Ventricular far-field estimation using autoencoders

Abstract

The invention is entitled "Ventricular Far-Field Estimation Using Autoencoders." The invention provides a method. The method includes receiving input intracardiac signals from a monitoring and treatment device. Each of the input intracardiac signals includes artifacts. The method includes encoding, by an autoencoder, the input intracardiac signals with an intracardiac dataset to produce a latent representation. The method also includes decoding, by the autoencoder, the latent representation to produce output intracardiac signals. The output intracardiac signals include the input intracardiac signals reconstructed without the signal artifacts.

Description

Technical Field

[0001] This invention relates to an artificial intelligence and machine learning autoencoder associated with ventricular far-field estimation and the identification and decomposition of near-field and far-field signals in cardiac electrical activity. Background Technology

[0002] Treatment of cardiac conditions such as arrhythmias often requires cardiac mapping (i.e., mapping the heart tissue, chambers, veins, arteries, and / or pathways). An electrocardiogram (ECG) is an example of cardiac mapping. An ECG is generated from electrical signals from the heart that describe its activity.

[0003] ECG is used during cardiac procedures to identify the potential origin of heart disease. Generally, when physicians use ECG to study cardiac activity, signal interference, artifacts, and noise associated with the underlying electrical signals of the ECG can significantly impair its accuracy. Signal interference can also be caused by processing signal regions with abrupt changes, peaks, and / or pacing signals, including high-frequency and harmonic regions. Due to these interferences, artifacts, and noise, physicians cannot distinguish the origin of the ventricles and atria in real time (e.g., during cardiac procedures), which increases the difficulty of diagnosing / treating heart disease. Therefore, there is a need for improved methods of cardiac mapping to remove such interferences, artifacts, and noise.

[0004] A monopolar signal is a combination of near-field and far-field signals. During ablation procedures, it is important to identify and isolate the near-field signal. When electrodes are inserted into muscles such as the heart muscle, each activation of the muscle generates an electric field. Each electrode captures all sources of the electric field at its placement location, including near-field signals close to the electrode and far-field signals farther away. Summary of the Invention

[0005] According to one embodiment, a method is provided. The method includes receiving input intracardiac signals from a monitoring and processing device. Each of these input intracardiac signals may include at least one artifact. The method includes encoding these input intracardiac signals using an autoencoder with an intracardiac dataset to produce a latent representation. The method further includes decoding the latent representation by the autoencoder to produce output intracardiac signals. These output intracardiac signals may include reconstructed input intracardiac signals without artifacts.

[0006] According to one embodiment, a method for decomposing near-field and far-field signals is provided. Measured signals can be received. The measured signals can be encoded by an autoencoder to generate a latent representation. This latent representation can be decoded by the autoencoder to decompose near-field and far-field components from the measured signals. Far-field ventricular measurements can be acquired. Measurement results can be acquired using multi-electrode catheters and surface ECG signals. Synthesized local field signals can be added. The acquired far-field signal and residual near-field signal can be detected. Decoding of the latent representation can be based on the detected acquired far-field signal and residual near-field signal.

[0007] According to one or more implementation schemes, the above-described method implementation schemes can be implemented as devices, systems, and / or computer program products. Attached Figure Description

[0008] A more detailed understanding can be obtained through the following specific embodiments provided by way of example and in conjunction with the accompanying drawings, wherein similar reference numerals in the drawings indicate similar elements, and wherein:

[0009] Figure 1 A diagram illustrating an exemplary system that can implement one or more features of the subject matter of this disclosure.

[0010] Figure 2 A block diagram of an exemplary system for remotely monitoring and transmitting patient biometrics according to one or more embodiments is shown;

[0011] Figure 3 A graphical depiction of an artificial intelligence system according to one or more implementation schemes is shown;

[0012] Figure 4 The following are shown according to one or more embodiments: Figure 3 A block diagram of the methods executed in an artificial intelligence system;

[0013] Figure 5 An example of a neural network according to one or more implementation schemes is shown;

[0014] Figure 6 A block diagram of a method according to one or more embodiments is shown;

[0015] Figure 7 A graphical depiction of signals according to one or more embodiments is shown;

[0016] Figure 8 A graphical depiction of signals according to one or more embodiments is shown;

[0017] Figure 9 A graphical depiction of signals according to one or more embodiments is shown;

[0018] Figure 10 A graphical depiction of the signal process according to one or more embodiments is shown;

[0019] Figure 11 A block diagram of a method according to one or more embodiments is shown; and

[0020] Figure 12 This is an exemplary flowchart of an exemplary method for decomposing near-field and far-field signals according to an implementation scheme. Detailed Implementation

[0021] This paper discloses an artificial intelligence and machine learning autoencoder (hereinafter referred to as an autoencoder). The autoencoder can be processor-executable code or software, which necessarily originates from processing operations performed by a medical device and the processing hardware of that medical device to provide improved ECGs for the treatment of cardiac conditions. According to one embodiment, the autoencoder can provide specific encoding and decoding methods for the medical device. These specific encoding and decoding methods may involve multi-step data manipulation of the heart's electrical signals, removing signal interference, signal artifacts, and signal noise from the electrical signals.

[0022] In this regard, and in operation, the autoencoder can receive input intracardiac signals (e.g., cardiac electrical signals including signal interference, signal artifacts, and signal noise). These intracardiac signals can be recorded and processed in real time by a monitoring and processing device (e.g., a catheter containing an autoencoder), and / or recorded and transmitted by the monitoring and processing device to a computing device containing an autoencoder.

[0023] This autoencoder encodes an input intracardiac signal using an intracardiac dataset (e.g., a pre-defined and approved electrical signal of the heart free from signal interference, artifacts, and noise). This encoding by the autoencoder produces a latent representation from the input intracardiac signal. The autoencoder can further decode this latent representation to produce an output intracardiac signal. The output intracardiac signal can be a reconstructed version of the input intracardiac signal free from signal interference, artifacts, and noise. An improved ECG for treating heart conditions is then generated based on the output intracardiac signal.

[0024] The technical benefits of autoencoders include the real-time generation of output intracardiac signals, which further enables the generation of improved ECGs for physicians (such as during cardiac procedures) to study cardiac activity and identify potential origins of heart disease. The improved ECGs are not obscured by signal interference, artifacts, and noise from the original input intracardiac signals, as these artifacts are removed during decoding. Furthermore, the technical benefits of autoencoders include the generation of improved ECGs with increased accuracy, where signal interference, artifacts, and noise have been removed, allowing for the separate real-time determination of the origins of the ventricles and atria.

[0025] In one implementation, an autoencoder can be used to train the system to decompose near-field and far-field signals detected by electrodes through analysis of a large number of data points. Bits can be selected as part of the training set to train the system to identify the far-field signal components.

[0026] A signal can be provided, and it can be attempted to be regenerated by providing a signal with a large amount of far-field signal and a signal with both a large amount of far-field and near-field signal. A signal can be provided to the autoencoder to reconstruct the far-field signal. Once the network is trained, it can output the far-field component from the provided signal.

[0027] Figure 1 An illustration is shown of an exemplary system 100 (e.g., a medical device) that can implement one or more features of the subject matter of this disclosure. All or part of the system 100 can be used to collect information for an intracardiac dataset (e.g., a training dataset), and / or all or part of the system 100 can be used to implement the autoencoder (e.g., a trained model) described herein.

[0028] System 100 may include components, such as catheter 105, configured to damage tissue regions of organs within the body. Catheter 105 may also be further configured to acquire biometric data, including electrical signals (e.g., intracardiac signals) of the heart. Although catheter 105 is shown as a pointed catheter, it should be understood that catheters of any shape, including one or more elements (e.g., electrodes), may be used to implement embodiments disclosed herein.

[0029] System 100 includes a probe 110 having an axis that can be navigated by a physician or medical professional 115 to a body part (such as the heart 120) of a patient 125 lying on a bed (or table) 130. Multiple probes may be provided depending on the implementation. However, for simplicity, a single probe 110 is described herein. It should be understood, however, that probe 110 may represent multiple probes.

[0030] Exemplary system 100 can be used to detect, diagnose, and treat cardiac conditions (e.g., using intracardiac signals). Cardiac conditions, such as arrhythmias (specifically atrial fibrillation), have been common and dangerous medical conditions, especially among older adults. In a patient with a normal sinus rhythm (e.g., patient 125), the heart (e.g., heart 120), comprising the atria, ventricles, and excitatory conduction tissues, is electrically stimulated to beat in a synchronized, patterned manner. This electrical stimulation can be detected as an intracardiac signal.

[0031] In patients with arrhythmias (e.g., patient 125), abnormal areas of cardiac tissue do not follow the synchronous beating cycle associated with normal conduction tissue as in patients with normal sinus rhythm. Instead, the abnormal areas of cardiac tissue conduct abnormally to adjacent tissues, thus disrupting the cardiac cycle into an asynchronous rhythm. This asynchronous rhythm can also be detected as an intracardiac signal. Such abnormal conduction is previously known to occur in various regions of the heart (e.g., heart 120), such as the sinoatrial (SA) node region, the conduction pathway along the atrioventricular (AV) node, or in the myocardial tissue that forms the walls of the ventricles and atria.

[0032] Furthermore, arrhythmias, including atrial arrhythmias, can be multi-wave reentrant, characterized by multiple asynchronous loops of electrical impulses dispersed around the atrial chambers and typically self-propagating (e.g., another example of intracardiac signals). Alternatively, or in addition to multi-wave reentrant arrhythmias, arrhythmias can also have a focal source, such as when an isolated area of ​​tissue within the atrium beats spontaneously in a rapid, repetitive manner (e.g., another example of intracardiac signals). Ventricular tachycardia (V-tach or VT) is a tachycardia or rapid rhythm originating in a single ventricle. This is a potentially life-threatening arrhythmia because it can lead to ventricular fibrillation and sudden cardiac death.

