System and method for determining the location of a catheter

Abstract

Systems, devices, and techniques for automatically generating a CPM matrix are disclosed. The system includes a processor that is configured to receive a plurality of historical sparse CPM matrices and a plurality of historical complementary CPM matrices, where each sparse CPM matrix is associated with a respective complementary CPM matrix; train a learning system based on the plurality of historical sparse CPM matrices and the plurality of historical complementary CPM matrices, where the learning system is trained to generate a complementary CPM matrix when a sparse CPM matrix is provided; receive a new sparse CPM matrix by the trained learning system; and generate a new complementary CPM matrix using the trained learning system.

Description

Technical Field

[0001] The present invention relates to artificial intelligence and machine learning associated with the optimization of cardiac mapping, including systems and methods for determining an improved mapping matrix (CPM matrix) between patient characteristics and catheter locations. Such a matrix can be used to determine catheter locations based on new received patient characteristics.

Background Art

[0002] Medical conditions such as arrhythmias (e.g., atrial fibrillation (AF)) are often diagnosed and treated via an in-body procedure. For example, pulmonary vein isolation (PVI) from the left atrial (LA) body is performed using ablation to treat AF. PVI, and many other minimally invasive catheter methods, require real-time visualization and mapping of the in-body surface.

[0003] Visualization and mapping of in-body body parts can be performed by mapping propagation of activation waves, fluoroscopy, computerized tomography (CT), and magnetic resonance imaging (MRI), and other techniques that may need to exceed the desired time or resources to provide visualization and mapping.

Summary of the Invention

Means for Solving the Problems

[0004] Systems, devices, and techniques are disclosed for generating an improved current-to-position (i.e., CPM) matrix. The system includes a processor that receives a plurality of historical sparse CPM matrices and a plurality of historical complementary CPM matrices, each sparse CPM matrix being associated with a respective complementary CPM matrix, trains a learning system based on the plurality of historical sparse CPM matrices and the plurality of historical complementary CPM matrices, where the learning system is trained to generate a complementary CPM matrix when given a sparse CPM matrix, receives a new sparse CPM matrix by the trained learning system, and generates a new complementary CPM matrix using the trained learning system. BRIEF DESCRIPTION OF THE DRAWINGS

[0005] A more detailed understanding can be obtained from the following description, taken in conjunction with the accompanying drawings, in which like reference numerals refer to like elements.

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Figure 12

Figure 13

Figure 14

Figure 15

Figure 16A

Figure 16B

Figure 16C

Figure 17

Figure 18a

Figure 18b

Figure 19

Figure 20

Figure 21

Figure 22

Figure 23

DETAILED DESCRIPTION OF THE INVENTION

[0006] According to an exemplary embodiment of the disclosed subject matter, a new current-to-position mapping (CPM) matrix can be generated. According to an embodiment of the present invention, the new CPM matrix may be generated from a sparse CPM matrix that is input into a trained learning system such as a neural network. As further disclosed herein, the CPM is generated based on electrical signals transmitted by catheter electrodes and received at a plurality of body patches on the surface of a patient's body. The electrical signals are correlated based on the known location of the catheter such that they are independently determined, for example, using a magnetic field (e.g., received and / or emitted by a magnetic catheter sensor, a localization pad, a body patch, etc.). However, since electrical signals and position data need to be acquired over a wide range of locations, generating a comprehensive CPM matrix is a time-consuming and laborious process. Therefore, a system and method are provided for using a learning system to obtain a supplemental CPM matrix from a given CPM matrix having only sparse data.

[0007] A "sparse" CPM matrix refers to a CPM matrix calculated from current data and position data, where it can be understood that the current data and position data are acquired at a first number of locations. Also, a "supplemental" CPM matrix is a CPM matrix calculated from current data and position data, where the current data and position data are acquired at a second number of locations, and it can also be understood that the second number of locations is greater than the first number of locations.

