were associated with an ablation response.

[0301] Patients with more than 50% of LA composed of LVZs (59.2 ± 6.5 mV) were more likely to identify more than 2 potential AFDs. There was a strong positive correlation between the proportion of LVZs present and the number of potential AFDs identified (r = 0.91; p < 0.001). However, LVZs alone had little predictive value in predicting AFD sites. s =0.91; p < 0.001). However, LVZs alone had little predictive value in predicting AFD sites.

[0302] The association of potential AFDs with LVZs was highly predictive of ablation response. Potential AFDs mapped to LVZs were more frequently associated with AF termination on ablation relative to potential AFDs mapped to non-LVZs. (Odds ratio = 20.0, 95% CI 1.1-351.5; p = 0.04).

[0303] Figures 3a-f

[0304] For spectral analysis, a total of 170 5-minute monopolar recordings were used. Of these 170 recordings, 84 were created pre-PVI and 86 were created post-PVI. The average dominant CL and average Min-CL obtained at potential AFDs post-PVI were 137.9 ± 64.2 ms and 131.2 ± 15.5 ms, respectively.

[0305] Co-localization of sites identified in spectral analysis and potential AFDs

[0306] i) Fastest dominant CL

[0307] Of the 92 potential AFDs identified after PVI, 47 (51.1%) were co-localized with the fastest main CL site. The main CL showed a sensitivity of 51.1% (95% CI 40.4-61.7%) and a specificity of 18.6% (95% CI 8.4-33.4%) in predicting AFDs. The positive and negative predictive values were 57.3% (95% CI 51.2-63.2%) and 15.1% (95% CI 8.4-25.6%), respectively.

[0308] ii) Min-CL sites within the highest decile and absolute Min-CL

[0309] On average, 4.1 ± 1.1 Min-CL sites were identified within the highest decile per patient. Of the 92 potential AFDs identified after PVI, 58 (63.0%) were co-localized with one of these sites. According to the Min-CL sites in the top decile, they showed a sensitivity of 63.0% (95% CI 52.3-72.9) and a specificity of 20.4% (95% CI 10.2-34.3) in identifying potential AFDs.

[0310] On average, 2.5 ± 0.9 absolute Min-CL sites were identified per patient, which were within 5% of the fastest Min-CL site. Of the 92 potential AFDs identified after PVI, 56 (60.9%) were co-localized with one of these sites.

[0311] iii) Lowest CLV sites within the lowest decile and absolute lowest CLV

[0312] On average, 3.8 ± 1.0 lowest CLV sites were identified within the lowest decile per patient. Of the 92 potential AFDs identified after PVI, 61 (66.3%) were co-localized with one of these sites. According to the lowest CLV sites in the lowest decile, the lowest CLV showed a sensitivity of 66.3% (95% CI 55.7-75.8) and a specificity of 24.4% (95% CI 12.4-40.3) in identifying potential AFDs.

[0313] On average, 2.3 ± 0.8 lowest CLV sites were identified per patient, which were within 5% of the overall lowest CLV sites. Of the 92 potential AFDs identified after PVI, 60 (65.2%) were co-localized with one of these sites.

[0314] Using spectral analysis to predict ablation response at potential AFDs

[0315] i) Fastest main CL

[0316] Of the 73 potential AFDs associated with response, only 39 (53.4%) were co-localised with the site of fastest primary CL ablation. The fastest primary CL showed a sensitivity of 53.4% (95% CI 41.4-65.2%) and a specificity of 26.3% (95% CI 13.4-43.1%) in predicting potential AFDs. Positive and negative predictive values were 58.2% (95% CI 51.1-65.0%) and 22.7% (95% CI 14.1-34.6%) respectively.

[0317] ii) Min-CL site within the top decile and absolute Min-CL

[0318] Of the 73 potential AFDs with study-defined ablation response, 54 (74.0%) were co-localised with one of the Min-CL sites (p<0.001). Figures 4a-b and Figure 5 ).

[0319] Again, co-localisation with these sites is shown to indicate high diagnostic accuracy in predicting potential AFDs with ablation response.