[0033] Atrial fibrillation (AF) is a type of arrhythmia that occurs when normal electrical impulses generated by the sinoatrial node (e.g., another example of intracardiac signaling) are overwhelmed by disordered electrical impulses originating from the atria and pulmonary veins (e.g., signal interference), which cause irregular impulses to be transmitted to the ventricles. This results in an irregular heartbeat that can last from minutes to weeks, or even years. AF is typically a chronic condition that slightly increases the risk of death, usually caused by stroke. First-line treatment for AF is medication to slow the heart rate or restore a normal rhythm. Additionally, people with AF are often given anticoagulants to protect against their risk of stroke. The use of such anticoagulants carries the inherent risk of internal bleeding. For some patients, medication is insufficient; their AF is considered drug-resistant, meaning it cannot be treated with standard medical interventions. Synchronized cardioversion can also be used to restore AF to a normal rhythm. Alternatively, catheter ablation can be used to treat patients with AF.

[0034] Catheter-based ablation therapy can include mapping the electrical properties of cardiac tissue (particularly the endocardium and cardiac volume) and selectively ablating cardiac tissue by applying energy. Cardiac mapping includes creating mapping maps of potentials propagating along cardiac tissue (e.g., voltage mapping) or mapping maps of arrival times to various tissue locations (e.g., local time activation (LAT) mapping). Cardiac mapping can be used to detect localized cardiac tissue dysfunction. Ablation, such as cardiac mapping-based ablation, can stop or alter unwanted electrical signals propagating from one part of the heart to another.

[0035] The ablation process disrupts unwanted electrical pathways by creating a non-conductive ablation focus. Various forms of energy delivery for creating ablation focuses have been disclosed, including the use of microwaves, lasers, and more commonly, radiofrequency energy to create conduction blocks along the heart tissue wall. In a two-step mapping-then-ablation procedure, electrical activity at various points in the heart is typically sensed and measured by inserting a catheter (e.g., catheter 105) containing one or more electrical sensors (e.g., at least one ablation electrode 134 of catheter 105) into the heart (e.g., heart 120) and acquiring data at multiple points. This data (e.g., biometric data including intracardiac signals) is then used to select the endocardial target region for ablation. This data is more accurate and better supports the selection of the endocardial target region for ablation than the underlying electrical signal of an ECG, which includes signal interference, signal artifacts, and signal noise, due to the use of an autoencoder employed by the exemplary system 100 (e.g., a medical device device). Signal interference, signal artifacts, and signal noise are collectively referred to herein as artifacts. Examples of artifacts include, but are not limited to, power noise (e.g., electrostatic and electromagnetic coupling between the circuit and a 50 or 60 Hz power line), Fluro noise (e.g., fluorescent lamps), contact noise (e.g., collisions between conduit electrodes), and deflection noise (e.g., electrostatic discharge during conduit deflection).

[0036] As clinicians treat increasingly challenging conditions such as atrial fibrillation and ventricular tachycardia, cardiac ablation and other cardiac electrophysiological procedures become increasingly complex. Treatment of complex arrhythmias currently relies on the use of three-dimensional (3D) mapping systems to reconstruct the anatomy of the cardiac chambers of interest. In this regard, the autoencoder employed in the exemplary system 100 (e.g., a medical device) provides the underlying output signal, enabling the generation of improved 3D mapping maps and / or ECGs for the treatment of cardiac conditions.

[0037] For example, cardiologists rely on software such as CARTO, produced by BiosenseWebster, Inc. (Diamond Bar, Calif.). ®3. A complex fragmented atrial electrocardiogram (CFAE) module of a 3D mapping system to generate and analyze intracardiac electrograms (EGMs). An autoencoder of an exemplary system 100 (e.g., a medical device) enhances the software to generate and analyze improved intracardiac electrograms (EGMs), enabling the determination of ablation points for the treatment of a range of cardiac conditions, including atypical atrial flutter and ventricular tachycardia.

[0038] Improved 3D mapping supported by autoencoders can provide multiple pieces of information about the electrophysiological properties of tissues, representing the anatomical and functional matrix of these challenging arrhythmias.

[0039] Cardiomyopathy with different etiologies (hypoxia, dilated (DCM), hypertrophic cardiomyopathy (HCM), arrhythmogenic right ventricular dysplasia (ARVD), left ventricular noncompaction (LVNC), etc.) has a recognizable matrix, characterized by a region of unhealthy tissue surrounded by normally functioning cardiomyocytes.

[0040] Abnormal tissue is typically characterized by low-voltage EGM. However, initial clinical experience in endocardial-epicardial mapping indicates that low-voltage regions are not always the sole arrhythmogenic mechanism present in such patients. In fact, low- or intermediate-voltage regions can exhibit fragmented and prolonged EGM activity during sinus rhythm, which corresponds to the critical isthmus identified during sustained and tissue ventricular arrhythmias, e.g., only in intolerable ventricular tachycardias. Furthermore, in many cases, fragmented and prolonged EGM activity is observed in regions showing normal or near-normal voltage amplitudes (>1–1.5 mV). While the latter regions can be assessed based on voltage amplitude, they cannot be considered normal based on intracardiac signals and thus represent the true arrhythmogenic matrix. 3D mapping enables the localization of the arrhythmogenic matrix on the endocardial and / or epicardial layers of the right / left ventricle, which can vary in distribution depending on the extent of the primary disease.

[0041] The matrix associated with these cardiac conditions is the presence of fragmented and elongated EGM in the endocardial and / or epicardial layers of the ventricular chambers (right and left). 3D mapping systems, such as CARTO... ® 3. It can locate the potential arrhythmogenic matrix of cardiomyopathy in abnormal EGM detection.

[0042] Electrode catheters (e.g., catheter 105) are used in medical practice. These catheters are used to stimulate and map electrical activity in the heart, and to ablate sites of abnormal electrical activity. In use, the electrode catheter is inserted into a major vein or artery, such as the femoral artery, and then guided to the cardiac chamber of interest. A typical ablation procedure involves inserting a catheter with at least one electrode at its distal end into the cardiac chamber. A reference electrode is provided, typically taped to the patient's skin, or alternatively, a second catheter positioned in or near the heart may be used to provide the reference electrode. Radiofrequency (RF) current is applied to the tip electrode of the ablation catheter, and the current flows through the surrounding medium (i.e., blood and tissue) to the reference electrode. The current distribution depends on the amount of contact between the electrode surface and the tissue, which has a higher conductivity than blood. Heating of the tissue occurs due to its resistance. The tissue is sufficiently heated to destroy cells in the cardiac tissue, resulting in the formation of a non-conductive ablation focus within the cardiac tissue. Heating of the electrode also occurs during this process due to conduction from the heated tissue to the electrode itself. If the electrode temperature becomes high enough, perhaps above 60 degrees Celsius, a thin, transparent coating of dehydrated hemoglobin can form on the electrode surface. If the temperature continues to rise, this dehydrated layer can become increasingly thick, causing blood to clot on the electrode surface. Because dehydrated biomaterials have a higher electrical resistance than endocardial tissue, the impedance to the flow of electrical energy into the tissue also increases. If the impedance increases sufficiently, an impedance rise occurs, and the catheter must be removed from the body and the tip electrode cleaned.

[0043] Treatment of cardiac conditions such as arrhythmias often requires detailed mapping of cardiac tissue, chambers, veins, arteries, and / or electrical pathways. For example, a prerequisite for successful catheter ablation is that the cause of the arrhythmia is accurately located within a cardiac chamber. Such localization can be accomplished via an electrophysiological study, during which potentials are spatially resolved using a mapping catheter introduced into the cardiac chamber. This electrophysiological study (so-called electroanatomical mapping) thus provides 3D mapping data that can be displayed on a monitor. In many cases, mapping and therapeutic functions (e.g., ablation) are provided by a single catheter or a group of catheters, such that the mapping catheter also functions as a therapeutic (e.g., ablation) catheter. In this case, an autoencoder can be directly stored and executed by catheter 105.

[0044] Mapping of cardiac regions (such as cardiac areas, tissues, veins, arteries, and / or electrical pathways of the heart (e.g., 120)) can lead to the identification of problem areas such as scar tissue, sources of arrhythmia (e.g., tachyarrhythmias), healthy areas, etc. Cardiac regions can be mapped to provide a visual rendering of the mapped cardiac regions using a display, as further disclosed herein. Additionally, cardiac mapping can include mapping based on one or more modalities, such as, but not limited to, local activation time (LAT), electrical activity, topology, bipolar mapping, dominant frequency, or impedance. Data corresponding to multiple modalities can be captured using a catheter inserted into the patient's body, and data corresponding to multiple modalities can be provided simultaneously or at different times for rendering based on the corresponding settings and / or preferences of a healthcare professional.

[0045] Cardiac mapping can be achieved using one or more techniques. As an example of a first technique, cardiac mapping can be achieved by sensing the electrical properties of cardiac tissue at precise locations within the heart (e.g., LAT). The corresponding data can be acquired via one or more catheters advanced into the heart using catheters with electrical and position sensors at their distal ends. Specifically, for example, position and electrical activity can initially be measured at approximately 10 to approximately 20 points on the inner surface of the heart. These data points are typically sufficient to generate a preliminary reconstruction or mapping map of the cardiac surface of satisfactory quality. The preliminary map can be combined with data taken from additional points to produce a more comprehensive map of cardiac electrical activity. In clinical settings, it is not uncommon to accumulate data at 100 or more sites to generate a detailed and comprehensive mapping map of cardiac chamber electrical activity. The resulting detailed map can then serve as a basis for determining therapeutic actions, such as tissue ablation, to alter the propagation of cardiac electrical activity and restore normal heart rhythm.

[0046] like Figure 1 As shown, a medical professional 115 can insert the shaft 137 through the sheath 136 while manipulating the distal end of the shaft 137 using a manipulator 138 near the proximal end of the catheter 105 and / or by deflecting it from the sheath 136. Figure 1 As shown in 140, catheter 105 can be fitted at the distal end of shaft 137. Catheter 105 can be inserted through sheath 136 in a collapsed state and then deployed within heart 120. As further described herein, catheter 105 may include at least one ablation electrode 134 and catheter needle.

[0047] According to the implementation plan, catheter 105 can be configured to ablate tissue areas of the heart chambers of heart 120. Figure 1Figure 150 shows a catheter 105 within a cardiac chamber of heart 120 in an enlarged view. As shown, catheter 105 may include at least one ablation electrode 134 coupled to the body of the catheter. According to other embodiments, multiple elements may be connected via a strip forming the shape of catheter 105. One or more other elements (not shown) may be provided, which may be any element configured to ablate or obtain biometric data, and may be an electrode, a transducer, or one or more other elements.

