Systems and methods for computing and detecting cardiac mapping annotations for detecting stable arrhythmic heartbeats - Patents.com

Abstract

To provide a system and method for detecting a mapping annotation for an electrophysiological (EP) mapping system.SOLUTION: The system includes a processor including a machine learning algorithm configured to: receive a first heartbeat at an identified cardiac spatial location including a first set of attribute information pieces corresponding to the first heartbeat; receive a second heartbeat at the identified cardiac spatial location including a second set of attribute information pieces corresponding to the second heartbeat; compare the first set of attribute information pieces with the second set of attribute information pieces; and determine which of the first heartbeat and the second heartbeat has optimal characteristics based on the compared attribute information pieces.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] (CROSS-REFERENCE TO RELATED APPLICATIONS) This application claims the benefit of U.S. Provisional Application No. 63 / 037,259, filed June 10, 2020, the contents of which are incorporated by reference as if fully set forth.

[0002] FIELD OF THE INVENTION The present disclosure relates to artificial intelligence and machine learning associated with selecting optimal heart beats at each spatial location in stable cardiac arrhythmias, and calculating and detecting optimal cardiac mapping annotations. [Background technology]

[0003] Treatment for cardiac conditions, such as cardiac arrhythmias, often requires obtaining detailed mapping of cardiac tissue, chambers, veins, arteries, and / or electrical pathways. For example, as a prerequisite for performing catheter ablation, the source of the cardiac arrhythmia must be precisely located within a cardiac chamber. Such localization can be achieved via electrophysiological studies, in which spatially resolved electrical potentials are detected by a mapping catheter introduced into the cardiac chamber. This electrophysiological study, often referred to as electrophysiological (EP) cardiac mapping or electroanatomical (EA) cardiac mapping, provides 3D mapping data that can be displayed on a monitor. Often, mapping and processing functions (e.g., ablation) are provided by a single catheter or a group of catheters, such that the mapping catheter simultaneously operates as a processing (e.g., ablation) catheter.

[0004] Mapping cardiac regions, such as the heart, cardiac tissue, veins, arteries, and / or electrical pathways, can result in the identification of problem areas, such as scar tissue, arrhythmia sources (e.g., electrical rotors), healthy regions, etc. As further disclosed herein, cardiac regions can be mapped using a display to provide a visual rendering of the mapped cardiac region. 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 catheters inserted into the patient's body and provided for rendering simultaneously or at different times based on corresponding settings and / or preferences.

[0005] Electrocardiograms (ECGs) and electrograms (EGMs) are examples of cardiac mapping. ECGs are generated by electrical signals from the heart and describe cardiac activity. ECGs are utilized during cardiac procedures to identify potential onset locations of cardiac conditions. ECG signals can also be used to map portions of the heart. EGMs can be recorded from each of the electrodes in contact with the cardiac surface relative to temporal references, such as the occurrence of P waves in sinus rhythm from a surface ECG. ECG and EGM signals can be utilized in conjunction with rule-based algorithms to determine cardiac mapping annotations, as described, for example, in U.S. Patent Publication No. 2018 / 0042504. However, physicians may need to manually correct erroneous mapping annotations during cardiac procedures. Summary of the Invention [Problem to be solved by the invention]

[0006] EP cardiac mapping systems traditionally acquire and record the first heart beat for each spatial location, even if that heart beat is poorly characterized. There is a need for improved methods and systems for selecting and detecting the best heart beat at each spatial location as a mapping annotation. In addition, there is a need for improved methods and systems for determining cardiac mapping annotations. [Means for solving the problem]

[0007] Described herein are methods, devices, systems, and models for selecting the optimal heart beat at each spatial location in stable cardiac arrhythmias, and for calculating and detecting optimal cardiac mapping annotations.

[0008] According to one aspect, the subject matter disclosed herein relates to a system for detecting heartbeats with optimal characteristics in an electrophysiological (EP) mapping system, including a processor including a machine learning algorithm configured to receive a first heartbeat at an identified heart space location including a first set of attribute information corresponding to the first heartbeat, receive a second heartbeat at the identified heart space location including a second set of attribute information corresponding to the second heartbeat, compare the first set of attribute information with the second set of attribute information, and determine which of the first heartbeat and the second heartbeat has optimal characteristics based on the compared attribute information.

[0009] According to another aspect, the subject matter disclosed herein relates to a method for detecting a heartbeat having optimal characteristics in an EP mapping system by a machine learning algorithm, the method including: receiving first data including a first heartbeat at an identified heart space location, the first data including first attribute information corresponding to the first heartbeat; receiving second data including the second heartbeat at the identified heart space location, the second data including second attribute information corresponding to the second heartbeat; comparing the first data with the second data; and outputting a binary decision of which of the first heartbeat and the second heartbeat has optimal characteristics based on the comparison.

[0010] According to yet another aspect, the subject matter disclosed herein relates to a system for detecting mapping annotations in an EP mapping system, including a processor including a machine learning algorithm configured to receive input data including attribute data for each of multiple heartbeats acquired at the same spatial location, and determine which heartbeats should be used as mapping annotations based on the heartbeats with the best attribute data.

[0011] According to yet another aspect, the subject matter disclosed herein relates to a method for detecting mapping annotations in an EP mapping system by a machine learning algorithm, the method including receiving input data including attribute data for each of a plurality of heartbeats acquired at the same spatial location, comparing the attribute data for each of the plurality of heartbeats to a predefined threshold, determining which heartbeat should be used as a mapping annotation based on the heartbeat with the best attribute data, and outputting the determination to the EP mapping system.

[0012] According to yet another aspect, the subject matter disclosed herein relates to a system for determining a mapping annotation for an EP mapping system, including a processor including a machine learning algorithm configured to receive first EP signal data obtained from a cardiac location, determine an initial mapping annotation based on the first EP signal data, receive data related to a manually revised mapping annotation, receive second EP signal data obtained from the cardiac location, and determine a new mapping annotation based on the second EP signal data and the manually revised mapping annotation.

[0013] According to yet another aspect, the subject matter disclosed herein relates to a method for determining a mapping annotation of an EP mapping system by a machine learning algorithm, the method including receiving first EP signal data obtained from a cardiac location, determining an initial mapping annotation based on the first EP signal data, receiving data related to a manually corrected mapping annotation at the cardiac location, receiving second EP signal data obtained from the cardiac location, and determining a new mapping annotation based on the second EP signal data and the manually corrected mapping annotation. [Brief explanation of the drawings]

[0014] The above and other features and advantages of the present invention will be apparent from the following more particular description of preferred embodiments of the invention, as illustrated in the accompanying drawings. [Figure 1] 1 is a block diagram of an exemplary system for remotely monitoring and communicating patient biometrics in accordance with the subject matter of the present application. [Figure 2] 1 is a system diagram of an exemplary computing environment in communication with a network in accordance with the subject matter of the present application. [Figure 3] 1 is a block diagram of an example device capable of implementing one or more features of the present disclosure in accordance with the subject matter of the present application. [Figure 4] 4 shows a graphical depiction of an artificial intelligence system incorporating the exemplary device of FIG. 3 in accordance with the subject matter of the present application. [Figure 5] 5 illustrates a method implemented in the artificial intelligence system of FIG. 4 according to the subject matter of the present application. [Figure 6] 1 illustrates an example of probability in naive Bayes calculations according to the subject matter of the present application. [Figure 7] 1 illustrates an exemplary decision tree in accordance with the subject matter of the present application. [Figure 8] 1 illustrates an exemplary random forest classifier in accordance with the subject matter of the present application. [Figure 9] 1 illustrates an exemplary logistic regression according to the subject matter of the present application. [Figure 10] 1 illustrates an exemplary support vector machine (SVM) in accordance with the subject matter of the present application. [Figure 11] 1 illustrates an exemplary linear regression model in accordance with the subject matter of the present application. [Figure 12] 1 illustrates an exemplary K-means clustering according to the subject matter of the present application. [Figure 13] 1 illustrates an exemplary ensemble learning algorithm in accordance with the subject matter of the present application. [Figure 14] 1 illustrates an exemplary neural network in accordance with the subject matter of the present application. [Figure 15] 1 illustrates a hardware-based neural network in accordance with the subject matter of the present application. [Figure 16] 1 illustrates electrocardiogram (ECG) signals produced by the contraction (depolarization) and relaxation (repolarization) of the atrial and ventricular muscles of the heart, in accordance with the subject matter of the present application. [Figure 17] 1 illustrates an exemplary cardiac ablation system in accordance with the subject matter of the present application that may implement one or more features of the disclosed subject matter. [Figure 18] 1 illustrates an exemplary embodiment of a convolutional neural network that can learn to receive attributes of two heartbeats and identify which heartbeat is better, according to the present application. [Figure 19]FIG. 1 is an exemplary flow diagram illustrating a method for training a neural network to learn to take attributes of two heartbeats and identify which heartbeat is better, in accordance with the subject matter of the present application. [Figure 20] 1 is an exemplary illustration of multiple electrodes sampling multiple contact points of a heart cavity, according to an exemplary embodiment of the present application. [Figure 21] 1 illustrates an exemplary embodiment of a recurrent neural network that can receive attributes of multiple heartbeats at the same spatial location and learn to identify the best mapping annotation, in accordance with the subject matter of the present application. [Figure 22] 10A-10C are exemplary diagrams illustrating the modification of mapping annotations based on the highest peak voltage of an ECG signal surrounding a reference annotation, in accordance with the subject matter of the present application. [Figure 23] FIG. 1 is an exemplary flow diagram illustrating a system and method for training a machine learning algorithm to determine mapping annotations in accordance with the subject matter of the present application. [Figure 24] 1A-1C are exemplary diagrams of EGM signals showing mapping annotations fixed by a rule-based system and EGM signals showing mapping annotations manually corrected by a physician. DETAILED DESCRIPTION OF THE INVENTION

[0015] Systems and methods are provided for selecting and detecting the best heart beat at each spatial location as a mapping annotation, and for improving the determination of cardiac mapping annotations based on machine learning.

[0016] 1 is a block diagram of an exemplary system 100 for remotely monitoring and communicating patient biometrics (i.e., patient data). In the example shown in FIG. 1, the system 100 includes a patient biometric monitoring and processing unit 102 associated with a patient 104, a local computing device 106, a remote computing system 108, a first network 110, and a second network 120.

[0017] According to an exemplary embodiment, the monitoring and processing device 102 may be a device internal to the patient's body (e.g., subcutaneously implantable). The monitoring and processing device 102 may be inserted into the patient via any applicable method, including oral injection, surgical insertion via a vein or artery, endoscopic procedure, or laparoscopic procedure.

[0018] According to an exemplary embodiment, the monitoring and processing device 102 may be a device external to the patient. For example, as described in more detail below, the monitoring and processing device 102 may include an attachable patch (e.g., attached to the patient's skin). The monitoring and processing device 102 may also include a catheter with one or more electrodes, a probe, a blood pressure cuff, a weight scale, a bracelet or smartwatch biometric tracker, a glucose monitor, a continuous positive airway pressure (CPAP) machine, or virtually any device that can provide input regarding the patient's health or biometrics.

[0019] According to an exemplary embodiment, the monitoring and processing unit 102 may include both components internal to the patient and components external to the patient.

[0020] A single monitoring and processing device 102 is shown in FIG. 1 . However, an exemplary system may include multiple patient biometric monitoring and processing devices. A patient biometric monitoring and processing device may communicate with one or more other patient biometric monitoring and processing devices. Additionally or alternatively, a patient biometric monitoring and processing device may communicate with a first network 110.

[0021] One or more monitoring and processing devices 102 may acquire patient biometric data (e.g., electrical signals, blood pressure, body temperature, blood glucose levels, or other biometric data) and may receive at least a portion of the patient biometric data representing the acquired patient biometric indicators, as well as additional information associated with the acquired patient biometric indicators from one or more other monitoring and processing devices 102. The additional information may be, for example, diagnostic information and / or additional information obtained from additional devices, such as wearable devices. Each monitoring and processing device 102 may process data including its own acquired patient biometric indicators as well as data received from one or more other monitoring and processing devices 102.

[0022] 1, the first network 110 is an example of a short-range network (e.g., a local area network (LAN) or a personal area network (PAN)). Information may be transmitted over the short-range network 110 between the monitoring and processing equipment 102 and the local computing device 106 using any one of a variety of short-range wireless communication protocols, such as Bluetooth, Wi-Fi, Zigbee, Z-Wave, near field communication (NFC), Ultraband, Zigbee, or infrared (IR).

[0023] In an exemplary embodiment, second network 120 may be a wired network, a wireless network, or may include one or more wired and wireless networks. For example, second network 120 may be a long-range network (e.g., a wide area network (WAN), the Internet, or a cellular network). Information may be transmitted over second network 120 using any one of a variety of long-range wireless communication protocols (e.g., TCP / IP, HTTP, 3G, 4G / LTE, or 5G / New Radio).

[0024] In one exemplary embodiment, the patient monitoring and processing device 102 may include patient biometric sensors 112, a processor 114, user input (UI) sensors 116, memory 118, and a transmitter-receiver (i.e., transceiver) 122. The monitoring and processing device 102 may continuously or periodically monitor, store, process, and communicate any number of various patient biometric indicators over the first network 110. Examples of patient biometric indicators include electrical signals (e.g., electrocardiogram (ECG) signals and brain biometrics), blood pressure data, blood glucose data, and temperature data. The patient biometric indicators may be monitored and communicated for treatment across any number of various diseases, such as cardiovascular diseases (e.g., arrhythmias, cardiomyopathies, and coronary artery disease) and autoimmune diseases (e.g., type 1 and type 2 diabetes).

[0025] In one embodiment, the patient biometric sensor 112 may include, for example, one or more sensors configured to sense a type of biometric patient biometric indicator. For example, the patient biometric sensor 112 may include electrodes configured to acquire electrical signals (e.g., cardiac signals, brain signals, or other bioelectrical signals), a temperature sensor, a blood pressure sensor, a blood glucose sensor, a blood oxygen sensor, a pH sensor, an accelerometer, and a microphone.