[0320] iii) lowest CLV site within the lowest decile and absolute lowest CLV

[0321] Of the 73 potential AFDs with study-defined ablation response, 56 (76.7%) were co-localised with one of the lowest CLV sites defined within the lowest decile (p<0.001). Figure 6 and Figures 4a-b ).

[0322] Of the 73 potential AFDs associated with ablation response, 55 (75.3%) were co-localised with one of the absolute lowest CLV sites (p<0.001). Figure 6 and Regional CL and frequency gradient and potential AFD Again, co-localisation with these sites is shown to indicate high diagnostic accuracy in predicting potential AFDs with ablation response.

[0323] iv) Prediction of AF termination

[0324] Potential AFDs co-localised with a Min-CL site according to any definition were more frequently associated with AF termination at ablation than CL slowing (24 / 24 (100%) vs 29 / 49 (59.2%); p<0.001). The odds ratio for predicting AF termination at ablation for potential AFDs was 34.1 (95% CI 2.0-592.3; P=0.02) in favour of.

[0325] Compared with CL slowing, potential AFDs co-located with the lowest CLV site according to either definition were more frequently associated with AF termination at resection (24 / 24 (100%) vs. 31 / 49 (63.2%); p < 0.001). Markings show an odds ratio of 28.8 (95% CI 1.7–501.6; p = 0.02) for predicting AF termination at resection at potential AFDs.

[0326] Figure 5

[0327] Of the 92 potential AFDs identified, 56 (60.9%) showed a significant gradient in the min-CL from the potential AFD to the peripheral pole. Figure 8 and Figure 5 (AD). Of the 73 potential AFDs with resection response, 61 (83.6%) showed the fastest-slowest Min-CL gradient from the potential AFD to the adjacent pole. The mean reduction of Min-CL from the potential AFD to the adjacent pole within 3 cm was 10.5 ± 4.2 ms (p = 0.01). The presence of a regional CL gradient at the potential AFD showed an advantage ratio of 5.1 (95% CI 1.3–20.3, p = 0.02) in predicting potential AFDs with resection response.

[0328] The mean frequency gradient (DF) of potential AFDs was 6.2 ± 0.7 Hz. In 56.5% (52 / 92) of cases, potential AFDs co-localized only with the site with the highest DF in the LA, with 41 out of 73 (56.2%) exhibiting a resection response co-localizing with the site of highest DF. Of the 92 identified potential AFDs, 59 (64.1%) showed a significant frequency gradient from the potential AFD to the adjacent pole. Figure 7 and Effect of PVI on spectral analysis data The average reduction in frequency gradient (DF) from the potential AFD to the adjacent pole was 1.8 ± 0.7 Hz (p = 0.01). When only 73 potential AFDs with a resection response were examined, 58 (80.8%) showed a frequency gradient. The presence of a regional frequency gradient at the potential AFD showed an advantage ratio of 5.8 (95% CI 1.4–23.2, p = 0.01) in predicting potential AFDs with a resection response. There was no significant difference in the frequency gradient between the AF termination case and the CL slowdown case at resection (18 / 24, 75.0% vs. 39 / 49, 79.6%; p = 0.78).

[0329] ​

[0330] Of the 92 potential AFDs identified after PVI, 42 were also detected on the pre-PVI map, of which 39 (92.9%) were associated with a study-defined ablation response. Patients in whom potential AFDs were identified pre-PVI had a lower mean CLV pre-PVI (obtained by taking the mean CLV of all electrodes) compared to patients in whom potential AFDs were identified only after PVI (31.4 ± 4.8 ms vs 49.5 ± 7.3 ms; p = 0.01).

[0331] At potential AFDs, PVI did not affect CL (131.0 ± 12.1 ms pre-PVI vs 131.2 ± 15.5 ms post-PVI; p = 0.96) and CLV (10.3 ± 3.9 ms pre-PVI vs 11.0 ± 5.4 ms post-PVI; p = 0.80).