[0048] According to the embodiments disclosed herein, an ablation electrode, such as at least one ablation electrode 134, can be configured to deliver energy to a tissue region of an organ in the body, such as the heart 120. The energy can be thermal and can cause damage to the tissue region by starting from the surface of the tissue region and extending into the thickness of the tissue region.

[0049] According to the embodiments disclosed herein, biometric data may include one or more of the following: LAT, electrical activity, topology, bipolar mapping, dominant frequency, impedance, etc. LAT can be a time point corresponding to a threshold activity of local excitation calculated based on a normalized initial starting point. Electrical activity can be any applicable electrical signal that can be measured based on one or more thresholds and sensed and / or amplified based on signal-to-noise ratio and / or other filters. Topology can correspond to the physical structure of a body part or a portion of a body part, and can correspond to variations in the physical structure relative to different parts of the body part or relative to different body parts. Dominant frequency can be a frequency or frequency range that is prevalent in a part of a body part and can differ in different parts of the same body part. For example, the dominant frequency of the pulmonary veins of the heart can differ from the dominant frequency of the right atrium of the same heart. Impedance can be a resistance measurement at a given region of a body part.

[0050] like Figure 1 As shown, probe 110 and conduit 105 can be connected to console 160. Console 160 may include computing device 161 employing an autoencoder as described herein. According to one embodiment, console 160 and / or computing device 161 include at least a processor and a memory, wherein the processor executes computer instructions relating to the autoencoder described herein, and the memory stores instructions for execution by the processor.

[0051] The computing device 161 can be any computing device including software and / or hardware, such as a general-purpose computer, having suitable front-end and interface circuitry 162 for transmitting and receiving signals to and from catheter 105, and other components for controlling system 100. The computing device 161 may include a real-time noise reduction circuitry system typically configured as a field-programmable gate array (FPGA), followed by an analog-to-digital (A / D) electrocardiogram (ECG) or electromyography (EMG) signal conversion integrated circuit. The computing device 161 can pass signals from the A / D ECG or EMG circuitry to another processor and / or can be programmed to perform one or more functions disclosed herein.

[0052] For example, one or more of these functions include receiving an input intracardiac signal, encoding the input intracardiac signal using an intracardiac dataset to generate a latent representation, and decoding the latent representation to generate an output intracardiac signal. The front-end and interface circuitry 162 includes an input / output (I / O) communication interface that enables the console 160 to receive signals from and / or transmit signals to at least one ablation electrode 134.

[0053] In some embodiments, computing device 161 may be further configured to receive biometric data, such as electrical activity, and determine whether a given tissue region is conductive. According to one embodiment, computing device 161 may be located external to console 160 and may be located, for example, in a catheter, an external device, a mobile device, a cloud-based device, or may be a standalone processor.

[0054] As described above, computing device 161 may include a general-purpose computer that can be software-programmed to perform the functions of the automatic encoder described herein. The software may be downloaded to the general-purpose computer electronically, for example, via a network, or alternatively or additionally set and / or stored on a non-transitory tangible medium such as magnetic storage, optical storage, or electronic storage (e.g., any suitable volatile and / or non-volatile memory, such as random access memory or hard disk drive). Figure 1 The exemplary configurations shown can be modified to implement the embodiments disclosed herein. The embodiments disclosed herein can be applied similarly using other system components and settings. Additionally, system 100 may include additional components such as elements for sensing electrical activity, wired or wireless connectors, processing and display devices, etc.

[0055] According to one embodiment, display 165 is connected to computing device 161. During the procedure, computing device 161 may facilitate the presentation of a body part rendering to a medical professional 115 on display 165 and store data representing the body part rendering in memory. In some embodiments, medical professional 115 may be able to manipulate the body part rendering using one or more input devices, such as a touchpad, mouse, keyboard, gesture recognition device, etc. For example, the input device may be used to change the position of catheter 105 so that the rendering is updated. In an alternative embodiment, display 165 may include a touchscreen that can be configured to accept input from medical professional 115 in addition to presenting the body part rendering. Display 165 may be located in the same location or a remote location, such as a single hospital or within a single healthcare provider network. Additionally, system 100 may be part of a surgical system configured to obtain anatomical and electrical measurements of a patient's organ, such as heart 120, and to perform cardiac ablation procedures. An example of such a surgical system is Carto, sold by Biosense Webster. ® system.

[0056] The console 160 can be connected via a cable to a surface electrode, which may include an adhesive skin patch attached to the patient 125. A processor, in conjunction with a current tracking module, determines the orientation coordinates of the catheter 105 within a body part of the patient 125 (e.g., the heart 120). The position coordinates may be based on impedance or electromagnetic field measured between the surface electrode and other electromagnetic components of the electrode or catheter 105 (e.g., at least one ablation electrode 134). Additionally or alternatively, a positioning pad may be located on the surface of the bed 130 and may be detachable from the bed 130.

[0057] System 100 may also, and optionally, use ultrasound, computed tomography (CT), magnetic resonance imaging (MRI), or other medical imaging techniques known in the art to acquire biometric data, such as anatomical measurements of the heart 120. System 100 may use catheters or other sensors that measure the electrical properties of the heart 120 to acquire ECG or electrical measurements. The biometric data, including anatomical and electrical measurements, can then be stored in a non-transitory tangible medium in console 160. The biometric data can be transferred from the non-transitory tangible medium to computing device 161. Alternatively or otherwise, biometric data can be transferred to a server 1760, which may be local or remote, using network 1762.

[0058] According to one or more embodiments, a catheter containing a position sensor can be used to determine the trajectory of points on the surface of the heart. These trajectories can be used to infer kinematic properties, such as the contractility of tissues. When trajectory information is sampled at a sufficient number of points in the heart 120, a mapping depicting such kinematic properties can be constructed.

[0059] Typically, a catheter 105 containing an electrical sensor at or near its distal end (at least one ablation electrode 134) is advanced to a point in the heart 120, where the sensor contacts the tissue and acquires data at that point, thereby measuring the electrical activity at that point in the heart 120. A drawback of mapping cardiac chambers using a catheter 105 containing only a single distal electrode is the time-consuming process of acquiring data point-by-point at the necessary number of points required for a detailed map of the overall chambers. Therefore, multi-electrode catheters have been developed to simultaneously measure electrical activity at multiple points within the cardiac chambers.

[0060] Multi-electrode catheters can be implemented using any suitable shape, such as linear catheters with multiple electrodes, balloon catheters comprising electrodes distributed across multiple ridges shaping the balloon, lasso or loop catheters with multiple electrodes, or any other suitable shape. Linear catheters can be fully or partially elastic, allowing them to twist, bend, or otherwise change shape based on received signals and / or based on external forces (e.g., cardiac tissue) applied to the linear catheter. Balloon catheters can be designed such that their electrodes remain in close contact with the endocardial surface when deployed into the patient. For example, balloon catheters can be inserted into lumens such as pulmonary veins (PVs). Balloon catheters can be inserted into the PV in a constricted state such that the balloon catheter does not occupy its maximum volume when inserted into the PV. Balloon catheters can inflate within the PV such that those electrodes on the balloon catheter contact the entire circular segment of the PV. Such contact with the entire circular portion of the PV or any other lumen enables effective mapping and / or ablation.

[0061] According to one example, a multi-electrode catheter can be advanced into a chamber of the heart at 120°. Anterior and posterior fluorescein (AP) and lateral fluorescein maps are obtained to establish the position and orientation of each electrode. The electrical activity map (EGM) relative to a time reference (e.g., starting from the P wave in the sinus rhythm from a surface ECG) can be recorded by each of the electrodes in contact with the cardiac surface. As further disclosed herein, the system can distinguish which electrodes record electrical activity from those that do not, due to their less close proximity to the endocardial wall. After recording the initial EGM, the catheter can be repositioned, and fluorescein maps and EGM can be recorded again. An electrogram can then be constructed iteratively based on the above process.

[0062] According to one example, cardiac mapping can be generated based on the detection of the intracardiac electrical potential field. Non-contact techniques for simultaneously acquiring large amounts of cardiac electrical information can be implemented. For example, a catheter with a distal end portion can be provided with a series of sensor electrodes distributed on its surface and connected to an insulating electrical conductor for connection to a signal sensing and processing device. The size and shape of the end portion allow the electrodes to be substantially spaced apart from the walls of the heart chambers. The intracardiac electrical potential field can be detected during a single heartbeat. According to one example, the sensor electrodes can be distributed on a series of circumferences located in planes spaced apart from each other. These planes can be perpendicular to the long axis of the end portion of the catheter. At least two additional electrodes can be provided adjacent to each other at the ends of the long axis of the end portion. As a more specific example, the catheter can include four circumferences, with eight electrodes spaced equally angularly apart on each circumference. Thus, in this specific embodiment, the catheter can include at least 34 electrodes (32 circumferential electrodes and 2 end electrodes).

[0063] According to another example, an electrophysiological cardiac mapping system and technique based on a non-contact and non-expanding multi-electrode catheter can be realized. EGM can be obtained via a catheter having multiple electrodes (e.g., 42 to 122 electrodes). According to this specific implementation, knowledge of the relative geometry of the probe and the endocardium can be obtained, for example, through independent imaging modalities (such as transesophageal echocardiography). After independent imaging, the non-contact electrodes can be used to measure the cardiac surface potential and construct a mapping map from it. The technique may include the following steps (after the independent imaging step): (a) measuring the potential using multiple electrodes disposed on a probe positioned on the heart 120; (b) determining the geometric relationship between the probe surface and the endocardial surface; (c) generating a coefficient matrix representing the geometric relationship between the probe surface and the endocardial surface; and (d) determining the endocardial potential based on the electrode potentials and the coefficient matrix.

[0064] According to another example, techniques and devices for mapping the potential distribution of cardiac chambers are available. An intracardiac multi-electrode mapping catheter assembly can be inserted into a patient's heart 120. The mapping catheter assembly may include a multi-electrode array with an integral reference electrode, or preferably, a mating reference catheter. The electrodes may be deployed in the form of a substantially spherical array. The electrode array may be spatially referenced to points on the endocardial surface via the reference electrode or via a reference catheter in contact with the endocardial surface. A preferred electrode array catheter may carry multiple individual electrode sites (e.g., at least 24). Furthermore, this exemplary technique is achieved by understanding the location of each electrode site in the array and understanding the cardiac geometry. These locations are preferably determined using techniques of impedance plethysmography.