[0026] In one exemplary embodiment, as described in more detail below, the patient biometric monitoring and processing device 102 may be an ECG monitor for monitoring cardiac ECG signals. The patient biometric sensor 112 of the ECG monitor may include one or more electrodes for acquiring the ECG signals. The ECG signals may be used in the treatment of various cardiovascular diseases.

[0027] In one exemplary embodiment, the transceiver 122 may include a separate transmitter and receiver. Alternatively, the transceiver 122 may include a transmitter and receiver integrated into a single device.

[0028] In one exemplary embodiment, the processor 114 may be configured to store patient data, such as patient biometric data acquired by the patient biometric sensors 112, in the memory 118 and to communicate the patient data over the first network 110 via the transmitter of the transceiver 122. Data from one or more other monitoring and processing devices 102 may also be received by the receiver of the transceiver 122, as described in more detail below.

[0029] According to one exemplary embodiment, the monitoring and processing device 102 includes a UI sensor 116, which may be a piezoelectric or capacitive sensor configured to receive user input, such as a tap or touch. For example, the UI sensor 116 may be controlled to implement a capacitive coupling in response to the patient 104 tapping or touching the surface of the monitoring and processing device 102. Gesture recognition may be implemented via any one of a variety of capacitive types, such as resistive capacitive, surface capacitive, projected capacitive, surface acoustic wave, piezoelectric, and infrared touch. The capacitive sensor may be positioned over a small area or length of the surface such that a tap or touch on the surface activates the monitoring device.

[0030] As described in more detail below, processor 114 may be configured to selectively respond to different tapping patterns of a capacitive sensor (e.g., single tap or double tap), which may be UI sensor 116, so that different tasks of the patch (e.g., data acquisition, storage, or transmission) may be initiated based on the detected pattern. In some embodiments, audible feedback may be provided to the user from processing unit 102 when a gesture is detected.

[0031] In an exemplary embodiment, the local computing device 106 of the system 100 may be configured to communicate with the patient biometric monitoring and processing device 102 and act as a gateway to the remote computing system 108 via the second network 120. The local computing device 106 may be, for example, a smartphone, smartwatch, tablet, or other portable smart device configured to communicate with other devices via the second network 120. Alternatively, the local computing device 106 may be a fixed or standalone device, such as, for example, a fixed base station including modem and / or router capabilities, a desktop or laptop computer using an executable program to communicate information between the processing device 102 and the remote computing system 108 via a wireless module in a PC, or a Universal Serial Bus (USB) dongle. Patient biometric indicators may be communicated between the local computing device 106 and the patient biometric monitoring and processing device 102 via a short-range wireless network 110, such as a LAN (e.g., PAN), using short-range wireless technology standards (e.g., Bluetooth, Wi-Fi, ZigBee, Z-Wave, and other short-range wireless standards). In some embodiments, the local computing device 106 may also be configured to display the acquired patient electrical signals and information related to the acquired patient electrical signals, as described in more detail below.

[0032] In some demonstrative embodiments, the remote computing system 108 may be configured to receive at least one of the monitored patient's biometric indicators and information associated with the monitored patient via a second network 120, which is a long-range network. For example, if the local computing device 106 is a cellular phone, the second network 120 may be a wireless cellular network, and information may be communicated between the local computing device 106 and the remote computing system 108 via a wireless technology standard, such as any of the wireless technologies described above. As described in more detail below, the remote computing system 108 may be configured to provide (e.g., visually display and / or audibly provide) at least one of the patient's biometric indicators and associated information to a medical professional (e.g., a physician).

[0033] 2 is a system diagram of an example computing environment 200 in communication with a second network 120. In some examples, computing environment 200 is incorporated into a public cloud computing platform (such as Amazon Web Services or Microsoft Azure), a hybrid cloud computing platform (such as HP Enterprise OneSphere), or a private cloud computing platform.

[0034] As shown in FIG. 2, computing environment 200 preferably includes a remote computing system 108, which is one example of a computing system on which embodiments described herein may be implemented.

[0035] The remote computing system 108 may perform various functions via the processor 220, which may include one or more processors. For example, functions may include analyzing monitored patient biometrics and related information and providing alerts, additional information, or instructions (e.g., via the display 266) according to physician-determined or algorithm-driven thresholds and parameters. As described in more detail below, the remote computing system 108 may be used to provide such a patient information dashboard (e.g., via the display 266) to a medical professional (e.g., a physician) so that the patient information may enable the medical professional to identify and prioritize patients with greater needs than others.

[0036] 2, the remote computing system 108 may include a communication mechanism, such as a bus 221, or other communication mechanism for communicating information within the remote computing system 108. The remote computing system 108 further includes one or more processors 220 coupled with the bus 221 for processing information. The processor 220 may include one or more central processing units (CPUs), graphics processing units (GPUs), or any other processor known in the art.

[0037] The remote computing system 108 may also include a system memory 230 coupled to the bus 221 for storing information and instructions executed by the processor 220. The system memory 230 may include computer-readable storage media in the form of volatile and / or nonvolatile memory, such as read-only system memory (ROM) 231 and / or random access memory (RAM) 232. The system memory RAM 232 may include other dynamic storage device(s) (e.g., dynamic RAM, static RAM, and synchronous DRAM). The system memory ROM 231 may include other static storage device(s) (e.g., programmable ROM, erasable PROM, and electrically erasable PROM). Additionally, the system memory 230 may be used to store temporary variables or other intermediate information during execution of instructions by the processor 220. A basic input / output system 233 (BIOS) may include routines for transferring information, which may be stored in the system memory ROM 231, between elements within the remote computing system 108, such as during start-up. RAM 232 may contain data and / or program modules that are immediately accessible to and / or presently being operated on by processor 220. System memory 230 may also include, for example, an operating system 234, application programs 235, other program modules 236, and program data 237.

[0038] In an exemplary embodiment, remote computing system 108 also includes a disk controller 240 coupled to bus 221 for controlling one or more storage devices for storing information and instructions, such as a magnetic hard disk 241 and a removable media drive 242 (e.g., a floppy disk drive, a compact disk drive, a tape drive, and / or a solid state drive). Storage devices may be added to remote computing system 108 using an appropriate device interface (e.g., small computer system interface (SCSI), integrated device electronics (IDE), USB, or FireWire).

[0039] The remote computing system 108 may also include a display controller 265 coupled to the bus 221 for controlling a monitor or display 266, such as a cathode ray tube (CRT) or liquid crystal display (LCD), for displaying information to a computer user. The illustrated remote computing system 108 includes a user input interface 260 and one or more input devices, such as a keyboard 262 and a pointing device 261, for interacting with a computer user and providing information to the processor 220. The pointing device 261 may be, for example, a mouse, trackball, or pointing stick for communicating directional information and command selections to the processor 220 and for controlling cursor movement on the display 266. The display 266 may provide a touchscreen interface, which may allow for the communication of directional information and command selections by the pointing device 261 and / or keyboard 262 to be supplemented or replaced.

[0040] The remote computing system 108 may perform some or each of the functions and methods described herein in response to the processor 220 executing one or more sequences of one or more instructions contained in a memory, such as the system memory 230. Such instructions may be read into the system memory 230 from another computer-readable medium, such as a hard disk 241 or a removable media drive 242. The hard disk 241 may include one or more data stores and data files used by the embodiments described herein. Data store contents and data files may be encrypted for improved security. The processor 220 may also be employed in a multi-processing configuration to execute one or more sequences of instructions contained in the system memory 230. In alternative embodiments, hardwired circuitry may be used in place of or in combination with software instructions. Thus, the embodiments are not limited to any specific combination of hardware circuitry and software.

[0041] As mentioned above, the remote computing system 108 may include at least one computer-readable medium or memory for retaining programmed instructions according to the embodiments described herein and for containing the data structures, tables, records, or other data described herein. As used herein, the term “computer-readable medium” refers to any non-transitory, tangible medium that participates in providing instructions to the processor 220 for execution. Computer-readable media can take many forms, including, but not limited to, non-volatile media, volatile media, and transmission media. Non-limiting examples of non-volatile media include optical disks, solid-state drives, magnetic disks, and magneto-optical disks, such as the hard disk 241 or the removable media drive 242. Non-limiting examples of volatile media include dynamic memory, such as the system memory 230. Non-limiting examples of transmission media include coaxial cables, copper wire, and fiber optics, including the wires that comprise the bus 221. Transmission media can also take the form of acoustic or light waves, such as those generated during radio wave and IR data communications.

[0042] The computing environment 200 may further include a remote computing system 108 operating in a networked environment using logical connections to the local computing device 106 and to one or more other devices, such as a personal computer (laptop or desktop), mobile device (e.g., a patient mobile device), server, router, network PC, peer device, or other common network node, and typically includes many or all of the elements described above for the remote computing system 108. When used in a networked environment, the remote computing system 108 may include a modem 272 for establishing communications over a second network 120, such as the Internet. The modem 272 may be connected to the system bus 221 via a network interface 270 or another appropriate mechanism.

[0043] The second network 120 as shown in FIGS. 1 and 2 may be any network or system commonly known in the art, including the Internet, an intranet, a LAN, a WAN, a metropolitan area network (MAN), a direct connection or series of connections, a cellular telephone network, or any other network or medium capable of facilitating communication between a computer system and another computer (e.g., local computing device 106).

[0044] 3 is a block diagram of an example device 300 capable of implementing one or more features of the present disclosure. The device 300 may be, for example, a local computing device 106. The device 300 may include, for example, a computer, a gaming device, a handheld device, a set-top box, a television, a mobile phone, or a tablet computer. The device 300 includes a processor 302, a memory 304, a storage device 306, one or more input devices 308, and one or more output devices 310. The device 300 may also optionally include an input driver 312 and an output driver 314. It is understood that the device 300 may include additional components not shown in FIG. 3 , including an artificial intelligence accelerator.

[0045] In various variations, processor 302 includes a CPU, a GPU, a CPU and a GPU located on the same die, or one or more processor cores, each of which may be a CPU or a GPU. In various variations, memory 304 is located on the same die as processor 302 or is located separately from processor 302. Memory 304 may include volatile or non-volatile memory, such as RAM, dynamic RAM, or cache.

[0046] The storage devices 306 include fixed or removable storage means, such as a hard disk drive, solid state drive, optical disk, or flash drive. The input devices 308 include, but are not limited to, a keyboard, keypad, touch screen, touchpad, detector, microphone, accelerometer, gyroscope, biometric scanner, or network connection (e.g., a wireless LAN card for transmitting and / or receiving wireless IEEE 802 signals). The output devices 310 include, but are not limited to, a display, speakers, printer, haptic feedback device, one or more lights, antenna, or network connection (e.g., a wireless LAN card for transmitting and / or receiving wireless IEEE 802 signals).

[0047] The input driver 312 communicates with the processor 302 and the input device 308, allowing the processor 302 to receive input from the input device 308. The output driver 314 communicates with the processor 302 and the output device 310, allowing the processor 302 to send output to the output device 310. Note that the input driver 312 and the output driver 314 are optional components, and that the device 300 would operate in the same manner if the input driver 312 and the output driver 314 were not present. The output driver 314 may include an accelerated processing device (APD) 316 coupled to a display device 318. The APD accepts computational commands and graphics rendering commands from the processor 302, processes these computational commands and graphics rendering commands, and provides pixel output to the display device 318 for display. As described in further detail below, the APD 316 includes one or more parallel processing units to perform computations according to the single instruction, multiple data (SIMD) paradigm. Thus, although various functions are described herein as being performed by or in conjunction with APD 316, in various variations, the functions described as being performed by APD 316 are additionally or alternatively performed by other computing devices that are not driven by a host processor (e.g., processor 302) and have similar capabilities for providing graphical output to display device 318. For example, it is contemplated that any computing system that performs processing tasks according to the SIMD paradigm may perform the functions described herein. Alternatively, it is contemplated that computing systems that do not perform processing tasks according to SIMD methods perform the functions described herein.

[0048] FIG. 4 shows a functional graphical depiction of an artificial intelligence system 400 incorporating the exemplary device of FIG. 3 . The system 400 includes data 410, a machine 420, a model 430, a plurality of predicted outcomes 440, and underlying hardware 450. The system 400 operates by training the machine 420 using the data 410 while building a model 430 that enables the prediction of a plurality of outcomes 440. The system 400 may operate on the hardware 450. In such a configuration, the data 410 may be associated with the hardware 450 and may originate, for example, from the monitoring and processing unit 102. For example, the data 410 may be ongoing data or output data associated with the hardware 450. The machine 420 may operate as a controller or data collection associated with the hardware 450 or may be associated with it. The model 430 may be configured to model the operation of the hardware 450 as well as model the data 410 collected from the hardware 450 to predict the results achieved by the hardware 450. The hardware 450 may be configured to use the predicted results 440 to provide a particular desired result 440 from the hardware 450 .

[0049] Figure 5 illustrates a general method 500 implemented in the artificial intelligence system of Figure 4. Method 500 includes, at 510, collecting data from hardware. This data may include historical data or other data currently collected from the hardware or various combinations thereof. For example, this data may include measurements taken during a surgical procedure and can be correlated with the outcome of the procedure. For example, cardiac temperature can be collected and correlated with the outcome of a cardiac procedure.

[0050] At 520, method 500 includes training a machine on the hardware. Training may include analyzing and correlating the data collected at 510. For example, in the case of the heart, temperature and outcome data may be trained to determine whether a correlation or association exists between cardiac temperature during treatment and outcome.

[0051] At 530, method 500 includes building a model related to the hardware and associated data. Building the model may include physical hardware or software modeling, algorithmic modeling, etc., as described below. This modeling may aim to represent the collected and trained data.

[0052] At 540, method 500 includes predicting an outcome of a model associated with the hardware. This outcome prediction may be based on the trained model. For example, for the heart, a temperature between 97.7°F and 100.2°F during the procedure may result in a positive outcome from the procedure, and for a given procedure, the outcome may be predicted based on the temperature of the heart during the procedure. This model, while rudimentary, is provided for illustrative purposes to facilitate understanding of the present invention.

[0053] The present systems and methods operate to train machines, build models, and predict outcomes using algorithms. These algorithms may be used to solve the trained models and predict outcomes related to hardware. These algorithms can generally be categorized into classification algorithms, regression algorithms, and clustering algorithms.