[0332] Potential AFDs also identified pre-PVI exhibited a sensitivity of 53.4% (95% CI 41.4-65.2) and a specificity of 90.0% (95% CI 55.5-99.8) in predicting potential AFDs with an ablation response. Positive and negative predictive values were 97.5% (95% CI 85.7-99.6) and 20.9% (95% CI 16.1-26.7), respectively.

[0333] It will be appreciated that certain embodiments of the present application as described below can be incorporated as code (e.g., software algorithms or programs) residing on firmware and / or computer usable media having control logic to enable execution on a computer system having a computer processor. Such computer system typically includes a memory storage configured to provide output from execution of the code, the code configuring the processor in accordance with the execution. The code can be arranged as firmware or software and can be organized as a set of modules, such as discrete code modules, function calls, procedure calls, or objects in an object-oriented programming environment. If implemented using modules, the code can include a single module or multiple modules that operate in coordination with one another.

[0334] Optional embodiments of the present application can be understood to include the components, elements, and features mentioned and / or indicated herein, individually or collectively, in any or all combinations. In addition, the specific integers mentioned herein are understood to include any known equivalents thereof, whether or not such equivalents are explicitly described herein.

[0335] While the illustrated embodiments of the application have been described, it will be understood by those skilled in the art that various changes, substitutions and alterations can be made, without departing from the spirit and scope of the application as defined by the recitations in the claims that follow, and that the scope of the application is not to be limited by any of the specific disclosed embodiments.

[0336] This application claims priority from GB 1915680.1, the contents of which and the contents of the abstract filed herewith are incorporated herein by reference.

Claims

1. A computer-implemented method for identifying one or more regions of the heart responsible for supporting or inducing abnormal heart rhythms, the computer-implemented method using electrocardiographic data recorded from multiple electrodes on a multipolar cardiac catheter, the electrocardiographic data being obtained from a corresponding series of sensing locations on the heart during a recording time period, the method comprising the steps of: The electrocardiogram identifies regions within the heart chambers that possess an electrical activation sequence, which characterizes these regions as potential drivers of abnormal heart rhythms. For each sensing location in or substantially around the region, the earliest activated electrode site is determined based on the primary activation. For each identified earliest activated electrode site: Calculate the value of each of a plurality of modifiers associated with the electrode site, the modifiers being determined based on the anatomical features of the site; Determine the sorting factor calculated based on the plurality of modifiers; Each of the earliest activated electrode sites is sorted according to its sorting factor; as well as The output identifies data about the region, which, according to the sorting, alters the salience of each of the earliest identified activation electrode sites. The modifiers determined based on the anatomical features include sensing locations that depend on the electrode sites.

2. The computer-implemented method according to claim 1, wherein, The output data includes visual indicators that change the salience of each of the earliest identified activated electrode sites according to the sorting.

3. The computer-implemented method according to claim 1, wherein, The anatomical features of the site are identified using an imaging system.

4. The computer-implemented method according to claim 3, wherein, The imaging system is selected from intracardiac echocardiography, cardiac magnetic resonance imaging, and cardiac computed tomography.

5. The computer-implemented method according to claim 1, wherein, The modifier, determined based on the anatomical features of the site, is determined based on the proximity of the electrode site to the anatomical structure.

6. The computer-implemented method according to claim 5, wherein, The degree of proximity is determined by calculating geodesic distances to one or more predetermined anatomical points.

7. The computer-implemented method according to claim 6, wherein, The predetermined anatomical points are selected from a set including: pulmonary veins or atrial appendages; junctions of appendages and veins; Marshall veins; mitral valve annulus; ligaments; and aneurysms.

8. A computer-implemented method for identifying one or more regions of the heart responsible for supporting or inducing abnormal heart rhythms, the computer-implemented method using electrocardiographic data recorded from multiple electrodes on a multipolar cardiac catheter, the electrocardiographic data being obtained from a corresponding series of sensing locations on the heart during a recording time period, the method comprising the steps of: The electrocardiogram identifies regions within the heart chambers that possess an electrical activation sequence, which characterizes these regions as potential drivers of abnormal heart rhythms. For each sensing location in or substantially around the region, the earliest activated electrode site is determined based on the primary activation. For each identified earliest activated electrode site: Calculate the value of each of a plurality of modifiers associated with the electrode site, the modifiers being determined based on the anatomical features of the site; Determine the sorting factor calculated based on the plurality of modifiers; Each of the earliest activated electrode sites is sorted according to its sorting factor; as well as, The output identifies data about the region, which modifies the significance of each of the earliest identified activation electrode sites according to the sorting, wherein modifiers are determined based on treatment outcomes at another earliest activated electrode site. The modifier determined based on the anatomical features depends on the sensing location of the electrode site.