[0065] According to another example, the cardiac mapping catheter assembly may include an electrode array defining multiple electrode sites. The mapping catheter assembly may also include a lumen to receive a reference catheter with a distal electrode assembly that can be used to probe the heart wall. The mapping catheter may include a braid of insulated wire (e.g., having 24 to 64 wires in the braid), and each wire can be used to form an electrode site. The catheter can be easily positioned in the heart 120 for obtaining electrical activity information from a first set of non-contact electrode sites and / or a second set of contact electrode sites.

[0066] According to another example, another catheter can be implemented for mapping electrophysiological activity within the heart. The catheter body may include a distal end adapted to deliver stimulation pulses for cardiac pacing or an ablation electrode for ablating tissue in contact with said end. The catheter may also include at least one pair of orthogonal electrodes to generate a differential signal indicative of local cardiac electrical activity adjacent to said orthogonal electrode.

[0067] According to another example, a process for measuring electrophysiological data in a heart chamber can be implemented. The method may partially include positioning a set of active and passive electrodes in the heart 120, supplying current to the active electrodes to generate an electric field in the heart chamber, and measuring the electric field at the site of the passive electrodes. The passive electrodes are contained in an array positioned on an inflatable balloon of a balloon catheter. In a preferred embodiment, the array reportedly has 60 to 64 electrodes.

[0068] According to another example, cardiac mapping can be achieved using one or more ultrasound transducers. The ultrasound transducer can be inserted into a patient's heart 120 and can collect multiple ultrasound slices (e.g., two-dimensional or three-dimensional slices) at various locations and orientations within the heart 120. The location and orientation of a given ultrasound transducer can be known, and the collected ultrasound slices can be stored so that they can be displayed at a later time. One or more ultrasound slices corresponding to the location of a probe (e.g., a treatment catheter) can be displayed after a period of time, and the probe can be overlaid on one or more ultrasound slices.

[0069] According to other examples, body patches and / or surface electrodes may be positioned on or near the patient's body. A catheter having one or more electrodes may be positioned within the patient's body (e.g., within the patient's heart 120), and the location of this catheter may be determined by the system based on signals transmitted and received between the one or more electrodes of the catheter and the body patch and / or surface electrodes. Additionally, the catheter electrodes may sense biometric data (e.g., LAT values) from within the patient's body (e.g., within the heart 120). This biometric data may be correlated with the determined location of the catheter, enabling the display of a rendering of the patient's body part (e.g., the heart 120) and the display of biometric data overlaid on the body shape.

[0070] See now Figure 2 A block diagram of an exemplary system 200 for remotely monitoring and transmitting biometric data (i.e., patient biometrics, patient data, or patient biometric data) is shown. Figure 2 In the example shown, system 200 includes a monitoring and processing device 202 (i.e., a patient data monitoring and processing device) associated with patient 204, a local computing device 206, a remote computing system 208, a first network 210, and a second network 211. According to one or more embodiments, the monitoring and processing device 202 may be... Figure 1 Example of catheter 105, patient 204 could be Figure 1 The example of patient 125, and the local computing device 206 may be Figure 1 Example of console 160.

[0071] The monitoring and processing device 202 includes a patient biometric sensor 212, a processor 214, a user input (UI) sensor 216, a memory 218, and a transmitter-receiver (i.e., transceiver) 222. In operation, the monitoring and processing device 202 acquires biometric data of the patient 204 (e.g., electrical signals, blood pressure, temperature, blood glucose levels, or other biometric data), and / or receives from one or more other patient biometric monitoring and processing devices at least a portion of biometric data representing any acquired patient biometrics and additional information associated with the acquired patient biometrics. The additional information may be, for example, diagnostic information and / or additional information obtained from an auxiliary device such as a wearable device.

[0072] The monitoring and processing device 202 may employ the autoencoder described herein to process data, including acquired biometric data and any biometric data received from one or more other patient biometric monitoring and processing devices. For example, in this regard, the autoencoder may include a neural network for learning latent representations (or data encodings) from the biometric data in an unsupervised manner when processing the data. Furthermore, the autoencoder can learn to detect specific data by training the neural network to ignore signal interference, signal artifacts, and signal noise by considering a clean dataset, without requiring pre-programming with specific rules.

[0073] Monitoring and processing device 202 can continuously or periodically monitor, store, process, and transmit any number of various patient biometrics (e.g., acquired biometric data) via network 210. Examples of patient biometrics, as described herein, include electrical signals (e.g., ECG signals and brain biometrics), blood pressure data, blood glucose data, and temperature data. Patient biometrics can be monitored and transmitted for the treatment of any number of various diseases, such as cardiovascular diseases (e.g., arrhythmias, cardiomyopathy, and coronary artery disease) and autoimmune diseases (e.g., type I and type II diabetes).

[0074] The patient biometric sensor 212 may include, for example, one or more transducers configured to convert one or more environmental conditions into electrical signals, thereby acquiring different types of biometric data. For example, the patient biometric sensor 212 may include one or more electrodes, temperature sensors (e.g., thermocouples), blood pressure sensors, blood glucose sensors, blood oxygen sensors, pH sensors, accelerometers, and microphones configured to acquire electrical signals (e.g., heart signals, brain signals, or other bioelectrical signals).

[0075] As described in more detail herein, the monitoring and processing device 202 may be for monitoring the heart (e.g., Figure 1 An ECG monitor that receives ECG signals from the heart (120). In this regard, the patient biometric sensor 212 of the ECG monitor may include one or more electrodes for acquiring ECG signals (e.g., ...). Figure 1 (The electrode in catheter 105). ECG signals can be used to treat various cardiovascular diseases.

[0076] In another example, the monitoring and processing device 202 could be a continuous glucose monitor (CGM) for continuously monitoring a patient's blood glucose levels for ongoing treatment of various conditions, such as type 1 and type 2 diabetes. In this regard, the CGM's patient biometric sensor 212 could include subcutaneously placed electrodes (e.g., Figure 1 The catheter 105 contains an electrode that monitors blood glucose levels from the patient's interstitial fluid. The CGM can be, for example, a component of a closed-loop system where blood glucose data is sent to an insulin pump to calculate insulin delivery without user intervention.

[0077] Processor 214 may be configured to receive, process, and manage biometric data acquired by patient biometric sensor 212, and transmit the biometric data to memory 218 via transceiver 222 for storage and / or across network 210. Data from one or more other monitoring and processing devices 202 may also be received by processor 214 via transceiver 222, as described in more detail herein. As described in more detail herein, processor 214 may be configured to selectively respond to different tap patterns (e.g., single or double taps) received from UI sensor 216 (e.g., a capacitive sensor therein), enabling different tasks of the patch (e.g., data acquisition, storage, or transmission) to be activated based on the detected pattern. In some embodiments, processor 214 may generate audible feedback relative to the detected gesture.

[0078] UI sensor 216 may include, for example, a piezoelectric or capacitive sensor configured to receive user input such as a tap or touch. For example, in response to a patient 204 tapping or touching the surface of monitoring and processing device 202, UI sensor 216 may be controlled to achieve capacitive coupling. Gesture recognition can be achieved via any of a variety of capacitance types, such as resistive capacitance, surface capacitance, projected capacitance, surface acoustic waves, piezoelectric, and infrared touch. The capacitive sensor may be positioned over a small area or along the length of a surface, such that a tap or touch on the surface activates the monitoring device.

[0079] Memory 218 is any non-transitory tangible medium, such as magnetic memory, optical memory, or electronic memory (e.g., any suitable volatile and / or non-volatile memory, such as random access memory or hard disk drive). According to one or more embodiments, memory 218 may store processor-executable code, software, or instructions for training algorithms and autoencoders.

[0080] Transceiver 222 may include a separate transmitter and a separate receiver. Alternatively, transceiver 222 may include a transmitter and receiver integrated into a single device.

[0081] According to one embodiment, the monitoring and processing device 202 may be a device located within the patient 204 (e.g., subcutaneously implantable). The monitoring and processing device 202 may be inserted into the patient 204 via any applicable means, including oral injection, surgical insertion via vein or artery, endoscopic procedure, or laparoscopic procedure.

[0082] According to one embodiment, the monitoring and processing device 202 may be a device located outside the patient 204. For example, as described in more detail herein, the monitoring and processing device 202 may include an attachable patch (e.g., which is attached to the patient's skin). The monitoring and processing device 202 may also include a catheter, probe, blood pressure cuff, scale, bracelet or smartwatch biometric tracker, glucose monitor, continuous positive airway pressure (CPAP) machine, or virtually any device that can provide input related to the patient's health or biometrics, having one or more electrodes.

[0083] According to one implementation, the monitoring and processing device 202 may include components inside the patient and components outside the patient.

[0084] Although Figure 2 A single monitoring and processing device 202 is shown, but the exemplary system may include multiple patient biometric monitoring and processing devices. For example, monitoring and processing device 202 may communicate with one or more other patient biometric monitoring and processing devices. In addition, or alternatively, one or more other patient biometric monitoring and processing devices may communicate with network 210 and other components of system 200.

[0085] The local computing device 206 and / or the remote computing system 208, together with the monitoring and processing device 202, can be any combination of software and / or hardware that stores, executes, and implements the automatic encoder and its functions, either individually or jointly. Furthermore, the local computing device 206 and / or the remote computing system 208, together with the monitoring and processing device 202, can be an electronic computer framework, including and / or employing any number and combination of computing devices and networks utilizing various communication technologies, as described herein. The local computing device 206 and / or the remote computing system 208, together with the monitoring and processing device 202, can be easily scaled, expanded, and modularized, with the ability to change for different services or reconfigure some features independently of others.

[0086] According to one embodiment, the local computing device 206 and the remote computing system 208, together with the monitoring and processing device 202, may include at least a processor and a memory, wherein the processor executes computer instructions concerning the automatic encoder, and the memory stores the instructions to be executed by the processor.