[0054] For example, classification algorithms are used in situations where the dependent variable, which is the variable to be predicted, is divided into multiple classes, and one class, i.e., the dependent variable, is predicted for a given input. Thus, classification algorithms are used to predict an outcome from a predetermined number of fixed predefined outcomes. Classification algorithms may include naive Bayes algorithms, decision trees, random forest classifiers, logistic regression, support vector machines (SVMs), and K-nearest neighbors (KNNs).

[0055] In general, the Naive Bayes algorithm follows Bayes' Theorem and follows a probabilistic approach, it being understood that other probability-based algorithms may be used and generally operate using similar principles of probability theory as those described below for the exemplary Naive Bayes algorithm.

[0056] Figure 6 shows an example of a naive Bayes calculation of probability. The Bayes Theorem probability approach essentially means that instead of jumping directly to the data, the algorithm has a set of prior probabilities for each target class. After the data is input, the naive Bayes algorithm may update the prior probabilities to form posterior probabilities. This is done using the following formula:

[0057]

number

[0058] The Naive Bayes algorithm and the Bayes algorithm in general can be useful when one needs to predict whether an input belongs to a given list of n classes or not. Because the probability of all n classes is very low, a probabilistic approach can be used.

[0059] For example, as shown in FIG. 6, a person's decision to play golf depends on factors including, but not limited to, the weather outside, as shown in a first data set 610. The first data set 610 lists the weather in a first column and the play outcomes associated with that weather in a second column. A frequency table 620 generates the frequency with which a particular event occurs. The frequency table 620 determines how often a person plays or does not play golf in each weather condition. From this, a likelihood table is compiled and initial probabilities are generated. For example, the probability that the weather is cloudy is 0.29, but the general probability of playing is 0.64.

[0060] Posterior probabilities may be generated from likelihood table 630. These posterior probabilities may be configured to answer questions about weather conditions and whether golf will be played in those weather conditions. For example, the probability that it is sunny outside and golf will be played is determined using the Bayesian formula: P(Yes|Sunny) = P(Sunny|Yes) × P(Yes) / P(Sunny) It may be represented by: According to likelihood table 630, P(sunny|yes) = 3 / 9 = 0.33, P(sunny)=5 / 14=0.36, P(yes) = 9 / 14 = 0.64 is. Therefore, P(yes|sunny) = .33 × .64 / .36 or approximately 0.60 (60%).

[0061] Generally, a decision tree is a tree structure similar to a flowchart, where each outer node represents a test of an attribute and each branch represents the result of that test. The leaf nodes contain the actual predicted label. A decision tree starts at the root of the tree and attribute values ​​are compared until a leaf node is reached. Decision trees can be used as classifiers when dealing with high-dimensional data or when little time is spent on data preparation. Decision trees can take the form of simple decision trees, linear decision trees, algebraic decision trees, deterministic decision trees, randomized decision trees, non-deterministic decision trees, and quantum decision trees. An exemplary decision tree is shown below in Figure 7.

[0062] 7 shows a decision tree 700 that follows the same structure as the Bayesian example above when deciding whether to play golf. In the decision tree, a first node 710 examines whether the weather is sunny 712, cloudy 714, and rainy 716 as options for progressing down the decision tree. If the weather is sunny, the tree branch continues to a second node 720 that examines the temperature. In this example, the temperature at node 720 can be high 722 or normal 724. If the temperature at node 720 is high 722, a predicted outcome of "No to Golf" 723 occurs. If the temperature at node 720 is normal 724, a predicted outcome of "Yes to Golf" 725 occurs.

[0063] Furthermore, from the first node 710, if the weather is cloudy 714, a predicted result of Golf "Yes" 715 occurs.

[0064] From the first node 710, as a result of rain 716, a third node 730 (again) checks the temperature. If the temperature is normal 732 at the third node 730, the answer is "yes" 733 to play golf. If the temperature is low 734 at the third node 730, the answer is "no" 735 to not play golf.

[0065] From this decision tree, a golfer will play golf when it is cloudy 715, when it is sunny with normal temperatures 725, and when it is raining with normal temperatures 733, but this golfer will not play when it is sunny with high temperatures 723 or raining with low temperatures 735.

[0066] A random forest classifier is a committee of decision trees, where each tree is given a subset of the data's attributes and makes a prediction based on that subset. The mode of the decision trees' actual predictions is taken into account to provide the final random forest answer. Random forest classifiers generally mitigate the overfitting that exists in standalone decision trees, making them more robust and accurate classifiers.

[0067] FIG. 8 illustrates an exemplary random forest classifier 800 for classifying clothing color. As shown in FIG. 8, the random forest classifier 800 includes five decision trees 810a, 810b, 810c, 810d, and 810e (collectively or generally referred to as decision trees 810). Each tree is designed to classify clothing color. Because the individual trees generally operate as the decision trees of FIG. 7, each tree and decision will not be described. In this figure, three of the five trees (810a, 810b, 810d) determine that the clothing is blue, one determines that the clothing is green (810c), and the remaining tree determines that the clothing is red (810e). The random forest receives these actual predictions of the five trees and calculates the mode of the actual predictions to provide the random forest's answer that the clothing is blue.

[0068] Logistic regression is another algorithm for binary classification tasks. It is based on the logistic function, also known as the sigmoid function. This S-shaped curve can take any real-valued value and map it between 0 and 1, asymptotically approaching these bounds. Logistic models can be used to model the probability of a particular class or event being present, such as pass / fail, win / lose, alive / dead, or healthy / sick. This may also be extended to model several classes of events, such as determining whether an image contains a cat, a dog, a lion, etc. Each object detected in an image is assigned a probability between 0 and 1, but these probabilities sum to 1.

[0069] In a logistic model, the log odds (log of the odds) of a value labeled "1" are a linear combination of one or more independent variables ("predictors"), each of which can be binary (two classes coded by indicator variables) or continuous (any real-valued value). The corresponding probability of a value labeled "1" is labeled as such because it can vary between 0 (definitely a value of "0") and 1 (definitely a value of "1"). The logistic function is so named because it converts log odds into probabilities. The unit of measurement for the log odds scale is called a logit, another name for the logistic unit. Similar models, such as probit models, can also be used, but with a sigmoid function instead of the logistic function. A distinctive feature of the logistic model is that increasing one of the independent variables multiplicatively scales the odds of a given outcome by a constant factor, and each independent variable has its own parameters. For binary dependent variables, this generalizes the odds ratio.

[0070] In a binary logistic regression model, the dependent variable has two levels (categories). Outputs with more than two values ​​are modeled by multinomial logistic regression, and if the categories are ordered, they are modeled by ordinal logistic regression (e.g., proportional odds ordinal logistic model). Although a logistic regression model itself simply models the probability of the output with respect to the inputs and does not perform statistical classification (it is not a classifier), it can be used to create a classifier, for example, by selecting a cutoff value and classifying inputs with probabilities greater than the cutoff as one class and inputs with probabilities less than the cutoff as the other class. This is a common way to create a binary classifier.

[0071] FIG. 9 illustrates an exemplary logistic regression 900. This exemplary logistic regression allows for the prediction of an outcome based on a set of variables. For example, based on an individual's grade point average, a school's acceptance outcome can be predicted. Predictions can be made based on the relationship between past history of grade point average and acceptance. Logistic regression 900 allows analysis of a grade point average variable 920 to predict an outcome 910 defined between 0 and 1. At the lower end of the S-curve 930, the grade point average 920 predicts an outcome 910 of not being accepted. At the upper end of the S-curve 940, the grade point average 920 predicts an outcome 910 of being accepted. Logistic regression can be used to predict home values, customer lifetime value in the insurance sector, and the like.

[0072] Using SVM, the data can be sorted to make the margin between the two classes as far apart as possible, which is called margin-maximizing separation. Unlike linear regression, which uses the entire dataset for its purpose, SVM is able to take support vectors into account while plotting the hyperplane.

[0073] FIG. 10 illustrates an exemplary SVM 1000. In the exemplary SVM 1000, data can be classified into two distinct classes, represented as squares 1010 and triangles 1020. The SVM 1000 operates by drawing a random hyperplane 1030. This hyperplane 1030 is monitored by comparing the distance (shown by lines 1040) between the hyperplane 1030 and the nearest data points 1050 from each class. The data points 1050 closest to the hyperplane 1030 are known as support vectors. The hyperplane 1030 is drawn based on these support vectors 1050, with the optimal hyperplane having the greatest distance from each support vector 1050. The distance between the hyperplane 1030 and the support vectors 1050 is known as the margin.

[0074] The SVM 1000 may be used for data classification by using the hyperplane 1030 to maximize the distance between the hyperplane 1030 and the support vectors 1050. Such an SVM 1000 may be used, for example, to predict heart disease.

[0075] KNN generally refers to a set of algorithms that make no assumptions about underlying data distributions and conduct reasonably short training phases. Generally, KNNs use a large number of data points divided into multiple classes to predict the classification of a new sample point. Operationally, KNNs are designed to predict the classification of a new sample point, using an integer N. The N entries in a model of the system that are closest to the new sample are selected. The most common classification of these entries is determined, and that classification is assigned to the new sample. KNNs generally require increasing storage space as the training set grows. This also means that estimation time increases linearly with the number of training points.

[0076] In regression algorithms, the output is continuous, so regression algorithms can be used when the target variable is a continuous variable. Linear regression is a common example of a regression algorithm. Linear regression can be used to measure true quality (such as housing costs, number of calls, or total transactions) by considering consistent variables. The relationship between the variables and the outcome is found by fitting a line of best fit (hence the name linear regression). This line of best fit is known as the regression line and is expressed directly in terms of Y=a×X+b. Linear regression is most often used in low-dimensional approaches.

[0077] 11 shows an exemplary linear regression model 1100. In this model, a predictor variable 1110 is modeled against a measurement variable 1120. Clusters of instances of the predictor variable 1110 and the measurement variable 1120 are plotted as data points 1130. The data points 1130 are then fitted to a best-fit line 1140. Subsequent predictions then use the best-fit line 1140 to predict the predictor variable 1110 given the measurement variable 1120. Linear regression can be used to model and predict financial portfolios, salary forecasts, real estate, and estimated arrival times for transportation.

[0078] Clustering algorithms may be used to model and train datasets. In clustering, inputs are assigned to two or more clusters based on feature similarity. Clustering algorithms generally learn patterns and useful insights from data without guidance. For example, unsupervised learning algorithms such as K-means clustering may be used to cluster viewers into similar groups based on interests, age, geography, etc.

[0079] K-means clustering is generally considered a simple unsupervised learning approach. In K-means clustering, similar data points can be grouped together and bound together in the form of clusters. One way to bind data points together is by calculating the centroid of the data points. When determining efficient clusters in K-means clustering, the distance between each point from the cluster centroid is evaluated. Depending on the distance between the data point and the centroid, the data is assigned to the closest cluster. The goal of clustering is to determine the unique groupings of a set of unlabeled data. The "K" in K-means represents the number of clusters formed. The number of clusters (essentially the number of classes into which new instances of data can be classified) can be determined by the user. This can be done, for example, using feedback and looking at the size of the clusters during training.

[0080] K-means is used when the data set has unique and well-separated points; otherwise, modeling may lead to inaccurate clusters if the clusters are not separated. Additionally, K-means may be avoided if the data set contains many outliers or is non-linear.

[0081] FIG. 12 illustrates K-means clustering. In K-means clustering, data points are plotted and assigned a K value. For example, for K=2 in FIG. 12, the data points are plotted as shown in depiction 1210. Next, at 1220, the points are assigned to similar centers. Cluster centroids are identified as shown in 1230. Once the centroids are identified, the points are reassigned to clusters such that the distance between the data points and the centroid of each cluster is minimized, as shown in 1240. New centroids of the clusters can then be determined as shown in depiction 1250. Once the data points are reassigned to clusters, new centroids of the clusters are formed, and an iteration or series of iterations can occur to minimize the size of the clusters and allow for optimal centroid determination. Next, new data points can be measured and compared to the centroids and clusters and identified with those clusters.

[0082] Ensemble learning algorithms may be used. These algorithms use multiple learning algorithms to achieve better predictive performance than would be possible from any of the constituent learning algorithms alone. Ensemble learning algorithms perform the task of searching a hypothesis space to find hypotheses that are suitable for making good predictions for a particular problem. Even if the hypothesis space contains hypotheses that are well suited to a particular problem, finding a good hypothesis can be very difficult. Ensemble algorithms combine multiple hypotheses to form a better hypothesis. The term ensemble is usually used to refer to the method of generating multiple hypotheses using the same base learner. The broader concept of a multi-classifier system also encompasses the hybridization of hypotheses that are not derived from the same base learner.

[0083] Because evaluating the predictions of an ensemble typically requires more computation than evaluating the predictions of a single model, ensembles can be thought of as a way to compensate for poor learning algorithms by performing a lot of extra computation. Fast algorithms such as decision trees are commonly used in ensemble methods such as random forests, but slower algorithms can also benefit from ensemble methods.

[0084] Ensembles are themselves supervised learning algorithms because they can be used to make predictions after training. A trained ensemble therefore represents a single hypothesis, although this hypothesis is not necessarily contained within the hypothesis space of the model it was built on. Ensembles can therefore have more flexibility in the functions they can represent. This flexibility theoretically allows them to overfit the training data more than a single model, but in practice, some ensemble methods (especially bagging) tend to mitigate problems associated with overfitting the training data.

[0085] Empirically, ensemble algorithms tend to perform better when there is a significant degree of diversity among the models. Therefore, many ensemble methods aim to promote diversity among the models they combine. While counterintuitive, more random algorithms (such as random decision trees) can be used to generate more powerful ensembles than highly cautious algorithms (such as entropy-reduced decision trees). However, the use of a variety of powerful learning algorithms has been shown to be more effective than using techniques that attempt to simplify models to promote diversity.

[0086] The number of component classifiers in an ensemble has a significant impact on the accuracy of prediction. This becomes even more important for online ensemble classifiers, where the size of the ensemble and the volume and velocity of the big data stream are determined a priori. Theoretical frameworks suggest that there is an ideal number of component classifiers in an ensemble, and that having more or fewer classifiers than this number will result in a decrease in accuracy. Theoretical frameworks also suggest that using the same number of independent component classifiers as class labels will result in the highest accuracy.