[0008] For example, if a "supplemental" CPM matrix provides a mapping between current data and position data at 1000 distinct points within the heart, the "sparse" CPM matrix may provide a mapping between these parameters at only 100 distinct points within the heart.

[0009] The object of the present invention is to estimate a supplementary CPM matrix based on a given sparse CPM matrix. A further object of the present invention is to determine the relationship between a given sparse CPM matrix and a given supplementary CPM matrix using a learning system.

[0010] A new CPM matrix can be generated such that new CPM data for a plurality of heart locations is generated based only on a subset of known electrical signals corresponding to the plurality of heart locations. Alternatively or additionally, the CPM matrix may be generated based on a correlation between a historical CPM and known catheter locations corresponding thereto. That is, new CPM data can be generated such that new CPM data for a plurality of heart locations is generated without any known catheter location corresponding to the plurality of catheter locations.

[0011] FIG. 1 is a block diagram of an exemplary system 100 for remotely monitoring and communicating a patient's biometric metrics (i.e., patient data). In the example illustrated in FIG. 1, system 100 includes a patient biometric measurement monitoring and processing device 102 associated with patient 104, a local computing device 106, a remote computing system 108, a first network 110, and a second network 120.

[0012] According to one embodiment, the monitoring and processing device 102 can be a device that is inside the patient's body (e.g., implantable subcutaneously). The monitoring and processing device 102 can be inserted into the patient via any applicable method including oral injection, surgical insertion via a vein or artery, endoscopic procedure, or laparoscopic procedure.

[0013] According to one embodiment, the monitoring and processing device 102 can be a device external to the patient. For example, as will be described in more detail below, the monitoring and processing device 102 can include an attachable patch (e.g., attached to the patient's skin). The monitoring and processing device 102 can also include a catheter, probe, blood pressure cuff, scale, bracelet or smartwatch biometric tracker, glucose monitor, continuous positive airway pressure (CPAP) machine, or substantially any device that can provide input regarding the patient's health or biometrics.

[0014] According to certain embodiments, the monitoring and processing device 102 can include both components internal to the patient and components external to the patient.

[0015] A single monitoring and processing device 102 is shown in FIG. 1. However, an exemplary system can include multiple patient biometric monitoring and processing devices. The patient biometric monitoring and processing devices can communicate with one or more other patient biometric monitoring and processing devices. Additionally or alternatively, the patient biometric monitoring and processing devices can communicate with a network 110.

[0016] One or more monitoring and processing devices 102 can acquire patient biometric data (e.g., electrical signals, blood pressure, body temperature, blood glucose levels, or other biometric data), and can receive at least a portion of the patient biometric data representing the acquired patient biometric metrics, as well as additional information associated with the patient biometric metrics acquired from one or more other monitoring and processing devices 102. The additional information can be, for example, diagnostic information and / or additional information obtained from additional devices such as wearable devices. Each monitoring and processing device 102 can process data including its own acquired patient biometric metrics, as well as data received from one or more other monitoring and processing devices 102.

[0017] In FIG. 1, network 110 is an example of a short - range network (e.g., a local area network (LAN) or a personal area network (PAN)). Information can be transmitted between the monitoring and processing device 102 and the local computing device 106 via the short - range network 110 using any one of various short - range wireless communication protocols such as Bluetooth, Wi - Fi, Zigbee, Z - Wave, near field communication (NFC), Ultra - wideband, Zigbee, or infrared (IR).

[0018] Network 120 can be a wired network, a wireless network, or may include one or more wired and wireless networks. For example, network 120 can be a long - range network (e.g., a wide area network (WAN), the Internet, or a cellular network). Information can be sent via network 120 using any one of various long - range wireless communication protocols (e.g., TCP / IP, HTTP, 3G, 4G / LTE, or 5G / New Radio).