[0087] The local computing device 206 of system 200 communicates with the monitoring and processing device 202 and can be configured to act as a gateway to the remote computing system 208 via a second network 211. For example, the local computing device 206 may be a smartphone, smartwatch, tablet, or other portable smart device configured to communicate with other devices via network 211. Alternatively, the local computing device 206 may be a fixed or standalone device, such as a fixed base station including, for example, modem and / or router capabilities, a desktop or laptop computer using an executable program to transmit information between the processing device 202 and the remote computing system 208 via the radio module of a PC, or a USB dongle. Biometric data can be transmitted between the local computing device 206 and the monitoring and processing device 202 via a short-range wireless network 210, such as a local area network (LAN) (e.g., a personal area network (PAN)), using short-range wireless technology standards (e.g., Bluetooth, Wi-Fi, ZigBee, Z-wave, and other short-range wireless standards). In some implementations, the local computing device 206 may also be configured to display the acquired patient electrical signals and information associated with the acquired patient electrical signals, as described in more detail herein.

[0088] In some implementations, the remote computing system 208 may be configured to receive at least one of monitored patient biometrics and patient-associated information via a network 211, which is a remote network. For example, if the local computing device 206 is a mobile phone, the network 211 may be a wireless cellular network, and information may be transmitted between the local computing device 206 and the remote computing system 208 via wireless technology standards such as any of the wireless technologies described above. As described in more detail herein, the remote computing system 208 may be configured to provide (e.g., visually displayed and / or audibly provided) patient biometrics and related information to healthcare professionals, physicians, or nurses.

[0089] exist Figure 2 In this context, network 210 is an example of a near-field network (e.g., a local area network (LAN) or a personal area network (PAN)). Information can be transmitted between monitoring and processing device 202 and local computing device 206 via near-field network 210 using any of a variety of near-field wireless communication protocols such as Bluetooth, Wi-Fi, Zigbee, Z-Wave, Near Field Communication (NFC), Ultraband, Zigbee, or Infrared (IR)).

[0090] Network 211 can be a wired network, a wireless network, or a network comprising one or more wired and wireless networks, such as an intranet, local area network (LAN), wide area network (WAN), metropolitan area network (MAN), direct connection or a series of connections, cellular telephone network, or any other network or medium capable of facilitating communication between local computing device 206 and remote computing system 208. Information can be transmitted via network 211 using any of a variety of remote wireless communication protocols (e.g., TCP / IP, HTTP, 3G, 4G / LTE, or 5G / new radio). Wired connections can be implemented using Ethernet, Universal Serial Bus (USB), RJ-11, or any other wired connection known in the art. Wireless connections can be implemented using Wi-Fi, WiMAX and Bluetooth, infrared, cellular networks, satellite, or any other wireless connection method. Additionally, several networks can operate independently or communicate with each other to facilitate communication within network 211. In some cases, remote computing system 208 can be implemented as a physical server on network 211. In other cases, remote computing system 208 can be implemented as a public cloud computing provider (e.g., Amazon Web Services (AWS)) on network 211. ® () virtual server.

[0091] Figure 3 Artificial intelligence system 300 according to one or more embodiments is shown. Artificial intelligence system 300 may include data 310, machine 320, model 330, multiple results 340 and underlying hardware 350. Figure 4 It shows in Figure 3 Method 400 is executed in an artificial intelligence system. For ease of understanding, refer to... Figure 2 right Figure 3 and Figure 4 Describe it.

[0092] Typically, the artificial intelligence system 300 trains the machine 320 using data 310 (e.g., Figure 2 The local computing device 206 simultaneously constructs a model 330 to achieve multiple results 340 (predictable) to operate the method 400. In this configuration, the artificial intelligence system 300 can operate relative to the hardware 350 (e.g., Figure 2 The monitoring and processing device 202 operates to train the machine 320, build a model 330, and use algorithms to predict results. These algorithms can be used to solve the trained model 330 and predict results 340 associated with the hardware 350. These algorithms can generally be categorized into classification, regression, and clustering algorithms.

[0093] At box 410, method 400 may include collecting data 310 from hardware 350. Machine 320 may operate and / or be associated with hardware 350 as a controller or data collection device. Data 310 (e.g., may originate from...) Figure 2 The biometric data from the monitoring and processing device 202 may be associated with the hardware 350. For example, data 310 may be data being generated or output data associated with the hardware 350. The data 310 may also include data currently collected from the hardware 350, historical data, or other data. For example, the data 310 may include measurements taken during a surgical procedure and may be associated with the outcome of the surgical procedure. For example, the temperature of the heart (e.g., the heart of patient 204) may be collected and associated with the outcome of a cardiac procedure.

[0094] At box 420, method 400 includes, for example, training machine 320 relative to hardware 350. This training may include analysis and correlation of data 310 collected in box 410. For example, in the case of the heart, the temperature and outcome data 310 may be trained to determine whether there is a correlation or association between the temperature of the heart (e.g., the heart of patient 204) and the outcome during cardiac procedures.

[0095] At box 430, method 400 may include constructing a model 330 based on data 310 associated with hardware 350. Constructing the model 330 may include physical hardware or software modeling, algorithmic modeling, etc. This modeling may attempt to represent the collected and trained data 310. According to one embodiment, model 330 may be configured to model the operation of hardware 350 and the data 310 collected from hardware 350 in order to predict outcomes achieved by hardware 350. According to one or more embodiments, model 330 may distinguish between ventricular far-field and atrial-based activation relative to an autoencoder, and generate differentiated mappings for atrial and ventricular activation.

[0096] At box 440, method 400 may include predicting multiple outcomes 340 of model 330 associated with hardware 350. Such predictions of multiple outcomes 340 may be based on a trained model 330. For example, to enhance understanding of this disclosure, in the case of the heart, a positive outcome from a cardiac procedure is produced if the temperature during the procedure is between 36.5 degrees Celsius and 37.89 degrees Celsius (i.e., 97.7 degrees Fahrenheit and 100.2 degrees Fahrenheit), which can be predicted in a given procedure based on the heart temperature during the cardiac procedure. Therefore, using the predicted outcomes 340, hardware 350 may be configured to provide a desired outcome 340 from hardware 350.

[0097] Now go to Figure 5An example of a neural network 500 according to one or more embodiments is shown. The neural network 500 can operate as a specific implementation of an autoencoder. The neural network 500 can be implemented in hardware such as machine 320 (e.g., Figure 2 The local computing device 206) and / or hardware 350 (e.g., Figure 2 This is implemented in the monitoring and processing equipment 202. A neural network is a network or circuit of neurons, or in the modern sense, an artificial neural network composed of artificial neurons or nodes.

[0098] For example, an ANN can involve a network of processing elements (artificial neurons) that can exhibit complex global behavior determined by the connections between the processing elements and their parameters. These connections in a network or circuit of neurons can be modeled as weights. Positive weights reflect excitatory connections, while negative values ​​represent inhibitory connections. The input can be modified by the weights and summed using a linear combination. The activation function controls the amplitude of the output. For example, an acceptable output range is typically between 0 and 1, or it could be between -1 and 1.

[0099] In most cases, ANNs are adaptive systems that change their structure based on information flowing through the network's external or internal systems. In more practical terms, neural networks are nonlinear statistical data modeling or decision-making tools that can be used to model complex relationships between inputs and outputs or to find patterns in data. Therefore, ANNs can be used for predictive modeling and adaptive control applications while being trained on datasets. Experience-based self-learning can occur within ANNs, drawing conclusions from complex and seemingly unrelated groups of information. The utility of artificial neural network models lies in their ability to infer functions from observations and also to use those functions. Unsupervised neural networks can also be used to learn representations of inputs that capture salient features of the input distribution, and recent deep learning algorithms can implicitly learn the distribution function of observed data. Learning in neural networks is particularly useful in applications where the complexity of the data or task makes manually designing such functions impractical.

[0100] Neural networks can be used in various fields. The tasks applied to ANNs often fall into the following broad categories: function approximation or regression analysis, including time series forecasting and modeling; classification, including pattern and sequence recognition, novelty detection, and order decision-making; and data processing, including filtering, clustering, blind signal separation, and compression.

[0101] Applications of ANNs can include nonlinear system recognition and control (vehicle control, process control), gaming and decision-making (backgammon, chat, competitions), pattern recognition (radar systems, facial recognition, object recognition), sequence recognition (gesture, speech, handwritten text recognition), medical diagnosis, financial applications, data mining (or knowledge discovery in databases, "KDD"), visualization, and email spam filtering. For example, semantic feature maps of user interests can be created from images trained for object recognition.

[0102] Now go to Figure 6 A block diagram of method 600 according to one or more embodiments is shown. Method 600 depicts the operation of neural network 500 (e.g., an autoencoder). [Go to...] Figure 5 In neural network 500, the input layer 510 is represented by multiple inputs such as 512 and 514. Relative to... Figure 6 In box 610, input layer 510 can receive the multiple inputs (e.g., input intracardiac signals) as initial operations. The multiple inputs can be ultrasound signals, radio signals, audio signals, or two-dimensional images. More specifically, the multiple inputs can be represented as input data (X), which is raw data recorded from the atria. The desired information may be located in the high-frequency region of the heart (e.g., the atria), and the autoencoder provides a better construct of the input intracardiac signals. According to one or more embodiments, the multiple inputs can be a combination of intracardiac ECG and surface ECG (to remove far-field noise from the intracardiac signals).

[0103] exist Figure 6 At box 620, neural network 500 can encode an input intracardiac signal using an intracardiac dataset to produce a latent representation. This latent representation may include one or more intermediate images derived from the input intracardiac signal. According to one or more embodiments, this latent representation is generated by an element-wise activation function (e.g., a sigmoid function or a rectified linear unit) of an autoencoder, which applies a weight matrix to the input intracardiac signal and adds a bias vector to the result. The weights and biases of the weight matrix and bias vector can be randomly initialized and then iteratively updated during training.

[0104] The intracardiac dataset can be a training dataset or clean data, comprising pre-determined and approved signals (i.e., clean examples) free from interference, artifacts, and noise. In one implementation, expert medical professionals, physicians, etc., can examine, edit, and remove signal interference, signal artifacts, and signal noise, and approve each electrical signal in the intracardiac dataset. In one implementation, the intracardiac data may contain approximately thousands or more electrical signals, where template matching and blanking are used to examine the signal morphology of each electrical signal. For example, denoising of any IC-ECG artifact can be performed using an intracardiac dataset (e.g., a database of "clean versions" of intracardiac ECG signals). Given the number of electrical signals and the complexity of examination, editing, and approval, the creation of the intracardiac dataset can be viewed as a data training component of a multi-step data manipulation performed by an autoencoder.