[0087] Some common types of ensembles include Bayesian optimal classifiers, bootstrap aggregation (bagging), boosting, Bayesian model averaging, Bayesian model combination, and model bucketing and stacking. Figure 13 shows an exemplary ensemble learning algorithm 1300 in which bagging is performed in parallel (1310) and boosting is performed sequentially (1320).

[0088] During bagging, multiple subsets are created from the original dataset with replacement. A base model is created for each of the subsets. The base models can be run in parallel and independent of each other. The final predictions can be determined by combining the predictions from all base models. Bagging can be an efficient way to reduce the variability of a model.

[0089] Boosting, on the other hand, is an iterative process in which each subsequent model attempts to correct the errors of the previous model. Successive models depend on the previous model. First, a subset may be created from the original dataset. Initially, all data points are given equal weight. Next, a base model is created on this subset. This model is then used to make predictions for the entire dataset. The error is calculated using the actual and predicted values. Incorrectly predicted observations are given a higher weight. Next, a second model is created and used to make predictions for the dataset. The second model attempts to correct the errors from the first model. Next, subsequent models are created, each correcting the errors of the previous model. The final model (i.e., a strong learner) is a weighted average of all previous models (i.e., weak learners). Thus, using boosting, several weak learners can be combined to form a strong learner.

[0090] A neural network is a network or circuit of neurons, or in the modern sense, an artificial neural network (ANN), composed of artificial neurons or nodes. The connections of biological neurons are modeled as weights. Positive weights reflect excitatory connections, while negative values ​​imply inhibitory connections. The inputs are modified by the weights and summed using linear combinations. An activation function can control the amplitude of the output. For example, the allowed range of the output is usually 0 to 1, but can also be -1 to 1.

[0091] These artificial networks can be used in predictive modeling, adaptive control, and other applications, and can be trained through datasets. Self-learning resulting from experience can occur within the network, allowing it to draw conclusions from complex and seemingly unrelated sets of information.

[0092] For completeness, biological neural networks consist of groups of chemically connected or functionally associated neurons. One neuron may be connected to many other neurons, and the total number of neurons and connections in the network may vary widely. Connections, called synapses, are usually formed from axons to dendrites, although dendritic synapses and other connections are also possible. Apart from electrical signaling, other forms of signaling result from the diffusion of neurotransmitters.

[0093] Artificial intelligence, cognitive modeling, and neural networks are information processing paradigms inspired by the way biological nervous systems process data. Artificial intelligence and cognitive modeling attempt to simulate some of the properties of biological neural networks. In the field of artificial intelligence, ANNs have been successfully applied to speech recognition, image analysis, and adaptive control to build software agents or autonomous robots (in computer and video games).

[0094] A neural network, in the case of an ANN or simulated neural network (SNN), is a group of interconnected natural or artificial neurons that uses a mathematical or computational model for information processing based on a connectionist approach to computation. In most cases, an ANN is an adaptive system that changes its structure based on external or internal information flowing through the network. More practically, neural networks are nonlinear statistical data modeling or decision-making tools. They can be used to model complex relationships between inputs and outputs and to find patterns in data.

[0095] An ANN comprises a network of simple processing elements (artificial neurons) that can exhibit complex global behavior determined by the connections between the processing elements and element parameters.

[0096] One classic type of artificial neural network is the recurrent Hopfield network. The utility of ANN models lies in the fact that they can be used to estimate functions from observations and even use functions. Unsupervised neural networks can be used to learn representations of inputs that capture salient features of the input distribution, and more recently, deep learning algorithms that can implicitly learn distribution functions for observed data. Training with neural networks is particularly useful in applications where the complexity of the data or the task makes the design of such functions impractical by hand.

[0097] Neural networks can be used in a variety of fields, and the tasks to which ANNs are applied tend to cover broad categories, including function approximation or regression analysis, including time series prediction and modeling, pattern and sequence recognition, classification, including novelty detection and sequential decision making, and data processing, including filtering, clustering, blind signal separation and compression.

[0098] Areas of application of ANNs include identification and control of nonlinear systems (vehicle control, process control), game playing and decision making (backgammon, chess, racing), pattern recognition (radar systems, face identification, object recognition), sequence recognition (gesture, speech, handwritten text recognition), medical diagnosis, financial applications, data mining (or knowledge discovery in databases, or "KDD"), visualization, and email spam filtering. For example, it is possible to create semantic profiles of user interests resulting from photographs trained for object recognition.

[0099] 14 illustrates an exemplary neural network 1400. The neural network 1400 has an input layer represented by multiple inputs, such as 1410a and 1410b. The inputs 1410a and 1410b are provided to a hidden layer, shown to include nodes 1420a, 1420b, 1420c, and 1420d. These nodes 1420a, 1420b, 1420c, and 1420d are combined to generate an output layer output 1430. While neural networks perform simple processing via a hidden layer of simple processing elements, nodes 1420a, 1420b, 1420c, and 1420d, these nodes can exhibit complex global behavior determined by the connections between the processing elements and element parameters.

[0100] Neural network 1400 may be implemented in hardware. FIG. 15 illustrates a hardware-based neural network 1500. In an exemplary embodiment, but not limited to, hardware-based neural network 1500 includes RAM 1510, preferably containing weights, biases, directions, etc., a multiplexer 1520, a finite state machine (FSM) control unit 1530, an operand selector 1540, an arithmetic logic unit 1560, and a multiplier unit 1570. Inputs 1580 are provided to neural network 1500, which generates outputs 1550. For example, in the embodiment illustrated in FIG. 15, inputs 1580 are provided to multiplexer 1520, which forwards the selected input to multiplexer unit 1570. Weights, biases, directions, etc. from RAM 1510 are also forwarded to multiplexer unit 1570. The multiplexer unit 1570 applies weights, biases, directions, etc. to selected inputs and forwards the results to the arithmetic logic unit 1560. The operand selector 1540 is configured to keep track of the number of operands being multiplied. Information from the FSM control unit 1530 is also forwarded to the arithmetic logic unit 1560. The overall operation of the neuron block is controlled using the FSM control unit 1530. The arithmetic logic unit 1560 performs one or more arithmetic operations and generates the output 1550.

[0101] Treatment for cardiac conditions, such as cardiac arrhythmias, often requires detailed mapping of cardiac tissue, chambers, veins, arteries, and / or electrical pathways. Such mapping can be performed via an EP study, in which spatially resolved electrical potentials are detected by a mapping catheter introduced into the ventricles. This EP study, so-called electroanatomical (EA) mapping, thus provides 3D mapping data that can be displayed on a monitor. Often, mapping and processing functions (e.g., ablation) are provided by a single catheter or a group of catheters, such that the mapping catheter simultaneously operates as a processing (e.g., ablation) catheter.

[0102] Cardiac mapping can be implemented using one or more techniques. As an example of a first technique, cardiac mapping can be implemented by sensing electrical properties of cardiac tissue as a function of precise location within the heart, e.g., local activation time (LAT). Corresponding data can be acquired with one or more catheters advanced into the heart using catheters having electrical and position sensors at their distal tips. By way of example, position and electrical activity can initially be measured at about 10 to about 20 points on the inner surface of the heart. These data points can generally be sufficient to generate a preliminary reconstruction or map of the cardiac surface of satisfactory quality. The preliminary map can be combined with data taken at additional points to generate a more comprehensive map of the cardiac electrical activity. In clinical settings, it is not uncommon to accumulate data at 100 or more sites to generate a detailed, comprehensive map of the cardiac chamber's electrical activity. The detailed map can then serve as a basis for making decisions regarding therapeutic actions, e.g., tissue ablation, to alter the propagation of the cardiac electrical activity and restore normal cardiac rhythm.

[0103] A catheter containing a position sensor can be used to determine the trajectory of each point on the heart's surface. These trajectories can be used to infer motion properties, such as the contractile force of the tissue. A map indicative of such motion properties can be constructed when trajectory information is sampled at a sufficient number of points within the heart.

[0104] Electrical activity at a point within the heart is typically measured by advancing a catheter containing an electrical sensor at or near its distal tip to that point within the heart, contacting tissue with the sensor, and acquiring data at that point. Multipolar electrode catheters can be implemented using any applicable shape, such as a linear catheter with multiple electrodes, a balloon catheter with multiple electrodes and containing electrodes distributed on multiple barbs forming a balloon, lasso, or loop catheter, or any other applicable shape.

[0105] According to one example, a multi-electrode catheter can be advanced into a cardiac chamber. An anterior-posterior (AP) and lateral fluorogram can be obtained to establish the position and orientation of each of the electrodes. An electrogram can be recorded from each of the electrodes in contact with the cardiac surface relative to a temporal reference, such as the occurrence of a P wave in sinus rhythm from a surface ECG. A system as further disclosed herein can distinguish between electrodes that record electrical activity and those that do not record because they are not in close proximity to the endocardial wall. After the first EGM is recorded, the catheter can be repositioned, and fluorograms and EGMs can be recorded again. An electrical map can then be constructed by iterating the above process.

[0106] According to one example, cardiac mapping can be generated based on the detection of intracardiac electrical fields. Non-contact techniques can be implemented to simultaneously acquire large amounts of cardiac electrical information. For example, a catheter having a distal end portion can include a series of sensor electrodes distributed over its surface and connected to insulated electrical conductors for connection to signal detection and processing means. The size and shape of the end portion can be such that the electrodes are significantly spaced from the walls of the ventricle. The intracardiac electrical fields can be detected during a single cardiac beat. According to one example, the sensor electrodes can be distributed on a series of circumferentially spaced apart planes. These planes can be perpendicular to the major axis of the catheter end portion. At least two additional electrodes can be provided adjacent the ends of the major axis of the end portion. As a more specific example, the catheter can include four circumferences with eight electrodes equiangularly spaced on each circumference. Thus, in this specific implementation, the catheter can include at least 34 electrodes (32 circumferentially and two at the ends).

[0107] According to another example, an EP cardiac mapping system and technique based on a non-contact, non-expanding multi-electrode catheter can be implemented. EGMs can be obtained with a catheter having multipolar electrodes (e.g., 42-122 electrodes). According to this implementation, knowledge of the relative geometry of the probe and endocardium can be obtained, for example, through an independent imaging modality such as transesophageal echocardiography. After the independent imaging, non-contact electrodes can be used to measure cardiac surface potentials from which a map can be constructed. This technique can include the following steps (after the independent imaging step): (a) measuring potentials with multiple electrodes placed on a probe positioned within the heart; (b) determining the geometric relationship between the probe surface and the endocardium surface; (c) generating a matrix of coefficients representing the geometric relationship between the probe surface and the endocardium surface; and (d) determining the endocardium potentials based on the electrode potentials and the matrix of coefficients.

[0108] According to another example, a technique and apparatus for mapping the electrical potential distribution of a cardiac chamber may be implemented. An intracardiac (IC) multi-electrode mapping catheter assembly may be inserted into a patient's heart. The mapping catheter assembly may include a multi-electrode array with an integral reference electrode, or preferably, a companion 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 by the reference electrode or by a reference catheter in contact with the endocardial surface. A suitable electrode array catheter may carry several individual electrode sites (e.g., at least 24 sites). Additionally, this exemplary technique may be implemented with knowledge of the location of each of the electrode sites on the array, as well as knowledge of the cardiac geometry. These locations are preferably determined by the technique of impedance plethysmography.

[0109] According to another example, a process for measuring electrophysiological data of a heart chamber may be implemented. The method may include, in part, positioning a set of active and passive electrodes on the heart, applying a current to the active electrodes thereby generating an electric field within the heart chamber, and measuring the electric field at the passive electrode sites. The passive electrodes are included in an array positioned on an inflatable balloon of a balloon catheter. In a preferred embodiment, the array is said to have 60-64 electrodes.

[0110] According to another example, cardiac mapping may be performed using one or more ultrasound transducers. The ultrasound transducers may be inserted into a patient's heart and may acquire multiple ultrasound slices (e.g., two-dimensional or three-dimensional slices) at various positions and orientations within the heart. The position orientation of a given ultrasound transducer may be known, and the acquired ultrasound slices may be stored for later display. One or more ultrasound slices corresponding to a later probe (e.g., a treatment catheter) position may be displayed, and the probe may be overlaid on one or more ultrasound slices.

[0111] According to another example, a body patch and / or body 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), and the position of the catheter may be determined by the system based on signals transmitted and received between one or more electrodes of the catheter and the body patch and / or body 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). The biometric data may be associated with the determined position of the catheter, such that a rendering of the patient's body part (e.g., heart) as determined by the position of the catheter may be displayed, showing the biometric data superimposed on the shape of the body part.

[0112] Electrical signals, such as ECG signals, are often detected before and / or during cardiac procedures. For example, ECG signals can be used to identify potential locations in the heart where signals due to arrhythmias originate. In general, an ECG is a signal that describes the electrical activity of the heart. ECG signals can also be used to map portions of the heart. ECG signals are generated by the contraction (depolarization) and relaxation (repolarization) of the atrial and ventricular muscles of the heart. As shown by signal 1600 in FIG. 16, the ECG signal includes a P wave 1610 (due to atrial depolarization), a QRS complex 1620 (due to atrial repolarization and ventricular depolarization), and a T wave 1630 (due to ventricular repolarization). To record ECG signals, electrodes can be placed at specific locations on the body or can be positioned within the body via a catheter. Artifacts (e.g., noise) are unwanted signals that blend with electronic signals, such as ECG signals, and can sometimes create barriers to diagnosing and / or treating cardiac conditions. Artifacts in the electrical signal can be baseline wander, power line interference, electromyogram (EMG) noise, power line noise, muscle noise, and the like.

[0113] Additionally, biometric (e.g., biopotential) patient monitors may use surface electrodes to make measurements of biopotentials, such as ECG or electroencephalogram (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 electrical current, known as electrical impedance, characterizes the effectiveness of the connection. Typically, the higher the impedance, the lower the fidelity of the measurement. Several mechanisms may be responsible for the lower fidelity.