[0019] The patient monitoring and processing device 102 can include a patient biometric sensor 112, a processor 114, a user input (UI) sensor 116, a memory 118, and a transceiver 122. The patient monitoring and processing device 102 can continuously or periodically monitor, store, process, and communicate the biometric metrics of any number of various patients via network 110. Examples of patient biometric metrics include electrical signals (e.g., ECG signals and brain biometric metrics), blood pressure data, blood glucose data, and body temperature data. The patient biometric metrics can be monitored and communicated for treating any number of various diseases such as cardiovascular diseases (e.g., arrhythmia, cardiomyopathy, and coronary artery disease), and autoimmune diseases (e.g., type I and type II diabetes).

[0020] The patient biometric measurement sensor 112 may include one or more sensors configured to sense, for example, the type of biometric indicators of the biometric measurement patient. For example, the patient biometric measurement sensor 112 may include electrodes configured to acquire electrical signals (e.g., cardiac signals, brain signals, or other bioelectrical signals), a body temperature sensor, a blood pressure sensor, a blood glucose sensor, a blood oxygen sensor, a pH sensor, an accelerometer, and a microphone.

[0021] As will be described in more detail below, the patient biometric measurement monitoring and processing device 102 may be an ECG monitor for monitoring the ECG signal of the heart. The patient biometric measurement sensor 112 of the ECG monitor may include one or more electrodes for acquiring the ECG signal. The ECG signal can be used for the treatment of various cardiovascular diseases.

[0022] In another example, the patient biometric measurement monitoring and processing device 102 may be a continuous glucose monitor (CGM) for continuously monitoring the blood glucose level of a patient on a continuous basis for the treatment of various diseases such as type I and type II diabetes. The CGM may include a subcutaneous electrode capable of monitoring the blood glucose level from the interstitial fluid of the patient. The CGM may be a component of a closed-loop system in which blood glucose data is sent to an insulin pump, for example, for the calculated delivery of insulin without user intervention.

[0023] 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.

[0024] The processor 114 may be configured to store patient data, such as patient biometric measurement data acquired by the patient biometric measurement sensor 112, in the memory 118 and communicate the patient data over the 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 will be described in more detail below.

[0025] According to one embodiment, the monitoring and processing device 102 may include a UI sensor 116 that includes, for example, a piezoelectric sensor or a capacitance sensor configured to receive user input such as a tap or a touch. For example, the UI sensor 116 may be controlled to implement capacitive coupling in response to the patient 104 tapping or touching the surface of the monitoring and processing device 102. Gesture recognition can be implemented by any one of various capacitive types, such as resistive-capacitive, surface capacitive, projected capacitive, surface acoustic wave, piezoelectric, and infrared touch. The capacitance sensor may be disposed over a small area or length of the surface such that a tap or touch on the surface activates the monitoring device.

[0026] As described in more detail below, the processor 114 may be configured to selectively respond to different tapping patterns (e.g., single tap or double tap) of the capacitance sensor, which may be the UI sensor 116, such 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 the processing device 102 when a gesture is detected.

[0027] The local computing device 106 of the system 100 communicates with the patient biometric monitoring and processing device 102 and may be configured to function as a gateway to the remote computing system 108 via the second network 120. The local computing device 106 can be, for example, a smartphone, smartwatch, tablet, or other portable smart device configured to communicate with other devices via the network 120. Alternatively, the local computing device 106 can be, for example, a desktop computer or laptop computer that uses an executable program for communicating information between the processing device 102 and the remote computing system 108 via a fixed base station including modem and / or router capabilities, a wireless module of a PC, or a fixed or stand-alone device such as a USB dongle. The patient's biometric metrics can be communicated between the local computing device 106 and the patient biometric monitoring and processing device 102 using a short-range wireless network 110 such as a local area network (LAN) (e.g., a personal area network (PAN)) and a short-range wireless technology standard (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 associated with the acquired patient electrical signals, as described in more detail below.