9. The computer-implemented method according to claim 8, wherein, The anatomical features of the site are identified using an imaging system.

10. The computer-implemented method according to claim 9, wherein, The imaging system is selected from intracardiac echocardiography, cardiac magnetic resonance imaging, and cardiac computed tomography.

11. The computer-implemented method according to claim 8, wherein, The modifier, determined based on the anatomical features of the site, is determined based on the proximity of the electrode site to the anatomical structure.

12. The computer-implemented method according to claim 11, wherein, The degree of proximity is determined by calculating geodesic distances to one or more predetermined anatomical points.

13. The computer-implemented method according to claim 12, wherein, The predetermined anatomical points are selected from a set including: pulmonary veins or atrial appendages; junctions of appendages and veins; Marshall veins; mitral valve annulus; ligaments; and aneurysms.

14. A computer-implemented method for identifying one or more regions of the heart responsible for supporting or inducing abnormal heart rhythms, the computer-implemented method using electrocardiographic data recorded from multiple electrodes on a multipolar cardiac catheter, the electrocardiographic data being obtained from a corresponding series of sensing locations on the heart during a recording time period, the method comprising the steps of: The electrocardiogram identifies regions within the heart chambers that possess an electrical activation sequence, which characterizes these regions as potential drivers of abnormal heart rhythms. For each sensing location in or substantially around the region, the earliest activated electrode site is determined based on the primary activation. For each identified earliest activated electrode site: Calculate the value of each of a plurality of modifiers associated with the electrode site, the modifiers being determined based on the anatomical features of the site; Determine the sorting factor calculated based on the plurality of modifiers; Each of the earliest activated electrode sites is sorted according to its sorting factor; as well as, Output data identifying the region, the data altering the salience of each of the identified earliest activated electrode sites according to the sorting, wherein if activation occurs across at least two electrodes and the angle from the vertex opposite the potential driving site to the two electrodes is less than a predetermined angle, then the potential driving site is classified as the earliest activated electrode site. At least one electrogrammatic activation was identified as occurring later than the potential driving site within a defined excitatory tissue arc. The modifier determined based on the anatomical features depends on the sensing location of the electrode site.

15. The computer-implemented method according to claim 14, wherein, The anatomical features of the site are identified using an imaging system.

16. The computer-implemented method according to claim 15, wherein, The imaging system is selected from intracardiac echocardiography, cardiac magnetic resonance imaging, and cardiac computed tomography.

17. The computer-implemented method according to claim 14, wherein, The modifier, determined based on the anatomical features of the site, is determined based on the proximity of the electrode site to the anatomical structure.

18. The computer-implemented method according to claim 17, wherein, The degree of proximity is determined by calculating geodesic distances to one or more predetermined anatomical points.

19. The computer-implemented method according to claim 18, wherein, The predetermined anatomical points are selected from a set including: pulmonary veins or atrial appendages; junctions of appendages and veins; Marshall veins; mitral valve annulus; ligaments; and aneurysms.

20. The computer-implemented method according to claim 14, wherein, The predetermined angle is less than 180 degrees.

21. The computer-implemented method according to claim 14, further comprising: The data is output to a monitor or medical scanning device to induce a display or navigation of the cavity geometry and guide the placement of catheters or electrodes for subsequent electrocardiogram data acquisition.

22. The computer-implemented method according to claim 21, further comprising: Identify sites for subsequent placement of the catheter or electrode to eliminate or confirm a previously determined first active electrode, or to extend the electrogram data to areas that were not previously scanned or not fully scanned.