[0114] FIG. 17 is a diagram of an exemplary system 1720 capable of implementing one or more features of the presently disclosed subject matter. All or a portion of the system 1720 can be used to collect information for a training dataset and / or to implement a training model. The system 1720 can include a component, such as a catheter 1740, configured to injure a tissue region of an internal organ. The catheter 1740 can also be further configured to obtain biometric data, including electronic signals. While the catheter 1740 is shown as a point catheter, it will be understood that any shape of catheter including one or more elements (e.g., electrodes) can be used to implement embodiments disclosed herein. The system 1720 includes a probe 1721 having a shaft that can be navigated by a physician 1730 into a body part, such as a heart 1726, of a patient 1728 reclining on a table 1729. According to an embodiment, multiple probes may be provided; for simplicity, a single probe 1721 is described in this example, although it will be understood that the probe 1721 may represent multiple probes. As shown in FIG. 17 , a physician 1730 may insert a shaft 1722 through a sheath 1723 while manipulating the distal end of the shaft 1722 using a manipulator near the proximal end of the catheter 1740 and / or deflection from the sheath 1723. As shown in inset 1725, the catheter 1740 may be fitted at the distal end of the shaft 1722. The catheter 1740 may be inserted through the sheath 1723 in a collapsed state and then expanded within the heart 1726. As further disclosed herein, the catheter 1740 may include at least one ablation electrode 1747 and a catheter needle.

[0115] According to embodiments, catheter 1740 can be configured to ablate a tissue region of a chamber of heart 1726. Inset 1745 shows a close-up of catheter 1740 inside a chamber of heart 1726. As shown, catheter 1740 can include at least one ablation electrode 1747 coupled to the body of the catheter. According to other embodiments, multiple elements can be connected via splines that form the shape of catheter 1740. One or more other elements (not shown) can be provided, which can be any element configured to perform ablation or acquire biometric data, and can be an electrode, a transducer, or one or more other elements.

[0116] According to embodiments disclosed herein, an ablation electrode, such as electrode 1747, can be configured to provide energy to a tissue region of a body organ, such as heart 1726. The energy can be thermal energy and can cause damage to the tissue region starting at the surface of the tissue region and extending through the thickness of the tissue region.

[0117] According to embodiments disclosed herein, the biological data may include one or more of LAT, electrical activity, topology, bipolar mapping, dominant frequency, impedance, etc. LAT may be a point at a threshold activation corresponding to local activation calculated based on a normalized initial starting point. Electrical activity may be any applicable electrical signal that can be measured based on one or more thresholds and detected and / or enhanced based on signal-to-noise ratio and / or other filters. Topology may correspond to the physical structure of a body part or a portion of a body part, or may correspond to changes in the physical structure for different portions of a body part or for different body parts. The dominant frequency may be a frequency or range of frequencies prevalent in a portion of a body part and may differ in different portions of the same body part. For example, the dominant frequency of the pulmonary veins of a heart may be different from the dominant frequency of the right atrium of the same heart. Impedance may be a resistance measurement in a given region of a body part.

[0118] 17 , the probe 1721 and catheter 1740 may be connected to a console 1724. The console 1724 may include a processor 1741, such as a general-purpose computer with suitable front-end and interface circuitry 1738 for sending signals to and receiving signals from the catheter, as well as for controlling other components of the system 1720. In some embodiments, the processor 1741 may be further configured to receive biometric data, such as electrical activity, and determine whether a given tissue region conducts electricity. According to one embodiment, the processor may be external to the console 1724, located, for example, in the catheter, in an external device, in a mobile device, in a cloud-based device, or may be a stand-alone processor.

[0119] As noted above, processor 1741 may include a general-purpose computer, which may be programmed with software to perform the functions described herein. The software may be downloaded to the general-purpose computer in electronic form, for example, over a network, or alternatively or additionally, may be provided and / or stored on non-transitory tangible media, such as magnetic, optical, or electronic memory. The exemplary configuration shown in FIG. 17 may be modified to implement embodiments disclosed herein. Embodiments of the present disclosure may be similarly applied using other system components and configurations. Additionally, system 1720 may include additional components, such as elements for sensing electrical activity, wired or wireless connectors, processing and display devices, etc.

[0120] According to one embodiment, the display 1727 connected to the processor (e.g., processor 1741) may be located at a remote location, such as a separate hospital, or a separate healthcare provider network. Additionally, the system 1720 may be part of a surgical system configured to obtain anatomical and electrical measurements of a patient's organs, such as the heart, and to perform cardiac ablation procedures. One example of such a surgical system is the Carto® system sold by Biosense Webster.

[0121] System 1720 can also, and optionally, obtain biometric data, such as anatomical measurements of the patient's heart, using ultrasound, computed tomography (CT), magnetic resonance imaging (MRI), or other medical imaging techniques known in the art. System 1720 can obtain electrical measurements using a catheter, ECG, or other sensors that measure electrical properties of the heart. The biometric data, including the anatomical and electrical measurements, can then be stored in memory 1742 of mapping system 1720, as shown in FIG. 17 . The biometric data can be transmitted from memory 1742 to processor 1741. Alternatively or additionally, the biometric data can be transmitted to server 1760, which can be local or remote, using network 1762.

[0122] The network 1762 may be any network or system commonly known in the art, such as an intranet, a LAN, a WAN, a MAN, a direct or series of connections, a cellular telephone network, or any other network or medium capable of facilitating communication between the mapping system 1720 and the server 1760. The network 1762 may be wired, wireless, or a combination thereof. Wired connections may be implemented using Ethernet, USB, RJ-11, or any other wired connection commonly known in the art. Wireless connections may be implemented using Wi-Fi, WiMAX, Bluetooth, IR, cellular networks, satellite, or any other wireless connection method commonly known in the art. Additionally, several networks may function alone or in communication with each other to facilitate communication within the network 1762.

[0123] In some cases, server 1760 may be implemented as a physical server. In other cases, server 1762 may be implemented as a virtual server, a public cloud computing provider (e.g., Amazon Web Services (AWS)).

[0124] According to an exemplary embodiment, server 1760 may be implemented as or in communication with a processor that stores a machine learning algorithm, such as neural network 1790. In another embodiment, neural network 1790 may be implemented within console 1724. For example, without limitation, neural network 1790 may be implemented in one or more CPU processors, one or more GPU processors, one or more FPGA chips, or an ASIC dedicated to performing deep learning calculations (for example), such as the Intel® Nervana™ Neural Network Processor. According to an exemplary embodiment, neural network 1790 may be located, without limitation, in a medical procedure room, on a server or processor within a hospital or medical facility, on a remote server or processor, or in the cloud.

[0125] The control console 1724 may be connected by cable 1739 to body surface electrodes 1743, which may include adhesive skin patches that are affixed to the patient 1728. The processor, in conjunction with the current tracking module, may determine position coordinates of the catheter 1740 within the patient's body part (e.g., heart 1726). The position coordinates may be based on impedance or electromagnetic fields measured between the body surface electrodes 1743 and electrodes 1747 or other electromagnetic components of the catheter 1740. Additionally or alternatively, location pads may be positioned on the surface of the table 1729 and may be separate from the table 1729.

[0126] The processor 1741 may include a real-time noise reduction circuit, typically configured as a field programmable gate array (FPGA), followed by an analog-to-digital (A / D) ECG or EMG signal conversion integrated circuit. The processor 1741 may communicate signals from the A / D ECG or EMG circuitry to another processor and / or may be programmed to perform one or more functions disclosed herein.

[0127] The control console 1724 may also include an input / output (I / O) communication interface that allows the control console to transfer signals to and / or from the electrodes 1747 .

[0128] During a procedure, processor 1741 may facilitate presentation of body part rendering 1735 to physician 1730 on display 1727 and store data representing body part rendering 1735 in memory 1742. Memory 1742 may include any suitable volatile and / or non-volatile memory, such as RAM or a hard disk drive. In some embodiments, physician 1730 may be able to manipulate body part rendering 1735 using one or more input devices, such as a touchpad, mouse, keyboard, gesture recognizer, etc. For example, the input device may be used to change the position of catheter 1740 so that rendering 1735 is updated. In an alternative embodiment, display 1727 may include a touchscreen that may be configured to receive input from physician 1730 in addition to presenting body part rendering 1735.

[0129] The processor 1741 may acquire multiple electrical signals measuring electrical properties of the heart obtained by the catheter 1740, ECG, or other sensors, as described above. The processor 1741 may apply an algorithm to select the optimal heart beat. In some embodiments, the algorithm may include dynamic filter filtering of the electrical signals to determine which electrical signals are integrated as "points" in a "point cloud" representing a desired anatomical structure, such as a heart chamber.

[0130] The algorithm may include multiple filters, including, but not limited to, position stability, internal distance, catheter filter, cycle length, cycle length stability, cavity activity, position density, respiration, pattern matching, LAT, LAT stability, unipolar slope, bipolar voltage, tissue proximity index (TPI), segmentation stability, and noise level. However, this list is not exhaustive, and dynamic filters may utilize other parameters.

[0131] In some embodiments, the algorithm may include two phases of filtering: a first phase and a second phase. In the first phase, one or more first filters may be applied to all of the collected electrical signals. The first filters may include, but are not limited to, position stability, internal distance, catheter filter, cycle length, ventricular activity, position density, respiratory cycle indication, and pattern matching. If the electrical signals meet certain criteria in the first phase, one or more second filters may be applied in the second phase. The second filters may include, but are not limited to, relative LAT stability, slope of unipolar signals, bipolar voltage, TPI, and segmentation stability.

[0132] In some embodiments, one or more parameters may be calculated in the second phase depending on the characteristics of the signal. For example, in the second phase, the signal may be analyzed to determine whether the signal has LAT. Furthermore, the bipolar voltage may be calculated and analyzed. The signal may also be analyzed to determine whether the electrodes are in close proximity to tissue. The force value (grams) may also be analyzed. The relative position of the point within the respiratory cycle may also be analyzed. In some embodiments, an algorithm determines whether the bipolar voltage is high or low. For example, a signal may be considered to have a "high" bipolar voltage if the bipolar voltage is equal to or greater than 0.1 mV and a "low" bipolar voltage if the bipolar voltage is less than 0.1 mV. However, as will be understood by those skilled in the art, the "high" and "low" bipolar voltage thresholds may vary based on many factors that may be specific to a particular patient. The thresholds may be entered into the system by the physician (e.g., via a graphical user interface) for the instance. If a signal with LAT is determined to have a "high" bipolar voltage, the signal's unipolar slope and relative LAT stability can be calculated and analyzed. The signal's unipolar slope, relative LAT stability, and TPI can be calculated and analyzed.

[0133] The algorithm may determine whether a signal is segmented. The algorithm may also calculate the TPI of the signal. The algorithm may calculate the segmentation stability of the signal. Each of these attributes may be an independent variable for machine learning.

[0134] Multiple second filters may be applied to signals that have passed the first filter. In some embodiments, the second filter applied to each signal is based on the characteristics of each signal. For example, the following filter may be applied to signals determined to have LAT and high voltage, such that the signals have one or more of the following characteristics: a unipolar slope of 0.03 mV / ms or greater and a relative LAT stability of 3 ms. The provided thresholds are merely exemplary, and various other thresholds may be utilized within the algorithm. Furthermore, signals determined to have LAT and low voltage, in addition to having a unipolar slope of 0.03 mV / ms or greater and a relative LAT stability of 3 ms, may also require a TPI indicating that the signal was acquired when the catheter 14 contacted tissue and passed through the second filter.

[0135] Machine learning may learn, for example, that a signal determined to be free of any LAT and not fragmented may require a TPI indicating that the signal was collected when the catheter made contact with the tissue to be considered a "good" point. In addition to having a TPI indicating contact, a signal determined to be free of any LAT and fragmented may require satisfactory fragmentation stability to be considered a "good" point.

[0136] Signals that do not pass the second filter can be discarded. Signals that pass the second filter can be selected to become points in the point cloud to create an EP map of a target mapping site, such as a heart chamber.

[0137] According to one embodiment, a neural network 1790 may be provided to select and detect the best heart beat at each spatial location as a cardiac mapping annotation and to improve the determination of the cardiac mapping annotation. According to one exemplary embodiment, the neural network 1790 receives input data to train the neural network. The input data is preferably a set of attributes related to the heart beats. Non-limiting examples of input data or attribute data may include: - ECG signals received from mapping electrodes around the fiducial annotation of the EP map; - ECG signals received by body surface electrode(s) around the reference annotation; - intracardiac EGM signals received by an electrode or by a bipolar electrode pair of a catheter such as a mapping catheter, the EMG signals may be bipolar or unipolar EGMs of electrodes of a mapping catheter around a reference annotation in the EP map; - the spatial position of the mapping electrodes at the reference annotation, with or without respiration correction; - Whether the beat is incorporated into the EP map (e.g., whether the beat passes the filter of the beat selector algorithm); - the local annotation times of the acquired heartbeats, including any corrections made by the physician; - TPI of the mapping electrode at the time of reference annotation, - the force value detected by the force sensor of the mapping catheter at the time of the reference annotation (typically measured in grams); the position of the fiducial annotation within the respiratory cycle when lung motion is being monitored; -Respiration gating state as a Boolean output, - the difference between the current respiratory cycle length and the previous cycle length, - the ratio of the current respiratory cycle length divided by the previous cycle length, the difference between the current respiratory cycle length and the average or median respiratory cycle length; - the ratio of the current respiratory cycle length divided by the average or mean respiratory cycle length; the distance of the mapping electrodes of the current reference annotation from the spatial location of the same electrodes of the previous reference annotation, which can be measured either by their actual positions or by their respiration-corrected positions; - an indication of whether any body surface activation (i.e., V interference) was present during this heartbeat; A separate Boolean value (i.e., 0 or 1) as an input indicating whether the beat has been removed from the map by a physician; and - Any other EP data measured by catheter electrodes.

[0138] In one embodiment, one or more of the input data are provided to neural network 1790. The input data can be stored in a variety of locations, including, but not limited to, at a hospital or medical facility, a remote server location, or in the cloud. The training data can be transferred manually or automatically to a storage device associated with the neural network in real time, at predetermined intervals, on demand, when mapping system 1720 is idle, etc. As a result, neural network 1790 can train in real time or at predetermined times or intervals.