[0028] In some embodiments, the remote computing system 108 may be configured to receive at least one of the biometric metrics of the monitored patient and the information associated with the monitored patient via the network 120, which is a long-distance network. For example, if the local computing device 106 is a mobile phone, the network 120 may be a wireless cellular network, and the 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 will be described in more detail below, the remote computing system 108 may be configured to provide (e.g., visually display and / or auditorily provide) at least one of the patient's biometric metrics and the associated information to a medical professional (e.g., a physician).

[0029] Figure 2 is a system diagram of an example of a computing environment 200 that communicates with the network 120. In some examples, the 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.

[0030] As shown in Figure 2, the computing environment 200 includes a remote computing system 108 (hereinafter, the computer system), which is an example of a computing system in which the embodiments described herein may be implemented.

[0031] The remote computing system 108 can perform various functions via a processor 220, which can include one or more processors. The functions can include analyzing the monitored patient's biometric metrics and associated information, and providing warnings, additional information, or instructions (e.g., via display 266) according to thresholds and parameters determined by a physician or algorithm-driven. As described in more detail below, the remote computing system 108 can be used to provide a patient information dashboard (e.g., via display 266) to healthcare providers (e.g., physicians), and the patient information can enable the healthcare provider to identify and prioritize patients who have more critical needs than other patients.

[0032] As shown in FIG. 2, the computer system 210 may include a communication mechanism such as a bus 221 or any other communication mechanism for communicating information within the computer system 210. The computer system 210 further includes one or more processors 220 coupled to the bus 221 for processing information. The processor 220 may include one or more CPUs, GPUs, or any other processor known in the art.

[0033] Computer system 210 also includes a system memory 230 coupled to bus 221 for storing information and instructions to be executed by processor 220. System memory 230 may include computer-readable storage media in the form of volatile and / or non-volatile memory such as read only system memory (ROM) 231 and / or random access memory (RAM) 232. System memory RAM 232 may include other dynamic storage devices (e.g., dynamic RAM, static RAM, and synchronous DRAM). System memory ROM 231 may include other static storage devices (e.g., programmable ROM, erasable PROM, and electrically erasable PROM). Additionally, system memory 230 can be used to store temporary variables or other intermediate information during execution of instructions by processor 220. Basic input / output system 233 (BIOS) may include routines that transfer information stored in system memory ROM 231 among elements within computer system 210, such as at startup. RAM 232 can be immediately accessible to processor 220 and / or may include data and / or program modules currently being operated on by processor 220. System memory 230 may further include, for example, operating system 234, application program 235, other program modules 236, and program data 237.

[0034] The illustrated computer system 210 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). The storage devices may be added to the computer system 210 using an appropriate device interface (e.g., a small computer system interface (SCSI), integrated device electronics (IDE), Universal Serial Bus (USB), or FireWire).

[0035] The computer system 210 may also include a display controller 265 coupled to bus 221 for controlling a monitor or display 266, such as a cathode ray tube (CRT) or a liquid crystal display (LCD), for displaying information to a computer user. The illustrated computer system 210 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, a trackball, or a pointing stick for communicating indication information and command selections to the processor 220 and controlling cursor movement on the display 266. The display 266 may provide a touch screen interface that enables input to complement or replace the communication of indication information and command selections by the pointing device 261 and / or the keyboard 262.

[0036] Computer system 210 may perform some or each of the functions and methods described herein in response to a processor 220 executing one or more sequences of one or more instructions included in a memory such as system memory 230. Such instructions can be read into system memory 230 from another computer-readable medium such as hard disk 241 or removable media drive 242. Hard disk 241 may include one or more data stores and data files used by the embodiments described herein. The data store contents and data files may be encrypted to improve security. Processor 220 may also be employed in a multiprocessing configuration to execute one or more sequences of instructions included in system memory 230. In alternative embodiments, hardwired circuitry may be used in place of, or in combination with, software instructions. Accordingly, embodiments are not limited to any specific combination of hardware circuitry and software.