[0105] like Figure 5 As shown, inputs 512 and 514 are provided to a hidden layer 530, depicted as including nodes 532, 534, 536, and 538. This encoding provides a dimensionality-reduced input signal. Dimensionality reduction is the process of reducing the number of random variables (of multiple inputs) under consideration by obtaining a set of principal variables. For example, dimensionality reduction can be feature extraction that transforms data (e.g., multiple inputs) from a high-dimensional space (e.g., more than 10 dimensions) to a low-dimensional space (e.g., 2-3 dimensions). The beneficial effects of dimensionality reduction techniques include reducing the time and storage space of data, improving data visualization, and improving parameter interpretation for machine learning. This data transformation can be linear or non-linear. The operations of receiving (box 610) and encoding (box 620) can be viewed as the data preparation part of a multi-step data manipulation performed by an autoencoder.

[0106] According to one implementation plan, data preparation may also include the collection of intracardiac electrocardiogram (IC-ECG) data from the atria (the upper chamber through which blood enters the heart), while simultaneously recording from the ventricles (the two lower chambers of the heart).

[0107] exist Figure 6 At frame 630, neural network 500 decodes the latent representation to generate the output intracardiac signal. In the case of IC-ECG, the output intracardiac signal can be a far-field estimate of the ventricle. For example... Figure 5As shown, nodes 532, 534, 536, and 538 can be combined to produce output 552 in output layer 550, where output layer 550 can reconstruct inputs 512 and 514 in a reduced dimension but without signal interference, signal artifacts, and signal noise. Neural network 500 can perform processing via hidden layer 530 of nodes 532, 534, 536, and 538 to exhibit complex global behavior determined by the connections between processing elements and element parameters. The target data of output layer 550 can include target data type one ventricular activity (Y1) and target data type two input data after far-field attenuation (Y2). The far field can cause problems regarding the generation and navigation of 3D mapping (e.g., the ventricular far field can interfere with atrial activation). Therefore, the technical effects and benefits of the autoencoder employing neural network 500 include improved accuracy of 3D mapping due to artifact removal (relative to the far field).

[0108] According to one or more implementation schemes, the model of the autoencoder using a neural network 500 can distinguish between ventricular far-field and atrial-based activations, and generate differentiated mappings for atrial and ventricular activations.

[0109] According to one implementation, the autoencoder can be a denoising autoencoder to find a measurement function (f, g) such that f(X) = Y1 and g(X) = Y2. In this regard, the task of the autoencoder can be to learn a measurement from X to X by reducing some dimension of the input X (e.g., constructing two neural networks (F, G) such that h = F(X) and X = G(h)). The dimension of h is smaller than the dimension of X. In the denoising autoencoder, although the architecture is similar, the denoising autoencoder learns a measurement from X to Y, where Y is a denoised version of X.

[0110] See Figure 7The diagram illustrates a graphical depiction of signal 700 according to one or more embodiments. As shown by signal 700, the ECG signal comprises a P wave 710 (due to atrial depolarization), a QRS complex 720 (due to atrial repolarization and ventricular depolarization), and a T wave 730 (due to ventricular repolarization). The ECG signal is generated by the contraction (depolarization) and relaxation (repolarization) of the atrial and ventricular myocardium of the heart. To record the ECG signal, electrodes may be placed at specific locations on the body or positioned within the body via a catheter. Artifacts (e.g., noise) are unwanted signals that are combined with electrical signals such as ECG signals and can sometimes hinder the diagnosis and / or treatment of cardiac conditions. Artifacts in electrical signals can include baseline drift, power line interference, EMG noise, power line noise, etc. In other words, examples of artifacts include, but are not limited to, power noise (e.g., electrostatic and electromagnetic coupling between the circuit and a 50 or 60 Hz power line), Fluro noise (e.g., fluorescent lamps), contact noise (e.g., collisions between conduit electrodes), and deflection noise (e.g., electrostatic discharge during conduit deflection).

[0111] Baseline drift can occur at the base axis (x-axis) of a signal, appearing "drifted" or moving up and down instead of straight. This can cause the entire signal to deviate from its normal base. In ECG signals, baseline drift can be caused by inappropriate electrode contact (e.g., electrode-skin impedance), patient movement, and periodic movements (e.g., breathing).

[0112] Figure 8 A graphical depiction of signal 810, shown in graph 800 according to one or more embodiments, is illustrated. For this purpose, signal 800 is a typical ECG signal affected by baseline drift 820. The frequency content of the baseline drift is in the range of 0.5 Hz. Increased body movement during exercise or stress testing increases the frequency content of the baseline drift. Depending on the specific embodiment, given that the baseline signal is a low-frequency signal, a finite impulse response (FIR) high-pass zero-phase forward-backward filter with a cutoff frequency of 0.5 Hz can be used to estimate and remove the baseline drift 820 in the ECG signal 810.

[0113] Electromagnetic fields induced by electric field lines represent a common noise source in electrical signals such as ECGs, as well as any other bioelectrical signals recorded from a patient's body. This type of noise is characterized by, for example, 50 or 60 Hz sinusoidal interference, possibly accompanied by multiple harmonics. This narrowband noise makes ECG analysis and interpretation more difficult because depictions of low-amplitude waveforms become unreliable and may introduce spurious waveforms. When ECG signals are superimposed with low-frequency ECG waves such as P-wave 710 and T-wave 730, it may be necessary to remove power line interference from these ECG signals.

[0114] The presence of muscle noise can interfere with many electrical signal applications, such as ECG, because low-amplitude waveforms can become blurred. Unlike baseline drift 820 and 50 / 60Hz interference, muscle noise does not present a different filtering problem because the spectral content of muscle activity significantly overlaps with that of the PQRST complex 720. Since the ECG signal 810 is a repetitive signal, techniques can be used to reduce muscle noise in a manner similar to the treatment of evoked potentials. Figure 9 A graphical depiction 900 of a signal 905 shown according to one or more embodiments is illustrated. In this regard, signal 905 is an ECG signal interfered with by EMG noise 910.

[0115] Instruments used to measure electrical signals such as ECG signals typically detect electrical interference corresponding to line or trunk frequencies. Although nominally set at 50Hz or 60Hz, line frequencies in most countries can differ from these nominal values ​​by several percentage points.

[0116] Various techniques can be implemented to remove electrical interference from electrical signals. Several of these techniques utilize one or more low-pass or notch filters. For example, a system for variable filtering of noise in ECG signals can be implemented. This system may have multiple low-pass filters, including, for example, a filter with a 3dB point at approximately 50Hz and a second low-pass filter with a 3dB point at approximately 5Hz.

[0117] According to another example, a system for rejecting line frequency components of an electrical signal can be implemented by passing the signal through two notch filters connected in series. A system with notch filters can be implemented, which may have either a low-pass coefficient and a high-pass coefficient for removing line frequency components from the ECG signal, or both. The system may also support the removal of burst noise and the calculation of heart rate from the notch filter output.

[0118] According to another example, a system having several units can be implemented for removing interference. The units may include an averaging unit for generating an average signal over several cardiac cycles, a subtraction unit for subtracting the average signal from an input signal to generate a residual signal, a filter unit for providing a filtered signal from the residual signal, and / or an addition unit for adding the filtered signal to the average signal.

[0119] According to another example, an analog-to-digital (A / D) converter can provide noise suppression by synchronizing the converter's clock with a phase-locked loop set to the line frequency.

[0120] Additionally, biometric (e.g., biopotential) patient monitors can use surface electrodes to measure biopotentials, such as ECG or EEG. The fidelity of these measurements is limited by the effectiveness of the electrode connection to the patient. The resistance of the electrode system to the flow of current (called impedance) characterizes the effectiveness of the connection. Generally, the higher the impedance, the lower the fidelity of the measurement. Several mechanisms can contribute to lower fidelity.

[0121] Signals from electrodes with high impedance are affected by thermal noise (or so-called Johnson noise), which is a voltage that increases with the square root of the impedance value. Furthermore, voltage noise from biopotential electrodes often exceeds the voltage noise predicted by Johnson. Additionally, amplifier systems measured by biopotential electrodes may exhibit reduced performance at higher electrode impedances. Damage is characterized by poor common-mode rejection, which increases contamination of bioelectrical signals by noise sources such as patient movement and electronic devices that may be used on or around the patient. These noise sources are particularly prevalent in protocol rooms and may include equipment such as electrosurgical units (ESUs), cardiopulmonary bypass pumps (CPBs), electrically driven surgical saws, lasers, and other sources.

[0122] During cardiac procedures, it is generally desirable to continuously measure electrode impedance in real time while monitoring the patient. This is typically done by injecting a very small current into the electrodes and measuring the resulting voltage, thus establishing impedance using Ohm's law. This current can be injected using a DC or AC source. Due to electrode impedance, it is often impossible to separate the voltage from voltage artifacts caused by interference. Interference increases the measured voltage and therefore the measured apparent impedance, causing the biopotential measurement system to erroneously detect an impedance higher than actually present. Typically, such monitoring systems have a maximum impedance threshold limit, which can be programmed to prevent operation if it detects impedance exceeding these limits. This is especially true for systems performing measurements of very small voltages, such as EEG. Such systems require very low electrode impedance.

[0123] High-resolution intracardiac electrograms (EGM) can guide cardiac ablation procedures. Cardiac ablation can be used to treat ventricular tachycardia (VT), where rapid and irregular heartbeats are caused by complex electrophysiological (EP) circuitry and reentry within the ventricle. Therefore, catheter ablation aims to target the origin of VT. Mapping the VT circuitry and identifying its source are crucial for the success of VT ablation. A major challenge in interpreting intracardiac EGM in the presence of VT is that the EGM signal can have complex morphologies, making it difficult to extract the local activation time (LAT) with sufficiently high spatial resolution. This, in turn, makes it difficult to accurately map the complex circuitry within the ventricle, which is essential for locating the relevant ablation target.