[0139] In one embodiment, the output of the neural network 1790 may include, but is not limited to: a decision, such as a discrete Boolean value, indicating whether the current heartbeat improved the attribute over the previous heartbeat; and -Determining accurate cardiac mapping annotations.

[0140] In the current state of the art, if an EA point in a cardiac EP mapping system is not good enough (e.g., because the ECG at this EA point was noisy), the physician removes this EA point and acquires a new EA point at approximately the same spatial location.

[0141] According to an exemplary embodiment of the present application, a machine learning algorithm is trained using a pair of deleted and reacquired EA points obtained at approximately the same spatial location (i.e., at the same location to some extent). Given two EA points, the machine learning algorithm learns that the reacquired EA point has better characteristics than the deleted EA point. Once the machine learning algorithm is trained with sufficient data, given any two EA points in a pair, the algorithm can predict which one has better characteristics. The EP mapping system can use this information to automatically replace the EA point with insufficient characteristics during the EP instance. Before acquiring a point for a spatial location, the mapping system automatically uses it in mapping and ignores any other points acquired at that location. In this system, multiple points can be acquired, and the system can determine which points should be used in the mapping system. The input of the machine learning algorithm preferably includes one or more of the input data described above.

[0142] According to an exemplary embodiment, neural network 1790 may include a recurrent neural network (RNN), such as a convolutional neural network (CNN) or a long short-term memory (LSTM) neural network. A CNN is a deep learning algorithm preferably used in the fields of computer vision and / or image recognition. A CNN assigns importance (learnable weights) to various aspects or features in an input image to distinguish one from another. An LSTM neural network is an RNN with feedback connections used for deep learning.

[0143] In one embodiment, the EP mapping system vendor may ship the system with a pre-trained network. Hospitals can preferably continue to train the system. In one embodiment, a single model may be maintained for all hospitals, or for a group of hospitals, or each hospital may maintain its own model.

[0144] In an exemplary embodiment, after each training, the trained model, including its output, can be run against a standard database, such as a gold standard database, to confirm its accuracy. In an exemplary embodiment, if the accuracy of the newly trained model is below a threshold, or alternatively, if the accuracy of the newly trained model is lower than the accuracy of the previous model, the model can be discarded. Similarly, if the accuracy of the newly trained model is equal to or higher than a threshold, or alternatively, if the accuracy of the newly trained model is higher than the accuracy of the previous model, the model can be published to an in-field EP mapping system. In an exemplary embodiment, publishing a new model can be performed manually, for example, by an operator downloading a file from a web address and uploading it to the mapping system 1720. Alternatively, the new model can be pushed to the in-field mapping system 1720 via the internet.

[0145] In EP mapping, a catheter, typically with multiple mapping electrodes positioned along the body of the catheter near its distal end, is inserted into a cavity of the heart along with an internal probe. Time-varying ECG signals are recorded at multiple contact points between the mapping electrodes and the cardiac tissue. The ECG electrodes are then moved to different contact locations within the cardiac tissue and the process is repeated. Metrics regarding cardiac function are then calculated from the local ECG signals, and the signals are spatially mapped across the surface of the cardiac cavity. The mapping assists medical professionals in identifying areas of cardiac dysfunction.

[0146] Electrical sources within the heart, such as the sinoatrial (SA) and atrioventricular (AV) nodes, initiate waves of electrical activity that propagate throughout the heart, triggering the atrial and ventricular muscle tissue to contract in a natural sinus rhythm. As the activity wavefront reaches multiple mapping electrodes during each cardiac cycle, unique ECG waveforms are sensed at the multiple mapping electrodes. These waveforms are time-shifted due to the different arrival times of the same wavefront at different electrodes contacting tissue at different spatial locations along the surface of the heart cavity.

[0147] The arrival times of ECG waveforms sensed at multiple mapping electrodes can be used to map the propagation time and / or velocity of activity waves across the heart, with the mapping of activity waves being performed with respect to a single time reference representing the cardiac cycle, referred to as the reference annotation time.

[0148] The reference annotation time can be calculated by processing ECG signals obtained from body surface electrodes or from an IC reference electrode on an additional catheter placed in contact with the surface of the heart chamber. Typically, the physician specifies whether the reference annotation time is calculated from the BS or IC channel depending on the suspected lesion. However, if the assigned ECG reference channel fails during the mapping procedure (e.g., due to poor reference electrode contact, system noise, or other impairments), remapping must be performed. Cardiac remapping is time-consuming and uncomfortable for the patient.

[0149] 20 is an exemplary diagram of multiple electrodes sampling multiple contact points within a heart cavity, according to an embodiment of the present application. In the exemplary embodiment shown in FIG. 20, an IC catheter 2024A is inserted into the coronary sinus 2062 of the heart 2034. The catheter preferably includes five reference ECG electrodes 2060A-2060E positioned at points along the length of the body of the catheter 2024A near its distal end 2065. The reference ECG electrodes 2060A-2060E are used to contact multiple tissue points on the surface of the coronary sinus and measure respective reference ECG signals at the multiple contact points.

[0150] Similarly, IC catheter 2024B is inserted into right ventricle 2034. The catheter includes five mapping electrodes 2070A-2070E at points along the body of catheter 2024B near distal end 2075. Mapping electrodes 2070A-2070E are used to contact multiple tissue points on the surface of the right ventricle and measure respective mapping ECG signals at the multiple contact points.

[0151] The five reference ECG electrodes 2060A-2060E and multiple mapping electrodes 2070A-2070E shown in FIG. 20 are shown merely for conceptual clarity and are not intended to limit the scope of the claims. In alternative embodiments, any suitable number of mapping and reference electrode configurations can be used. For example, instead of a single electrode for unipolar ECG detection, electrodes can be arranged in pairs for bipolar ECG detection. Any suitable number of mapping and reference catheters can be used in any suitable configuration. Reference catheter 2024A and mapping catheter 2024B, or any number of catheters, can be navigated to suitable locations within heart 2034 to perform the functions described herein.

[0152] An ECG signal interface, such as processor 1741 (shown in FIG. 17), receives and processes signals from body surface electrodes, such as body surface electrode 1743 (shown in FIG. 17), IC ECG reference electrodes 2060A-2060E, and mapping electrodes 2070A-2070E. Cardiac mapping is performed by moving multiple mapping electrodes 2070A-2070E over the surface of the cardiac tissue and recording the ECG at each contact point along with the position of the electrodes when the ECG waveform was recorded. Alternatively, cardiac mapping may involve inserting the reference and / or mapping electrodes externally into the patient's body, for example through the thoracic cavity, so as to contact the surface of the epicardial tissue.

[0153] The data recorded at each mapping point is also referred to as a mapping annotation. However, because the activation wavefront propagates to different electrodes at spatially distinct points on the endocardium, local ECG signals arrive at different points, and therefore different electrodes, at different times. The difference in arrival time of the ECG signal at a given ECG electrode indicates the local activation wavefront velocity. The measured time difference of the ECG signal arriving at a particular mapping electrode for a particular cardiac cycle or beat relative to a single timing reference representing the cardiac cycle timing of the heart is known herein as the LAT. Similarly, the single timing reference is referred to as the reference annotation time.

[0154] A typical cardiac map includes a mapping of the LAT at multiple different points on the cardiac surface, a propagation map showing activation wavefronts at different times throughout the heart, and the intrinsic voltage of the ECG at the same given point. A user, such as a physician, can then use variations from the expected activation wavefront and / or voltage shown on the cardiac map to detect areas of cardiac dysfunction, such as atrial or ventricular tachycardia. Ablation therapy may be used, for example, to correct the dysfunction.

[0155] In some embodiments, the processor 1741 uses reference ECG signals acquired from the BS electrode 1743 and / or from the IC reference electrodes 2060A-2060E in calculating the reference annotation time. The acquired IC mapping ECG signals are acquired as the mapping electrodes are moved across the cardiac tissue to be recorded, and the ECG is recorded to obtain mapping data points. However, during a medical procedure, the IC reference electrodes are in contact with the cardiac tissue and do not move. Depending on the type of lesion expected, the physician typically specifies whether the reference annotation time is to be calculated from an ECG acquired from the BS electrode 1743 or from the IC reference electrodes 2060A-2060E. [Example]

[0156] In one exemplary embodiment, as described in more detail below with respect to Figures 18 and 19, a machine learning algorithm such as a neural network can utilize attributes of two heartbeats as input, learn to identify which heartbeat exhibits better characteristics based on predetermined criteria, and output a binary result. Exemplary attributes for identifying which heartbeat is better include, but are not limited to, heartbeats with less noise, heartbeats with more pronounced LAT activation, heartbeats with more pronounced late potentials, heartbeats with more stable divisions, heartbeats with less V-interference, etc. In one embodiment, the heartbeats are derived from stable arrhythmias.

[0157] Conventional EP mapping systems acquire the first heartbeat for each spatial location or use a rule-based system to select the best heartbeat for each spatial location. For example, if a physician acquires the first heartbeat at a particular spatial location, they stop acquiring subsequent heartbeats even if the first heartbeat exhibits insufficient characteristics. The system and method disclosed in this embodiment collects multiple heartbeats at a particular spatial location and learns to select the best heartbeat at each spatial location through machine learning.

[0158] FIG. 18 is an exemplary embodiment of a CNN 1800 that can receive attributes of two heartbeats and learn to identify which heartbeat is better, according to the present application.

[0159] Referring to FIG. 18 , in one exemplary embodiment, the neural network 1800 is a CNN. The CNN 1800 preferably receives input data 1810. In one embodiment, the input data 1810 preferably includes a first set of input data related to attributes of a first heartbeat and a second set of input data related to a second heartbeat obtained at the same spatial location as the first heartbeat. The first set of input data and the second set of input data may include inputs In-1, In-2, In-3, . . . , In-n, where “n” is the last input of the plurality. For example, the first set of input data may include inputs In-1 and In-2, and the second set of input data may include In-3 and In-n. The first set of input data and the second set of input data may be input to the CNN during the same cycle or during separate cycles. The input data 1810 preferably includes attributes of the heartbeat, including, but not limited to, the following: - intracardiac EGM signals received by an electrode or by a bipolar electrode pair of a catheter such as a mapping catheter, the EMG signals may be bipolar or unipolar EGMs of electrodes of a mapping catheter around a reference annotation in the EP map; - ECG signals received from mapping electrodes around the reference annotation; - ECG signals received by body surface electrode(s) around the reference annotation; - TPI of the mapping electrode at the time of reference annotation, - the force value detected by the force sensor of the mapping catheter at the time of the reference annotation (typically measured in grams); - the spatial position or distance vector of the mapping electrodes at the reference annotation, which may be with or without respiration correction; - respiratory state vectors around the reference annotation, the position of the fiducial annotation within the respiratory cycle when lung motion is being monitored; - the difference between the current respiratory cycle length and the previous cycle length, - the ratio of the current respiratory cycle length divided by the previous cycle length, the difference between the current respiratory cycle length and the average or median respiratory cycle length; - the ratio of the current respiratory cycle length divided by the average or mean respiratory cycle length; - An indication of whether the physician manually accepted or deleted beats from the EP mapping system; and - The distance of the mapping electrode of the current reference annotation from the spatial position of the same electrode of the previous reference annotation, which can be measured either by its actual position or by its respiration-corrected position. In other words, the distance the electrode has traveled since the last heartbeat, e.g., by its actual position or by its respiration-corrected position, as an indication of the stability of the catheter.

[0160] Input data 1810 is provided to a first hidden layer 1820 including nodes HL1-1, HL1-2, HL1-3, . . . , HL1-n, and optionally to a second hidden layer 1830 including nodes HL2-1, HL2-2, HL2-3, . . . , HL2-n, which combine to generate output 1840. Hidden layers 1820, 1830 use data from the inputs to determine whether the first or second heartbeat has better characteristics, such as, but not limited to, less noise, more pronounced LAT activation, more pronounced late potentials, more stable disruption, less V-interference, etc. Over time, the neural network learns weights to apply to the input data.

[0161] In some embodiments, the output 1840 is a value indicating whether the first or second heartbeat of a pair of heartbeats taken at the same spatial location is better, for example. For example, in the CNN 1800, all inputs 1810 are fed simultaneously to the neural network to calculate the output 1840.

[0162] FIG. 19 is an exemplary flow diagram illustrating a method for training a neural network 1900 to learn to take attributes of two heartbeats and identify which heartbeat is better, according to the present application.

[0163] At 1910, the CNN 1800 receives from a database a pair of deleted and undeleted beats, preferably obtained at the same spatial location. The received data is input data 1810 for the CNN 1800 and preferably includes a set of attribute data corresponding to each beat of the beat pair obtained at the same spatial location. For example, the set of attribute data can be any of the input data 1810 described above for each beat.

[0164] At 1920, the hidden layers 1820, 1830 of the CNN 1800 preferably compare a set of attribute information between the deleted and non-deleted heartbeats to determine whether the first or second heartbeat has better features based on predetermined criteria, such as, but not limited to, less noise, more obvious LAT activation, more obvious late potentials, more stable fragmentation, heartbeats with less V-interference, or manual deletion by a physician. For example, if the first heartbeat was manually deleted by a physician and the second heartbeat was manually acquired by a physician, the CNN 1800 considers the second heartbeat to have better features than the first heartbeat. The CNN 1800 compares the attribute information associated with each heartbeat to learn optimal heartbeat features.

[0165] At 1930, the CNN 1800 outputs a result indicating whether the non-canceled beats have better features than the deleted beats.

[0166] At 1940, the CNN 1800 preferably repeats steps 1910, 1920, and 1930 for additional pairs of deleted and uncanceled beats obtained at the same spatial location until the CNN 1800 is trained.

[0167] In one embodiment, once the CNN 1800 is trained, it can be used to replace a suboptimal heartbeat in an EP mapping system with a heartbeat at the same spatial location that has better characteristics, either in real time or after a procedure. For example, the EP mapping system initially acquires a first heartbeat at spatial location X, Y, and Z. If another heartbeat is recorded at the same spatial location, the system runs the CNN using attribute data from each heartbeat as input data and outputs a decision on which heartbeat has better characteristics. If the first heartbeat is determined to be better, the second heartbeat is discarded from the EP mapping system. If the second heartbeat is determined to be better, the first heartbeat is automatically replaced with the second heartbeat in the EP mapping system.