23. The computer-implemented method according to claim 22, further comprising: The site is identified based on the location of the first identified activation site or based on the gradient in the activation vector and the signal leading score.

24. The computer-implemented method according to claim 14, further comprising: It receives electrogram data essentially in real time and provides indications of when sufficient timing has been achieved for the corresponding sites.

25. The computer-implemented method according to claim 24, wherein, The steps for determining whether sufficient acquisition timing has been performed include one or more of the following: counting down a predetermined time period, acquiring a predetermined number of activation cycles for at least a minimum number of electrodes on the flow multipolar mapping catheter, and acquiring a predetermined time or a predetermined number of activation cycles, wherein the activation modes across multiple electrodes are kept in a pattern-matching sequence.

26. The computer-implemented method according to claim 24 or 25, wherein, The step of determining whether sufficient acquisition timing has been performed includes: determining that the activation mode of the site being scanned has reached a predetermined level of statistical determinism.

27. A computer-implemented method for identifying one or more regions of the heart responsible for supporting or inducing abnormal heart rhythms, the computer-implemented method using electrocardiographic data recorded from multiple electrodes on a multipolar cardiac catheter, the electrocardiographic data being obtained from a corresponding series of sensing locations on the heart during a recording time period, the method comprising the steps of: The electrocardiogram identifies regions within the heart chambers that possess an electrical activation sequence, which characterizes these regions as potential drivers of abnormal heart rhythms. For each sensing location in or substantially around the region, the earliest activated electrode site is determined based on the primary activation. For each identified earliest activated electrode site: Calculate the value of each of a plurality of modifiers associated with the electrode site, the modifiers being determined based on the anatomical features of the site; Determine the sorting factor calculated based on the plurality of modifiers; Each of the earliest activated electrode sites is sorted according to its sorting factor; as well as, The output identifies data about the region, which, according to the sorting, alters the salience of each of the earliest identified activation electrode sites. The method further includes: deriving the wavefront direction by determining which electrode in each electrode pair leads, and processing the wavefront direction to determine whether the earliest activated electrode is a true AFD or represents a passively activated site activated beyond the measurement boundary. The modifier determined based on the anatomical features depends on the sensing location of the electrode site.

28. The computer-implemented method according to claim 27, wherein, The anatomical features of the site are identified using an imaging system.

29. The computer-implemented method according to claim 28, wherein, The imaging system is selected from intracardiac echocardiography, cardiac magnetic resonance imaging, and cardiac computed tomography.

30. The computer-implemented method according to claim 27, wherein, The modifier, determined based on the anatomical features of the site, is determined based on the proximity of the electrode site to the anatomical structure.

31. The computer-implemented method according to claim 30, wherein, The degree of proximity is determined by calculating geodesic distances to one or more predetermined anatomical points.

32. The computer-implemented method according to claim 31, wherein, The predetermined anatomical points are selected from a set including: pulmonary veins or atrial appendages; junctions of appendages and veins; Marshall veins; mitral valve annulus; ligaments; and aneurysms.

33. A computer-implemented method for analyzing electrograms acquired from the heart to identify one or more regions of the heart responsible for supporting or inducing abnormal heart rhythms, the computer-implemented method using electrogram data recorded from multiple electrodes on a multipolar cardiac catheter, the electrogram data being obtained from a corresponding series of sensing locations on the heart within a recording time period, the method comprising the steps of: By analyzing the sequence of electrical activation, specific regions within the heart's chambers are defined as potential drivers of abnormal rhythms. The region is weighted according to factors including the following: During the same acquisition period, the direction of the wavefront generated by activation travels from the potential driving site to one or more nearby electrodes relative to that driving site; The activation sequence and timing are within a biologically reasonable range, and the reference conduction velocity and pathway are within a reasonable activation sequence. Activation occurs across at least two electrodes, where the angle between the vertex opposite the potential driving site and the two electrodes is less than a predetermined angle, and At least one electrogram activation was identified as occurring within a defined excitable tissue arc later than the potential driving site, and Further acquisition performed at adjacent locations opposite the missing arc of the first potential driving site did not identify any potential driving sites at similar locations. as well as Based on the weighting, refined potential drivers are displayed in a highlighted manner, the display relating to a computer representation of the heart chambers. Specifically, if activation occurs across at least two electrodes, and the angle between the vertex opposite the potential driving site and the two electrodes is less than a predetermined angle, then the potential driving site is classified as the earliest activated electrode site. For each earliest activated electrode site: Calculate the value of each of a plurality of modifiers associated with the electrode site, the modifiers being determined based on the anatomical features of the site; Determine the sorting factor calculated based on the plurality of modifiers; Each of the earliest activated electrode sites is sorted according to its sorting factor; and Output data identifying the region, the data altering the salience of each of the earliest identified activation electrode sites according to the sorting; The modifier is determined based on anatomical features, wherein the modifier determined based on the anatomical features depends on the sensing location of the electrode site.