[0037] As described above, computer system 210 may include at least one computer-readable medium or memory for holding instructions programmed 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 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 discs, solid state drives, magnetic disks, and magneto-optical disks, such as hard disk 241 or removable media drive 242. Non-limiting examples of volatile media include dynamic memory, such as system memory 230. Non-limiting examples of transmission media include coaxial cables, copper wire, and fiber optics, including the wires that make up bus 221. Transmission media can also take the form of acoustic or light waves, such as those generated during radio and infrared data communications.

[0038] Computing environment 200 may further include a computer system 210 that operates in a networked environment using logical connections to a local computing device 106 and to one or more other devices, such as a personal computer (laptop or desktop), a mobile device (e.g., a patient mobile device), a server, a router, a network PC, a peer device, or other common network node, and typically includes many or all of the elements described above with respect to computer system 210. When used in a network environment, computer system 210 may include a modem 272 for establishing communications via a network 120, such as the Internet. Modem 272 may be connected to system bus 221 via network interface 270 or via some other suitable mechanism.

[0039] Network 120, as shown in FIGS. 1 and 2, can be any network or system generally known in the art, including the Internet, an intranet, a local area network (LAN), a wide area network (WAN), a metropolitan area network (MAN), a direct connection or a series of connections, a cellular phone network, or any other network or medium that can facilitate communication between computer system 610 and other computers (e.g., local computing device 106).

[0040] FIG. 3 is a block diagram of an exemplary device 300 that can implement one or more features of the present disclosure. Device 300 can be, for example, local computing device 106. Device 300 can include, for example, a computer, a gaming device, a handheld device, a set-top box, a television, a cellular phone, or a tablet computer. 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. Device 300 can also optionally include an input driver 312 and an output driver 314. It is understood that device 300 can include additional components (not shown in FIG. 3) including an artificial intelligence accelerator.

[0041] In various alternatives, processor 302 includes a central processing unit (CPU), a graphics processing unit (GPU), a CPU and GPU located on the same die, or one or more processor cores, each of which can be a CPU or a GPU. In various alternatives, memory 304 is located on the same die as processor 302 or separately from processor 302. Memory 304 includes volatile or non-volatile memory, such as random access memory (RAM), dynamic RAM, or cache.

[0042] The memory device 306 includes fixed or removable memory means, such as a hard disk drive, a solid state drive, an optical disk, or a flash drive. The input device 308 includes, but is not limited to, a keyboard, a keypad, a touch screen, a touch pad, a detector, a microphone, an accelerometer, a gyroscope, a biometrics scanner, or a network connection (e.g., a wireless local area network card for transmitting and / or receiving wireless IEEE802 signals). The output device 310 includes, but is not limited to, a display, a speaker, a printer, a tactile feedback device, one or more lights, an antenna, or a network connection (e.g., a wireless local area network card for transmitting and / or receiving wireless IEEE802 signals).

[0043] The input driver 312 communicates with the processor 302 and the input device 308, enabling the processor 302 to receive inputs from the input device 308. The output driver 314 communicates with the processor 302 and the output device 310, enabling the processor 302 to send outputs to the output device 310. Note that the input driver 312 and the output driver 314 are optional components, and if the input driver 312 and the output driver 314 do not exist, the device 300 operates in the same manner. The output driver 316 includes an accelerated processing device ("APD") 316 coupled to the display device 318. The APD receives calculation commands and graphic rendering commands from the processor 302, processes those calculation commands and graphic rendering commands, and provides pixel outputs to the display device 318 for display. As will be described in more detail below, the APD 316 includes one or more parallel processing units that execute calculations according to the single-instruction-multiple-data ("SIMD") paradigm. Thus, although various functions are described herein as being performed by the APD 316 or in cooperation with the APD 316, in various alternatives, the functions described as being performed by the APD 316 are additionally or alternatively not driven by the host processor (e.g., the processor 302) and are performed by other computing devices having similar capabilities to provide graphical outputs to the display device 318. For example, it is contemplated that any processing system that executes processing tasks according to the SIMD paradigm can perform the functions described herein. Alternatively, it is contemplated that computing systems that do not execute processing tasks according to the SIMD paradigm can perform the functions described herein.