[0168] The training model described above is preferably supervised by a physician, on-site at a hospital, or at a remote training facility.

[0169] An advantage of utilizing the CNN 1800 to receive attribute data of two heartbeats and learn to identify which heartbeat is better, according to the present application, is that the CNN 1800 can learn to mimic a physician's decision by looking at heartbeats that have been removed by the physician. Additionally, by determining the best heartbeat at a spatial location, the physician can more accurately determine the heartbeat's characteristics (e.g., LAT value, peak-to-peak bipolar voltage value, ECG signal, and EGM signal).

[0170] In the embodiment described above, the machine learning took two heartbeats and determined which was better. The system was trained by a pair of heartbeats (one deleted and one re-acquired at approximately the same spatial location on the heart). Below, we describe another embodiment with the same objective. In this embodiment, each heartbeat is assigned a score. For example, in one embodiment, the first and second heartbeats may each be assigned a score of 0, 0.5, or 1. If the doctor deletes a heartbeat, the heartbeat receives a score of 0; if the doctor acquires a new heartbeat at approximately the same location, the heartbeat receives a score of 1; and all other heartbeats receive a score of 0.5. Unlike the embodiment described above, which takes two heartbeats and predicts which is better, this embodiment takes a set of attributes for a single heartbeat and predicts the score for the single heartbeat. If the heartbeat has characteristics of a heartbeat that doctors typically delete, the machine learning will likely output a number around 0. If the heartbeat has the characteristics that a physician would typically reacquire after removing the inappropriate ones, the machine learning will likely output a number around 1. In this embodiment, performance is a continuous variable (not a discrete category). We train the machine learning with performances of only 0, 0.5, and 1, but the machine learning can output any real number. In this embodiment, during run-time (i.e., during the estimation phase, while the physician is performing an electrophysiological study on the patient), if the system needs to understand which heartbeat is better, it can predict the performance of each heartbeat independently and consider the heartbeat with the highest performance to be the best heartbeat. [Example]

[0171] In one exemplary embodiment, a machine learning algorithm, such as a neural network, can be trained to receive EP data representing multiple heartbeats taken at the same spatial location and select and output the best mapping annotation for the EP mapping system based on any of the inputs described above.

[0172] In conventional rule-based EP mapping systems, the mapping annotation is typically determined based on the highest peak voltage 2201 of the ECG signal around the reference annotation or the highest negative derivative 2202 of the ECG signal around the reference annotation, as shown in FIG. 22. A physician can accept the mapping annotation or move the mapping annotation if the results are not satisfactory. Examples of reasons why a physician may move the mapping annotation include, but are not limited to, heartbeats with significant noise, less obvious LAT, less obvious late potentials, less stable splitting, more V interference, etc.

[0173] In accordance with the subject matter of the present application, a neural network can receive attributes of multiple heartbeats at the same spatial location and train to identify the best mapping annotation based on the input data described above. In one embodiment, the heartbeats are obtained from stable arrhythmias.

[0174] In an embodiment of the subject matter of the present application, a neural network or machine learning system is first trained to find the same mapping annotations calculated by the rule-based system. The rule-based system uses multiple rules to select an action. The multiple rules can be defined by a human programmer. In some embodiments, the multiple rules are in the form of if-then statements. For example, initially, a large number of ECG signals and corresponding mapping annotations calculated by the rule-based system for each ECG signal are fed into the machine learning system to train it. This is expected to result in a machine learning system that is nearly as accurate as the rule-based system. When physicians begin using the system in various locations, the initial results are similar to the rule-based system. However, each time a physician manually corrects the mapping annotations, this information is used to train the machine learning system, which is preferably a self-learning system. In one embodiment, a single model can be maintained for all hospitals, or each hospital or group of hospitals can maintain its own model. After the machine learning system is trained, the model's performance can be validated against a gold standard to ensure its accuracy exceeds a predetermined threshold and / or is better than the accuracy of previous models. Over time, machine learning systems make machine learning algorithms better than rule-based systems.

[0175] Referring to FIG. 21 , in one exemplary embodiment, the neural network is an RNN 2100. The RNN 2100 preferably receives input data 2110. In one embodiment, the input data 2110 may include attribute data for each of multiple heartbeats 2110-1, 2110-2, 2110-3, ..., 2110-n obtained from the same spatial location. For example, the attribute data may be derived from one of the mapping electrodes 2070A-2070E shown in FIG. 20 . The attribute data for each heartbeat may include any of the attributes described herein. Each cycle of the RNN receives all attribute data for each heartbeat, as opposed to providing attribute data for a single heartbeat. Additionally, as shown with reference to arrow 2140 in FIG. 21 , if the accuracy of a newly trained model is above a threshold, or alternatively, if the accuracy of the newly trained model is higher than the accuracy of the previous model, the model may be used as input for the neural network.

[0176] Inputs 2110 are provided to the RNN 2100 and combined to generate output 2130, such as an identification of the best mapping annotation. The more heartbeat samples provided to the RNN 2100, the more accurate the output 2130 will be.

[0177] In one embodiment, to train the model, the physician's receipt of a heartbeat as a mapping annotation, along with other heartbeats obtained from the spatial location, may be input to the RNN 2100. The physician's received heartbeat will have better features than other heartbeats obtained at the same spatial location, and the RNN 2100 is expected to output the physician's received heartbeat as the best mapping annotation. The RNN 2100 compares the attribute information associated with each heartbeat with the physician's received heartbeats to learn optimal heartbeat features. Over time, the RNN 2100 learns the attribute features of the physician's received heartbeats to train the model.

[0178] The output of the RNN, such as output 2130, can be used to train the RNN 2100. For example, as shown with reference to arrow 2140 in FIG. 21, if the accuracy of output 2130 is above a threshold, output 2130 can be used as input 2110 of neural network 2100.

[0179] Additionally, for example, as discussed above, after each training, the trained model including its output can be run against a standard database, such as a gold standard database, to confirm its accuracy. In an exemplary embodiment, if the accuracy of the newly trained model is below a threshold, or alternatively, if the accuracy of the newly trained model is lower than the accuracy of the previous model, the model can be discarded. Similarly, if the accuracy of the newly trained model is equal to or higher than a threshold, or alternatively, if the accuracy of the newly trained model is higher than the accuracy of the previous model, the model can be released to an in-field EP mapping system.

[0180] In one embodiment, once the RNN 2100 is trained, it can be used to improve the mapping annotations of an EP mapping system in real time or post-procedure.

[0181] In one embodiment, training of the RNN 2100 can be supervised at the facility where the cardiac procedure is performed, such as a hospital or medical facility, or at a remote location, such as a training center. [Example]

[0182] In one exemplary embodiment, a machine learning algorithm, such as a neural network, can receive data representing heartbeats and any manual modifications to the mapping annotations, such as modifications performed by a physician, and learns to calculate mapping annotations for the EP mapping system.

[0183] In conventional mapping annotation systems, such as rule-based systems, mapping annotations are determined based on intracardiac ECG activation. ECG activation is traditionally determined in several ways, such as, but not limited to, the occurrence of bipolar EGMs, the time of maximum bipolar amplitude, or by combining various features of the ECG signal, such as, but not limited to, the manner described in U.S. Patent Publication No. 2018 / 00042504. A physician is required to review the determined mapping annotation for accuracy. If the mapping annotation is inaccurate, the physician must manually move the mapping annotation to correct it.

[0184] 23 is an exemplary flow diagram illustrating a method for training a machine learning system, such as a neural network, to compute mapping annotations of intracardiac signals, such as EGM signals, in accordance with the present application. In one embodiment, the neural network can be any type of neural network, such as a CNN or RNN, similar to those described with reference to FIGS. 18 and 21.

[0185] 23 illustrates a preferred system and method for training an initial model for a machine learning system, according to one embodiment. At 2310, the machine learning system preferably receives input data related to the EP signal. For example, the input data may include, but is not limited to, one or more of the following: - bipolar EGM signal of the mapping catheter around the fiducial annotation, - unipolar EGM signal of the mapping catheter around the fiducial annotation, - ECG signals received by body surface electrode(s) around the reference annotation; -Opening of the Window of Interest (WOI), -End of WOI, - spatial locations, such as 3D coordinates of mapping electrodes around the fiducial annotation, - TPI of the mapping electrodes around the fiducial annotation, - the force value detected by the force sensor of the mapping catheter at the time of the reference annotation (typically measured in grams); - respiratory state vectors around the reference annotation, the position of the fiducial annotation within the respiratory cycle when lung motion is being monitored; - the difference between the current respiratory cycle length and the previous cycle length, - the ratio of the current respiratory cycle length divided by the previous cycle length, the difference between the current respiratory cycle length and the average or median respiratory cycle length; - the ratio of the current respiratory cycle length divided by the average or mean respiratory cycle length; - An indication of whether the physician manually accepted or deleted beats from the EP mapping system; and - The distance of the mapping electrode of the current reference annotation from the spatial position of the same electrode of the previous reference annotation, which can be measured either by its actual position or by its respiration-corrected position. In other words, the distance the electrode has traveled since the last heartbeat, e.g., by its actual position or by its respiration-corrected position, as an indication of the stability of the catheter.

[0186] A rule-based algorithm is preferably applied to the input data at 2312, and mapping annotations are computed by the rule-based algorithm at 2314. The mapping annotations are output as an initial model for further training the machine learning system at 2316. The initial model is preferably trained to mimic the rule-based algorithm.

[0187] In some embodiments, the signal may be shifted to obtain a shifted signal. The shifted signal may be used as additional input data at 2310. For example, if bipolar and two unipolar signals are used as input data, the signal may be shifted 1 ms left, 2 ms left, 3 ms left, 4 ms left, 5 ms left, 6 ms left, 7 ms left, 8 ms left, 9 ms left, 10 ms left, 1 ms right, 2 ms right, 3 ms right, 4 ms right, 5 ms right, 6 ms right, 7 ms right, 8 ms right, 9 ms right, and 10 ms right. Thus, in this example, 21 different signals may be obtained. When the shifted signal is fed to a neural network, the neural network may learn that the signal is time-invariant, which allows training with less input data.

[0188] In some embodiments, the signal can be enhanced by adding low-frequency noise and / or high-frequency noise. The low-frequency noise can represent breathing effects, and the high-frequency noise can represent electromagnetic noise. For example, four signals can be obtained from a single signal: the signal itself, the signal plus 1 Hz noise, the signal plus 450 Hz noise, and the signal plus 1 Hz and 450 Hz noise. In this way, the convolutional layers of the CNN can learn to ignore low-frequency and high-frequency noise more quickly and efficiently.

[0189] In some embodiments, a combination of time shifting and noise addition may be used, for example, the signal may be shifted 1 ms to the left with 1 Hz noise added, 2 ms to the left with 1 Hz noise added, etc.

[0190] 23 illustrates an embodiment of continuous improvement training of a deployed model of a machine learning system, according to one embodiment. At 2320, the deployed model is deployed to an EP mapping system in an environment such as a hospital or medical testing facility. In one embodiment, the deployed model may be an initial model or a new model.

[0191] In some embodiments, the deployed model may be downloaded from a central server. The data downloaded from the central server may be signed with a private key, known only to the central server. A public key corresponding to the aforementioned private key may be known to the medical device downloading the deployed model from the central server. In some embodiments, a public / private key mechanism for HTTP or a similar protocol such as SFTP may also be used to ensure the security of the download. In these embodiments, the medical device may verify the validity of the central server's URL address and Secure Sockets Layer (SSL) certificate.

[0192] At 2322, the EP mapping system uses the deployed model to calculate mapping annotations in real time, for example, but not by way of limitation, if the deployed model is an initial model, a rule-based algorithm is used to calculate mapping annotations.

[0193] At 2324, if the mapping annotations are inaccurate, a physician can manually correct the mapping annotations in the EP mapping system. The machine learning system preferably receives data regarding mapping annotations manually corrected by a physician in the EP mapping system. The physician's corrections may be provided to the machine learning system in real time, through the cloud, or via an offline device such as a memory stick, CD, or digital versatile disc (DVD). In one embodiment, the physician's corrections may be manually confirmed before they are provided to the machine learning system. EGM signal A 2410 of FIG. 24 shows an example of mapping annotations corrected by a rule-based system. EGM signal B 2420 of FIG. 24 shows an example of mapping annotations manually corrected by a physician.

[0194] In some embodiments, the physician's modifications may be signed by the medical device before being provided to the machine learning system. The machine learning system may verify the physician's modifications to ensure they actually originate from the medical device. In some embodiments, a public / private key method may be used. For example, a private key may be known by the medical device and a public key may be known by the machine learning system. In further embodiments, the private key may be maintained in a tamper-resistant hardware security module of the medical device. Additionally or alternatively, the medical device's private key may be stored in a Trusted Platform Module integrated into the main board of the medical device. In further embodiments, the public / private key method may be combined with IP whitelisting (e.g., private key P1 originates from IP address xxx.x.xxx.x). In some embodiments, multiple medical devices contain the same private key. In other embodiments, a different private key may be deployed to each medical device. In some embodiments, for added security, the user may be required to provide a username and password while uploading data. The username and password may be combined with the private key method and / or IP whitelisting.

[0195] At 2326, the machine learning system further trains the developed model, preferably by using new EP signals and mapping annotations manually corrected by a physician in the EP mapping system.

[0196] At 2326, the machine learning system further trains the deployed model, preferably by using new EP signals and the physician-corrected mapping annotation(s). As discussed above, initially, the machine learning system calculates the mapping annotations based on the results of the rule-based algorithm. However, as the neural network receives more mapping annotations that have been manually corrected by the physician, the machine learning system learns from the corrections made by the physician and calculates the mapping annotations more accurately.