[0124] Compared to bipolar signals, unipolar EGM signals typically have a much lower signal-to-noise ratio, making bipolar signals the current primary tool for LAT extraction. However, unipolar signals may offer better spatial and temporal resolution, which can significantly improve the calibration of VT circuits. Therefore, advanced digital signal processing (DSP) methods can be applied to extract accurate LATs from noisy unipolar signals. DSP methods or systems are designed to reduce or attenuate noise from a signal and can include a variety of digital filters. Linear smoothing filters (e.g., low-pass or high-pass filters) or any other smoothing operator that can convolve with the signal can be used to reduce or attenuate noise. Nonlinear filters (e.g., median filters for noise reduction) can be used to reduce or attenuate noise. Wavelet transforms, which achieve both noise reduction and feature preservation, can be used. Statistical denoising methods can be used, which can use ambient or neighboring signals or any other modality to reduce unwanted components in the signal.

[0125] The bipolar signal originates from two adjacent monopole electrodes. The monopole signal originates from the monopole electrode and the reference electrode, and is a combination of far-field and near-field contributions. The main noise source in the monopole signal is the far-field signal generated by voltage depolarization of the distant tissue. Due to the large distance between the monopole electrode and the reference electrode, the far-field signal is usually not fully accounted for and is not completely removed from the signal compared to the bipolar signal. Because the two monopole signals forming the monopole pair have very similar far fields, the difference between them is almost zero except for the case where there is local activity at each monopole in the monopole signal. This local activity is called the near-field signal and can be indicated by small peaks on the bipolar signal.

[0126] When a monopolar electrode is located beneath scar tissue that does not generate electrical activity, the near field can have a lower amplitude than the far field compared to the case caused by healthy tissue. This makes it particularly difficult to distinguish between far-field and near-field signals using classical DSP methods. In this case, the near field can be very low and negligible. Therefore, the bipolar signal may not have any activation and can be effectively zero. This type of monopolar signal can be represented as a pure far-field signal because there is no obvious local activity. In this case, the bipolar signal may appear flat, and the two monopolar signals may be almost identical. These types of signals can be obtained by placing the catheter in a location that does not contact the heart muscle or as surface ECG signals, which are essentially far-field signals. These types of monopolar signals can be used as training datasets to enable neural networks to learn this type of activity and distinguish the far-field component from the mixed monopolar signal.

[0127] While the far-field contribution is considered noise to be removed in the current context, it can contain useful information in other contexts. This provides an additional motivation to separate these two types of contributions.

[0128] Deep learning (DL), based on deep neural networks (DNNs), has become a disruptive technology in the application of computer algorithms to various fields, such as computer vision and DSP. DL allows the extraction of complex patterns and data from signals and images, often in situations where such extraction was previously impossible or only possible through time-consuming manual analysis. Therefore, applying DL to intracardiac EGM is particularly attractive, where reducing procedural time and increasing clinical success rates are key objectives.

[0129] Machine learning (ML) is a set of algorithms and statistical models used for data analysis to perform specific tasks. Deep learning (DL) is a subset of machine learning algorithms that sets model parameters during the training process to allow accurate prediction of the desired output on unseen data. ML and DL techniques allow the analysis of highly complex spatiotemporal information that is difficult for classical algorithms to analyze. While machine learning is typically based on feature extraction using a heuristic list about the data, DL is based on learning from examples and typically does not require feature extraction from the data. The main difference between DL and traditional ML is that the training process requires a large amount of data. Given a sufficient amount of data, DL-based algorithms generally outperform traditional ML algorithms.

[0130] Therefore, DL is a useful tool for decomposing the near-field and far-field components in ECG signals, and specifically in VT signals. This allows for activation detection of only near-field activity. This is useful because, when mapping ventricular activity, the far-field can be stronger and mask near-field activity, thus misleading the annotation mechanism. This is also useful in the case of atrial fibrillation (AFIB), because the strong ventricular signal may be incorrectly annotated as atrial activity.

[0131] Therefore, it is expected that DL methods will shorten the overall clinical process by providing medical professionals (e.g., cardiologists and electrophysiologists) with insights that can currently only be obtained through manual data analysis by trained clinicians, and identify deep data patterns that cannot currently be identified manually or using classical algorithms (e.g., DSP and computer vision), and thus allow for the identification of ablation targets in more complex situations that are currently untreatable.

[0132] DL training can be unsupervised. That is, while a large body of pre-recorded unipolar EGM signals exists, and additional signals can be collected if needed, a major challenge in applying DL to remove far-field noise is the lack of baseline truth data for training DL models. Any far-field and near-field signal decomposition is an evaluation and does not necessarily correspond to the true far-field and near-field signals at a specific electrode. Therefore, DL methods can be unsupervised rather than supervised. Surface ECG can be used with distal electrodes as baseline truth for the far-field components.

[0133] Figure 10 Graphical depictions (10A, 10B, 10C, 10D, 10E and 10F) of a far-field removal signal process 1000 according to one or more embodiments are shown. Figure 10 The signals in 10A to 10E are intracardiac (IC) ECG signals recorded from different locations along the coronary sinus (CS). Figure 10 In 10A, signal 1021 represents the surface IC ECG signal. Boundary line 1032 represents the Local Activation Time (LAT). Boundary line 1032 also exists Figure 10 In 10C, 10D, 10E, and 10F. Boundary line 1043 indicates the QRS location. Boundary line 1043 also exists in... Figure 10 In 10A, 10C, 10D, 10E, and 10F. Figure 10 In the diagram, the X-axis represents time, while the Y-axis represents mV.

[0134] like Figure 10 As shown, 1054 represents the far-field component of the IC ECG signal. Figure 10 10C, 10D, 10E, and 10F illustrate the process of increasing the amount of far-field removal in signal 1054, where signal 1065 represents the IC ECG signal after far-field removal. Far-field removal can be achieved, for example, by creating a blanking period, during which IC ECG signal 1065 may be zero.

[0135] Figure 11 A block diagram of method 1100 according to one or more embodiments is shown. According to one embodiment, method 1100 can be implemented by a noise-reducing autoencoder. Any combination of software and / or hardware (e.g., local computing device 206 and remote computing system 208 together with monitoring and processing device 202) can store, execute, and implement the noise-reducing autoencoder and its functions individually or jointly. The noise-reducing autoencoder can be trained to reconstruct input from its own corrupted version in order to force the hidden layer (e.g., Figure 5The hidden layer 530 discovers more robust features (i.e., useful features that will constitute a better, higher-level representation of the input) and prevents it from learning properties (i.e., always returning to the same values). In this respect, the denoising autoencoder can encode the input (e.g., to retain information about the input) and reverse the effects of the corrupting process of the input randomly applied to the autoencoder.

[0136] According to one or more embodiments, the denoising autoencoder can implement a long short-term memory neural network architecture, a convolutional neural network architecture, or other similar architecture. The architecture of the denoising autoencoder can be configured relative to multiple layers, multiple connections (e.g., encoder / decoder connections), regularization techniques (e.g., differential pressure or batch normalization), and optimized features.

[0137] Long Short-Term Memory (LSTM) neural network architectures may include feedback connections and can process single data points (e.g., such as images) as well as entire sequences of data (e.g., such as speech or video). A unit in an LSM neural network architecture may consist of a cell, an input gate, an output gate, and a forget gate, where the cell remembers a value over any time interval, and the gates regulate the flow of information into and out of the cell.

[0138] A convolutional neural network (CNN) architecture can be a shared-weight architecture with translation invariance, where each neuron in one layer is connected to all neurons in the next layer. Regularization techniques for CNN architectures can leverage hierarchical patterns in the data and assemble more complex patterns using smaller, simpler ones. If a denoising autoencoder implements a CNN architecture, other configurable aspects of that architecture can include the number of filters at each stage, the kernel size, and the number of kernels per layer.

[0139] Method 1100 begins at box 1105, where the denoising autoencoder receives a “clean and approved” intracardiac dataset from multiple electrical signals. As noted herein, expert medical professionals, physicians, etc., can examine and edit the dataset to remove signal interference, signal artifacts, and signal noise, and approve each electrical signal in the intracardiac dataset. At box 1110, the denoising autoencoder builds a model based on the clean and approved intracardiac dataset (e.g., Figure 3 Model 330).

[0140] At box 1115, the noise-reducing autoencoder can receive an input intracardiac signal that includes at least far-field artifacts. The input intracardiac signal can be recorded by one or more monitoring and processing devices (e.g., a five-ray catheter with twenty electrodes, a basket catheter with sixty-four electrodes, multiple surface leads, etc.). The far field can cause problems regarding the generation and navigation of 3D mapping (i.e., ventricular far field can interfere with atrial activation).

[0141] At box 1120, the denoising autoencoder encodes the input intracardiac signal using a model (from box 1110). This encoding provides a dimensionality-reduced input intracardiac signal by indicating how the model should reduce dimensionality, removing at least far-field artifacts. The result of the encoding is a latent representation. At box 1130, the denoising autoencoder decodes the latent representation to produce an output intracardiac signal.

[0142] At box 1135, the noise-reducing autoencoder can map the output intracardiac signal. For example, the noise-reducing autoencoder (using its underlying architecture) finds a mapping function (f, g) such that f(X) = Y1 and g(X) = Y2.

[0143] At box 1140, an ECG can be generated based on the mapped output intracardiac signal. The ECG can be generated by a computing device performing a noise-reducing autoencoder or by another device. The improved ECG, which has been improved due to the removal of signal interference, signal noise, and signal artifacts, can then be presented to medical professionals. The improved ECG can significantly reduce the time spent on each cardiac case.

[0144] As noted herein, during intracardiac electrocardiography mapping, the mapping catheter can record both atrial and ventricular activation. In some cases, the ventricular far field can interfere with atrial activation (e.g., signal interference), which can affect the clinical understanding and interpretation of Carto mapping. According to one or more embodiments, the technical effects and benefits of a noise-reducing autoencoder may include distinguishing between ventricular far field and atrial-based activation, and generating differentiated mappings for atrial and ventricular activation (e.g., the noise-reducing autoencoder uses a model during decoding to distinguish between ventricular far field and atrial-based activation within one or more output intracardiac signals).