[0044] Figure 4 illustrates a graphic depiction of an artificial intelligence system 200 incorporating the exemplary device of Figure 3. System 400 includes data 410, a machine 420, a model 430, a plurality of outcomes 440, and underlying hardware 450. System 400 operates by training machine 420 using data 410 while building model 430 that enables prediction of the plurality of outcomes 440. System 400 can operate with respect to hardware 450. In such a configuration, data 410 can be related to hardware 450 and can, for example, originate from device 102. For example, data 410 can be in-progress data or output data related to hardware 450. Machine 420 can operate as a controller or data collection associated with hardware 450 or can be associated therewith. Model 430 can be configured to model the operation of hardware 450 and the data 410 collected from hardware 450 to predict the outcomes achieved by hardware 450. Hardware 450 can be configured to provide a predetermined desired outcome 440 from hardware 450 using the predicted outcome 440.

[0045] Figure 5 illustrates a method 500 executed in the artificial intelligence system of Figure 4. Method 500 includes, at step 510, collecting data from hardware. This data can include currently collected data, historical data, or other data from the hardware. For example, this data can include measurements during a surgical procedure and can be associated with the outcome of the procedure. For example, the temperature of the heart can be collected and correlated with the outcome of a heart procedure.

[0046] Method 500 includes, at step 520, training a machine on the hardware. The training can include analysis and correlation of the data collected at step 510. For example, in the case of the heart, temperature and outcome data can be trained to determine whether there is a correlation or association between the temperature of the heart during the procedure and the outcome.

[0047] Method 500 includes, at step 530, building a model based on data related to hardware. Building the model may include physical hardware or software modeling, algorithm modeling, etc., as described below. This modeling may aim to represent the collected and trained data.

[0048] Method 500 includes, at step 540, predicting the outcome of a model related to hardware. This prediction of the outcome may be based on the trained model. For example, in the case of the heart, if a positive outcome from the treatment is obtained when the temperature during the treatment is between 97.7 and 100.2, then the outcome in a given treatment can be predicted based on the temperature of the heart during the treatment. This model is a preliminary one but is provided for illustrative purposes to deepen the understanding of the present invention.

[0049] The present system and method operate to train a machine, build a model, and use an algorithm to predict an outcome. These algorithms can be used to solve the trained model and predict the outcome related to hardware. These algorithms can generally be classified into classification algorithms, regression algorithms, and clustering algorithms.

[0050] For example, a classification algorithm is used in a situation 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. Therefore, a classification algorithm is used to predict the outcome from a predetermined number of fixed pre-defined outcomes. Classification algorithms may include naive Bayes algorithm, decision tree, random forest classifier, logistic regression, support vector machine, and k-nearest neighbor method.

[0051] Generally, the naive Bayes algorithm follows the Bayes theorem and a probabilistic approach. Other algorithms based on probability theory can also be used and generally operate using the same principles of probability theory as those described below for the exemplary naive Bayes algorithm.

[0052] Figure 6 illustrates an example of the probability of simple Bayes calculation. The probabilistic approach of Bayes' theorem 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 simple Bayes algorithm can update the prior probabilities to form posterior probabilities. This is given by the following formula.

[0053]

Number

[0054] This simple Bayes algorithm and the Bayes algorithm can generally be useful when it is necessary to predict whether an input belongs to a given list of n classes. Since the probabilities of all n classes are very low, a probabilistic approach can be used.