34. The computer-implemented method according to claim 33, wherein, The anatomical features of the site are identified using an imaging system.

35. The computer-implemented method according to claim 34, wherein, The imaging system is selected from intracardiac echocardiography, cardiac magnetic resonance imaging, and cardiac computed tomography.

36. The computer-implemented method according to claim 33, wherein, The modifier, determined based on the anatomical features of the site, is determined based on the proximity of the electrode site to the anatomical structure.

37. The computer-implemented method according to claim 36, wherein, The degree of proximity is determined by calculating geodesic distances to one or more predetermined anatomical points.

38. The computer-implemented method according to claim 36, wherein, The predetermined anatomical points are selected from a set including: pulmonary veins or atrial appendages; junctions of appendages and veins; Marshall veins; mitral valve annulus; ligaments; and aneurysms.

39. A computer system for identifying one or more regions of the myocardium responsible for supporting or inducing abnormal heart rhythms using electrocardiographic data recorded from multiple electrodes on a multipolar cardiac catheter, the electrocardiographic data being obtained from a corresponding series of sensing locations on the heart during a recording period, the system comprising: processor; The first memory is used to store the received electrical diagram data; as well as A second memory having program code stored therein, which, when executed by the processor, causes the system to: Electrocardiography identifies regions within the heart's chambers that possess an electrical activation sequence, which characterizes these regions as potential drivers of abnormal heart rhythms. For each sensing location in or substantially around the region, the earliest activated electrode site is determined based on the primary activation. For each identified earliest activated electrode site: Calculate the value of each of a plurality of modifiers associated with the electrode site, the modifiers being determined based on the anatomical features of the site; Determine the sorting factor calculated based on the plurality of modifiers; Each of the earliest activated electrode sites is sorted according to its sorting factor; as well as The output identifies data about the region, which depends on the sorting to change the salience of each of the earliest identified activation electrode sites, wherein the program code, when executed by the processor, causes the system to determine the modifier based on the anatomical features by accessing the data to obtain values ​​depending on the location of the electrode sites.

40. The computer system according to claim 39, wherein, When executed by the processor, the program code causes the system to output a visual instruction that alters the salience of each of the earliest identified activation electrode sites according to the sorting.

41. The computer system according to claim 39, wherein, When executed by the processor, the program code causes the system to output data to a display or medical scanning device to induce the display or navigation of the cavity geometry and guide the placement of catheters or electrodes for subsequent electrocardiogram data acquisition.

42. The computer system according to claim 39, wherein, The anatomical features of the site are identified using an imaging system.

43. The computer system according to claim 42, wherein, The imaging system is selected from intracardiac echocardiography, cardiac magnetic resonance imaging, and cardiac computed tomography.

44. The computer system according to claim 39, wherein, The modifier, determined based on the anatomical features of the site, is determined based on the proximity of the electrode site to the anatomical structure.

45. The computer system according to claim 44, wherein, The proximity of the electrode site to the anatomical structure is determined by calculating geodesic distances to one or more predetermined anatomical points.

46. ​​The computer system according to claim 45, wherein, The predetermined anatomical points are selected from a set including: pulmonary veins or atrial appendages; junctions of appendages and veins; Marshall veins; mitral valve annulus; ligaments; and aneurysms.

Images