[0197] At 2328, the machine learning system preferably outputs the new model. In one embodiment, the output may further include at least one of a mapping annotation and a Boolean value indicating whether the mapping annotation can be computed. For example, the machine learning system may learn that it cannot compute the mapping annotation under some circumstances, such as insufficient force, an unstable catheter, etc. The Boolean output may be implemented as a neuron that outputs a real number. Optionally, a logit function may be applied to the output number, and the Boolean value is considered accurate when the output number exceeds a predetermined threshold and inaccurate when the output number is below the predetermined threshold. In one embodiment, a user may set the predetermined threshold, for example, by operating a slider bar in a graphical user interface.

[0198] At 2328, the new model including its output may optionally be run against a standard database, such as a gold standard database, to verify its accuracy. In an exemplary embodiment, if the accuracy of the newly trained model is below a threshold, or alternatively, if the accuracy of the newly trained model is lower than the accuracy of the previous model, the model may be discarded. Similarly, if the accuracy of the newly trained model is equal to or higher than a threshold, or alternatively, if the accuracy of the newly trained model is higher than the accuracy of the previous model, the model may be released to an EP mapping system in the field. Once the new model is validated, it may be deployed to the EP mapping system at 2320.

[0199] If the accuracy of the newly trained model is lower than the accuracy of the previous model but exceeds a threshold, the results of the newly trained model may be sent to a human expert, who may compare the predicted results of the gold standard database with the results calculated by the newly trained model. If the human expert believes the results of the newly trained model are better than the gold standard, the human expert may mark the results of the newly trained model as predicted, and the gold standard database may be updated accordingly.

[0200] In one embodiment, once the machine learning system is trained, it can be used to improve the mapping annotations of the EP mapping system in real time or post-procedure.

[0201] In one embodiment, training of the machine learning system can be supervised at the facility where the cardiac procedure is performed, such as a hospital or medical facility, or at a remote location, such as the cloud or a training center.

[0202] Although features and elements are described above in particular combinations, those skilled in the art will understand that each feature or element can be used alone or in combination with other features and elements. Additionally, although the process steps are described above in a particular order, the steps can be performed in any other desired order.

[0203] The methods, processes, modules, and systems described herein may be implemented in a computer program, software, or firmware embodied in a computer-readable medium for execution by a computer or processor. 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, ROM, RAM, registers, cache memory, semiconductor memory devices, magnetic media such as internal hard disks and removable disks, magneto-optical media, and optical media such as CD-ROM disks and DVDs. A processor in association with software may be used to implement a radio frequency transceiver for use in a WTRU, UE, terminal, base station, RNC, or any host computer.

[0204] Further embodiments of the present specification may be formed by adding to an embodiment one or more elements from any one or more other embodiments of the present specification and / or by substituting one or more elements from an embodiment with one or more elements from one or more other embodiments of the present specification.

[0205] It is understood, therefore, that the disclosed subject matter is not limited to the particular embodiments disclosed, but is intended to encompass all modifications that are within the spirit and scope of the present disclosure as defined by the appended claims, the above description, and / or as illustrated in the accompanying drawings.

[0206] [Embodiment] (1) A system for detecting heartbeats having optimal characteristics of an electrophysiological (EP) mapping system, said system comprising: 1. A processor, comprising: receiving a first heartbeat at an identified cardiac spatial location, the first heartbeat including first attribute information corresponding to the first heartbeat; receiving a second heartbeat at the identified heart space location, the second heartbeat including second attribute information corresponding to the second heartbeat; comparing the first attribute information with the second attribute information; and determining which of the first heartbeat and the second heartbeat has optimal characteristics based on the compared attribute information. (2) The system of embodiment 1, wherein the machine learning algorithm further outputs the determination of the heartbeat having optimal characteristics to the EP mapping system. (3) The system of embodiment 1, wherein at least one of the first heartbeat including first attribute information and the second heartbeat including second attribute information is stored in a database in communication with the processor. (4) The system of embodiment 3, wherein one of the first heartbeat and the second heartbeat is a heartbeat that is manually received in the EP mapping system, and the other of the first heartbeat and the second heartbeat is a heartbeat that is deleted by a physician from the EP mapping system at the same spatial location up to a threshold tolerance limit. (5) The system of embodiment 4, wherein the machine learning algorithm learns to determine which of the first heart beat and the second heart beat has optimal characteristics based on the results of the heart beat manually received in the EP mapping system and the heart beat deleted by the physician in the EP mapping system, and the manually received heart beat and the deleted heart beat are in the same position up to a threshold tolerance limit.

[0207] (6) The attribute information is intracardiac electrogram signals received by electrodes of the catheter; electrocardiogram signals received from the mapping electrodes; an electrocardiogram signal received by one or more body surface electrodes; a tissue proximity indication (TPI) of said mapping electrode at the time of reference annotation; a force value detected by a force sensor of the mapping catheter; the spatial position of the mapping electrode at the reference annotation; a respiratory state vector around the reference annotation; the position of said reference annotation within the respiratory cycle; the difference between the current respiratory cycle length and the previous respiratory cycle length; the ratio of the current respiratory cycle length to the previous respiratory cycle length; the difference between the current respiratory cycle length and an average or median respiratory cycle length; the ratio of the current respiratory cycle length to the average or median respiratory cycle length; an indication of whether a physician manually accepted or deleted a heartbeat from the EP mapping system; and The system of embodiment 1, further comprising at least one of: the distance of the mapping electrode at the current reference annotation from the spatial position of the same electrode at the previous reference annotation. (7) The system of embodiment 1, wherein the machine learning algorithm determines which of the first and second heartbeats has optimal characteristics based on at least one of the heartbeats having less noise, the heartbeats having a more pronounced local activation time (LAT), the heartbeats having a more pronounced late potential, the heartbeats having a more stable fraction, and the heartbeats having less V interference. (8) The system of embodiment 1, wherein the determination of whether the first heartbeat or the second heartbeat has optimal characteristics based on the compared attribute information is a binary decision. (9) The system of embodiment 1, wherein the machine learning algorithm is a neural network. (10) The system of embodiment 9, wherein the neural network is a convolutional neural network.

[0208] (11) A method for detecting heartbeats with optimal characteristics in an electrophysiological (EP) mapping system by a machine learning algorithm, the method comprising: receiving first data including a first heartbeat at an identified cardiac spatial location, the first data including first attribute information corresponding to the first heartbeat; receiving second data including a second heartbeat at the identified cardiac spatial location, the second data including second attribute information corresponding to the second heartbeat; comparing the first data with the second data; and outputting a determination of which of the first heartbeat and the second heartbeat has best characteristics based on the comparing. (12) The method of embodiment 11, further comprising outputting the determination to the EP mapping system. (13) A system for detecting mapping annotations in an electrophysiological (EP) mapping system, the system comprising: 1. A processor, comprising: receiving input data including attribute data for each of a plurality of heartbeats acquired at the same spatial location; comparing the attribute data for each of a plurality of heartbeats to a predefined threshold; and determining, based on the attribute data, which heartbeats to use as the mapping annotations. (14) The system of embodiment 13, wherein the machine learning algorithm further outputs the determination of the heartbeat to the EP mapping system for use as the mapping annotation. (15) The system of embodiment 13, wherein one of the plurality of heartbeats is manually acquired by a physician in the EP mapping system.

[0209] (16) The system of embodiment 15, wherein the machine learning algorithm learns to determine which of the multiple heartbeats should be used as the mapping annotation based on the results of the heartbeats obtained by the physician in the EP mapping system. (17) The attribute data is intracardiac electrogram signals received by electrodes of the catheter; electrocardiogram signals received from the mapping electrodes; an electrocardiogram signal received by one or more body surface electrodes; the local annotation time of the obtained heartbeat, a tissue proximity indication (TPI) of said mapping electrode at the time of reference annotation; a force value detected by a force sensor of the mapping catheter; the spatial position of the mapping electrode at the reference annotation; Respiratory gating state, a respiratory state vector around the reference annotation; the position of said reference annotation within the respiratory cycle; the difference between the current respiratory cycle length and the previous respiratory cycle length; the ratio of the current respiratory cycle length to the previous respiratory cycle length; the difference between the current respiratory cycle length and an average or median respiratory cycle length; the ratio of the current respiratory cycle length to the average or median respiratory cycle length; an indication of whether a physician manually accepted or deleted a heartbeat from the EP mapping system; an indication of whether any body surface activation was present during this beat; and The system of embodiment 13, further comprising at least one of: the distance of the mapping electrode at the current reference annotation from the spatial position of the same electrode at the previous reference annotation. (18) The system of embodiment 13, wherein the machine learning algorithm determines which of the plurality of heartbeats to use as the mapping annotation based on at least one of the heartbeats having less noise, the heartbeats having a more pronounced local activation time (LAT), the heartbeats having a more pronounced late potential, the heartbeats having a more stable fraction, and the heartbeats having less V-interference. (19) The system of embodiment 13, wherein the machine learning algorithm is a neural network. (20) The system described in embodiment 19, wherein the neural network is a recurrent neural network (RNN).

[0210] (21) The system of embodiment 20, wherein the heartbeats determined to be used as the mapping annotations are inputs for training the RNN. (22) The system of embodiment 13, wherein the determination of the heartbeat to use as the mapping annotation is compared against a gold standard database to confirm its accuracy. (23) A method for detecting mapping annotations in an electrophysiological (EP) mapping system by a machine learning algorithm, comprising: receiving input data including attribute data for each of a plurality of heartbeats acquired at the same spatial location; comparing the attribute data for each of a plurality of heartbeats to a predefined threshold; determining which heartbeats to use as the mapping annotations based on the comparing; and outputting the determination to the EP mapping system. (24) one of the plurality of heartbeats is manually acquired by a physician on the EP mapping system; 24. The method of claim 23, wherein the method further includes training the machine learning algorithm to learn to determine which of the multiple heartbeats should be used as the mapping annotation based on the results of the heartbeats obtained by the physician in the EP mapping system.

Claims

1. 1. An electrophysiological (EP) mapping system for detecting heart beats with optimal characteristics for improving cardiac mapping annotation in an EP mapping system, the EP mapping system comprising: Memory and A sensor, a processor communicatively coupled to the memory and the sensor; Including, The processor: receiving, from the sensor, a first heartbeat at an identified heart space location, the first heartbeat including first attribute information corresponding to the first heartbeat; receiving, from the sensor, a second heartbeat at the identified heart space location, the second heartbeat including second attribute information corresponding to the second heartbeat; comparing the first attribute information with the second attribute information; and determining which of the first heartbeat and the second heartbeat has optimal characteristics based on a comparison of the first attribute information and the second attribute information; The EP mapping system, wherein the machine learning algorithm determines which of the first heart beat and the second heart beat has optimal characteristics based on at least one of the heart beats having less noise, the heart beats having a more pronounced local activation time (LAT), the heart beats having a more pronounced late potential, the heart beats having a more stable fraction, and the heart beats having less V-interference, and the determined heart beat is used to annotate the cardiac mapping.

2. The EP mapping system of claim 1 , wherein the machine learning algorithm further outputs the determination of the heartbeat having optimal characteristics to the EP mapping system.

3. 2. The EP mapping system of claim 1, wherein at least one of the first heartbeat including first attribute information and the second heartbeat including second attribute information is stored in a database in communication with the processor.

4. 4. The EP mapping system of claim 3, wherein one of the first heart beat and the second heart beat is a heart beat manually received at the EP mapping system, and the other of the first heart beat and the second heart beat is a heart beat deleted by a physician from the EP mapping system at the same spatial location up to a threshold tolerance limit.

5. 5. The EP mapping system of claim 4, wherein the machine learning algorithm learns to determine which of the first heart beat and the second heart beat has optimal characteristics based on results of the heart beats manually received at the EP mapping system and the heart beats deleted by the physician at the EP mapping system, and the manually received heart beats and the deleted heart beats are co-located up to a threshold tolerance limit.

6. The first attribute information and the second attribute information are intracardiac electrogram signals received by electrodes of the catheter; electrocardiogram signals received from the mapping electrodes; an electrocardiogram signal received by one or more body surface electrodes; a tissue proximity indication (TPI) of the mapping electrode at the time of reference annotation; a force value detected by a force sensor of the mapping catheter; the spatial position of the mapping electrode at the reference annotation; a respiratory state vector around the reference annotation; the position of said reference annotation within the respiratory cycle; the difference between the current respiratory cycle length and the previous respiratory cycle length; the ratio of the current respiratory cycle length to the previous respiratory cycle length; the difference between the current respiratory cycle length and an average or median respiratory cycle length; the ratio of the current respiratory cycle length to the average or median respiratory cycle length; an indication of whether a physician manually accepted or deleted a heartbeat from the EP mapping system; and 3. The EP mapping system of claim 1, wherein the distance of the mapping electrode at the current reference annotation from the spatial position of the same electrode at a previous reference annotation.

7. An EP mapping system as described in claim 1, wherein the determination of whether the first heartbeat or the second heartbeat has optimal characteristics based on a comparison with the first attribute information and the second attribute information is a binary decision.

8. The EP mapping system of claim 1 , wherein the machine learning algorithm is a neural network.

9. The EP mapping system of claim 8 , wherein the neural network is a convolutional neural network.

10. 1. A method of operating an electrophysiological (EP) mapping system for detecting heart beats with optimal features for improving cardiac mapping annotation in an EP mapping system by a machine learning algorithm, the method comprising: a processor of the EP mapping system including the machine learning algorithm receiving first data from a sensor of the EP mapping system, the first data including a first heartbeat at an identified cardiac spatial location, the first data including first attribute information corresponding to the first heartbeat; receiving, by the processor, second data from the sensor, the second data including the second heartbeat at the identified cardiac spatial location, the second data including second attribute information corresponding to the second heartbeat; comparing the first data with the second data; and outputting a determination of which of the first heartbeat and the second heartbeat has optimal characteristics based on the comparing. A method of operating an EP mapping system, wherein the machine learning algorithm determines which of the first and second heart beats has optimal characteristics based on at least one of the heart beats having less noise, the heart beats having a more pronounced local activation time (LAT), the heart beats having a more pronounced late potential, the heart beats having a more stable fraction, and the heart beats having less V-interference, and the determined heart beat is used to annotate the cardiac mapping.

11. The method of claim 10 further comprising outputting the determination to the EP mapping system.

Images