[0145] Figure 12 This is an exemplary flowchart of a method 1200 for decomposing near-field and far-field signals according to an embodiment. During the training phase, far-field ventricular measurements (1210) can be acquired. These can be unipolar signals. Measurements can be acquired using multi-electrode catheters and / or surface ECG. Numerous far-field measurements may exist. Far-field measurements can be pure far-field signals. In one embodiment, a pure far-field signal may originate from a recording where the bipolar signal is zero or nearly zero. Therefore, the near field in the unipolar signal may be absent or very small. In one embodiment, simulation can be used for pure far-field measurements using, for example, specialized simulation software capable of generating pure far-field signals. This can be accomplished by controlling the source generating the ECG signal and using only the far source. In one embodiment, an expert can determine the degree of pure far-field. In one embodiment, a surface ECG may primarily contain the far field. In one embodiment, measurements from areas of scar tissue can be used for pure far-field measurements that may not contain local activity, thus ignoring the near field.

[0146] Synthetic local field signals (1220) can be added. These signals can be, for example, analogous to ECG signals. A large number of unipolar signals can be introduced during the training phase, allowing the algorithm to learn to recognize unipolar signal patterns. Furthermore, the algorithm can be exposed to far-field signals and may be able to learn to detect these far-field signals.

[0147] This algorithm can be configured to evaluate or learn both pure far-field signals and combinations or mixtures (real or synthetic hybrid signals) of far-field and near-field signals. The algorithm can detect or predict the far-field component from the hybrid signal, which is the common part of all electrodes (far-field). The near-field is unique to each electrode because it has localized activity affecting only a small area of ​​tissue, while the far-field contributes much more (in both signal amplitude and dispersion within the tissue), hence it is called the common part because it is shared by a large number of electrodes. By subtracting the far-field component from the original signal, the common part of the electrodes can be the pure near-field signal.

[0148] Data can be routinely collected from multiple patients and fed to the system at the EP procedure (1240). The data may include regular monopolar signals, which are a combination of far-field and near-field signals. These signals can be collected in any VT procedure using a multi-electrode catheter. Another approach is to use a synthetic signal combining far-field and near-field components. For example, analog or synthetic data can be used to generate a pure far-field signal that can be used as a gold standard. These signals can be generated using specialized simulation software or any other simulation program that can control the ECG signal source. The data may include unique monopolar signals that include only far-field contributions and no near-field components. This monopolar signal can be obtained by placing the catheter in a position that does not contact the heart muscle. These signals may include ECG values ​​and 3D positions (for each electrode, such that the signal = ...) V(x, y, z, t) The data may include surface ECG signals, which are essentially far-field signals. The data may also include manually annotated unipolar signals that identify certain characteristics (specifically, LATs) of the underlying near-field signals. Surface ECG signal data and / or manually annotated data can be used to aid in training and / or validating any DL model.

[0149] Before processing the unipolar data to extract near-field contributions in the training phase (1230), a preprocessing filtering step can be performed to remove irrelevant signals and artifacts. Manual annotations can be provided for the unipolar signals. The preprocessing filtering can be evaluated by the user (semi-automatically), while in a later stage, the annotated data can be used to train a regular classification convolutional neural network (CNN) to automatically perform the filtering.

[0150] The neural network is trained to obtain a far-field signal estimate (1230). By understanding the far-field contribution, the near-field signal is the residual signal remaining after removing the far-field signal from the regular unipolar signal. The bipolar signal can then be reconstructed between two sets of unipolar signals.

[0151] In the far-field attenuation model (1250), since the far-field and near-field signals are known from the neural network training (1230), the autoencoder can automatically decompose the near-field and far-field signals from the measured signal (1240).

[0152] Neural networks can be implemented through a variety of exemplary methods, including autoencoders and conjoined networks, which can be applied to the decomposition or separation of far-field and near-field contributions (1250).

[0153] Autoencoders (AEs) are a class of unsupervised DNNs that learn a reduced-dimensional representation of a given dataset, enabling them to produce new data that are statistically similar to the original dataset.

[0154] In the context of the current method, if an appropriately chosen AE is trained using an EGM signal that contains only far-field contributions and no near-field contributions, then that AE can be used to extract far-field contributions from any arbitrary EGM signal. The training phase aims to make the input X and output X' (pure far-field) equal, while the prediction phase maps any arbitrary EGM signal to the far-field signal that includes it. Several AE types exist that include, for example, exploitable elements (e.g., various AEs, reconstruction AEs, denoising AEs, adversarial AEs).

[0155] A conjoined network comprising two identical parts is trained in a fully supervised manner to distinguish between similar and dissimilar feature pairs. Then, when presented with a reference feature and a new feature, the network predicts whether the new feature is plausible (i.e., whether it is similar to the reference). In recent years, the concept of conjoined networks has been extended to DNNs and successfully applied to face recognition and facial verification. More recently, conjoined neural networks have been applied to unsupervised learning for visual representation and medical diagnosis. Conjoined networks can be utilized because two unipolar signals from very close electrodes (which form a bipolar pair) typically have very similar far-field components. Therefore, if two unipolar signals are fed into each part of the conjoined network, a value function can be constructed that tends to make the outputs of the two parts equal. To avoid obtaining trivial solutions (such as identical zero signals), constraint terms can be added to the value function. These are, for example, terms that tend to minimize the difference between the results and the average of the two input signals.

[0156] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible specific embodiments of systems, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of instructions comprising one or more executable instructions for implementing a specified logical function. In some alternative embodiments, the functions indicated in the blocks may not occur in the order shown in the drawings. For example, depending on the function involved, two consecutively shown blocks may actually be executed substantially simultaneously, or the blocks may sometimes be executed in reverse order. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, may be implemented by a system based on dedicated hardware that performs the specified function or action, or by a combination of dedicated hardware and computer instructions.

[0157] While features and elements have been specifically described above, those skilled in the art will recognize that each feature or element can be used alone or in any combination with other features and elements. Furthermore, the methods described herein can be implemented in a computer program, software, or firmware incorporated in a computer-readable medium for execution by a computer or processor. As used herein, a computer-readable medium should not be construed as a transient signal itself, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through waveguides or other transmission media (e.g., light pulses passing through fiber optic cables), or electrical signals transmitted through wires.

[0158] Examples of computer-readable media include electronic signals (transmitted via wired or wireless connections) and computer-readable storage media. Examples of computer-readable storage media include, but are not limited to, registers, cache memory, semiconductor memory devices, magnetic media such as internal hard disks and removable disks, magneto-optical media, optical media such as optical discs (CDs) and digital versatile discs (DVDs), random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), static random access memory (SRAM), and memory sticks. A processor associated with software can be used to implement a radio frequency transceiver used in a WTRU, UE, terminal, base station, RNC, or any host computer.

[0159] It should be understood that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting. As used herein, unless the context clearly indicates otherwise, the singular forms “an,” “a,” and “the / described” include plural references. It should also be understood that the terms “comprising” and / or “including”, when used in this specification, specify the presence of the stated features, integers, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof.

[0160] The descriptions of various embodiments herein are presented for illustrative purposes and are not intended to be exhaustive or limited to the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments. The terminology used herein is chosen to best explain the principles of the embodiments, their practical application or improvement relative to commercially available technologies, or to enable others skilled in the art to understand the disclosed embodiments.

Claims

1. A method for far-field estimation of the ventricle, comprising: Receive one or more input intracardiac signals from a monitoring and processing device, wherein each of the one or more input intracardiac signals includes one or more signal artifacts; The one or more input intracardiac signals are encoded by an autoencoder using an intracardiac dataset to produce a potential representation; The autoencoder decodes the latent representation to generate one or more output intracardiac signals, the one or more output intracardiac signals comprising the one or more input intracardiac signals reconstructed without the artifacts of the one or more signals. The autoencoder includes a model that distinguishes between the far-field ventricular and atrial activation-based signals within the one or more output intracardiac signals during decoding. The method further includes: inputting far-field ventricular measurements, wherein a synthesized local field signal is added to a training algorithm; Furthermore, the training algorithm is used to separate the ventricular far field and the atrial-based activation.

2. The method of claim 1, wherein, Decoding the latent representation to generate one or more output intracardiac signals includes reconstructing the one or more output intracardiac signals based on the latent representation, including a reduced-dimensional representation.

3. The method of claim 1, wherein, The one or more input intracardiac signals are recorded by the patient biometric sensors of the monitoring and processing device.

4. The method of claim 1, wherein, The intracardiac dataset includes predetermined and approved signals free from the one or more signal artifacts.

5. The method of claim 1, wherein, The method further includes generating an electrocardiogram (ECG) based on the one or more output intracardiac signals, the ECG being free of the one or more signal artifacts.

6. A system for ventricular far-field estimation, comprising: A memory that stores processor-executable instructions for the automatic encoder; and A processor configured to execute processor-executable instructions of the autoencoder, such that the system: Receive one or more input intracardiac signals from a monitoring and processing device, wherein each of the one or more input intracardiac signals includes one or more signal artifacts; The one or more input intracardiac signals are encoded using an intracardiac dataset to generate a potential representation; The latent representation is decoded to generate one or more output intracardiac signals, the one or more output intracardiac signals comprising the one or more input intracardiac signals reconstructed without the artifacts of the one or more signals. The autoencoder includes a model that distinguishes between the far-field ventricular and atrial activation-based signals within the one or more output intracardiac signals during decoding. The processor is also configured to execute the processor-executable instructions of the autoencoder such that the system: inputs far-field ventricular measurements, wherein a synthesized local field signal is added to a training algorithm; and separates the ventricular far field and the atrial-based activation through the training algorithm.

7. The system of claim 6, wherein, Decoding the latent representation to generate one or more output intracardiac signals includes reconstructing the one or more output intracardiac signals based on the latent representation, including a reduced-dimensional representation.

8. The system of claim 6, wherein, The one or more input intracardiac signals include biometric data.

9. The system of claim 6, wherein, The one or more input intracardiac signals are recorded by the patient biometric sensors of the monitoring and processing device.

10. The system of claim 6, wherein, The intracardiac dataset includes predetermined and approved signals free from the one or more signal artifacts.

11. The system of claim 6, wherein, The processor is further configured to execute the processor-executable instructions of the autoencoder such that the system generates an electrocardiogram (ECG) based on the one or more output intracardiac signals, the ECG being free of the one or more signal artifacts.

12. The system of claim 6, wherein, The autoencoder includes a noise-reducing autoencoder.

Images