[0055] For example, as illustrated in Figure 6, a person who plays golf depends on factors including weather other than that shown in the first data set 610. The first data set 610 illustrates the weather in the first column and the result of the play related to that weather in the second column. In the frequency table 620, the frequency of occurrence of a particular event is generated. In the frequency table 620, the frequency with which a person plays golf or does not play golf under each weather condition is determined. 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.

[0056] Posterior probabilities can be generated from the likelihood table 630. These posterior probabilities can be configured to answer questions about the weather conditions and whether golf is played under those weather conditions. For example, the probability that it is sunny outside and golf is played can be expressed by the following Bayes formula. P(Yes|Sunny) = P(Sunny|Yes) × P(Yes) / P(Sunny)

[0057] According to the likelihood table 630, P(sunny|yes)=3 / 9 = 0.33, P(sunny)=5 / 14 = 0.36, P(yes)=9 / 14 = 0.64 are as follows.

[0058] Therefore, P(yes|sunny)=0.33×0.64 / 0.36, that is, approximately 0.60 (60%).

[0059] Generally, a decision tree is a tree structure similar to a flowchart. Each external node represents a test of an attribute, and each branch represents the result of that test. The leaf nodes contain the actual predicted labels. The decision tree starts from the root of the tree and the attribute values are compared until a leaf node is reached. A decision tree can be used as a classifier when dealing with high-dimensional data and when little time is spent on data preparation. A decision tree can take the form of a simple decision tree, a linear decision tree, an algebraic decision tree, a deterministic decision tree, a randomized decision tree, a non-deterministic decision tree, and a quantum decision tree. An exemplary decision tree is provided below in FIG. 7.

[0060] FIG. 7 illustrates a decision tree along the same structure as the above Bayesian example when determining whether to play golf. In the decision tree, the first node 710 examines the cases where the weather is sunny 712, cloudy 714, and rainy 716 as options to proceed down the decision tree. If the weather is sunny, the branch of the tree continues to the second node 720 that examines the temperature. The temperature at node 720 can be high 722 or normal 724 in this example. If the temperature at node 720 is high 722, a prediction result of golf "no" 723 occurs. If the temperature at node 720 is normal 724, a prediction result of golf "yes" 725 occurs.

[0061] Furthermore, from the first node 710, a result of cloudy 714, golf "yes" 715 occurs.

[0062] From the weather 710 of the first node, as a result of rain 716, the third node 730 again checks the temperature. If the temperature at the third node 730 is normal 732, the answer is "yes" 733 to play golf. If the temperature at the third node 730 is low 734, the answer is "no" 735 not to play golf.

[0063] From this decision tree, a certain golfer will play golf when it is cloudy 715, when it is sunny 725 with normal temperature, and when it is rainy 733 with normal temperature. However, this golfer will not play when it is sunny and hot 723 or when it is rainy and cold 735.

[0064] A random forest classifier is a committee of decision trees. Each decision tree is given a subset of the attributes of the data and makes a prediction based on that subset. The mode of the actual prediction values of the decision trees is considered, and the final random forest answer is provided. A random forest classifier generally becomes a more robust and accurate classifier by reducing overfitting that exists in stand-alone decision trees.

[0065] FIG. 8 illustrates an exemplary random forest classifier for classifying the color of clothing. As shown in FIG. 8, the random forest classifier includes five decision trees 8101, 8102, 8103, 8104, and 8105 (collectively or generally referred to as decision tree 810). Each of the trees is designed to classify the color of clothing. Since each individual tree operates generally like the decision tree of FIG. 7, consideration of each of the trees and the decisions made is not provided. In this figure, three of the five trees (8101, 8102, 8104) determine that the clothing is blue, one determines that the clothing is green (8103), and the remaining tree determines that the clothing is red (8105). The random forest receives these actual prediction values of the five trees, calculates the mode of the actual prediction values, and provides the random forest answer that the clothing is blue.

[0066] Logistic regression is another algorithm for binary classification tasks. Logistic regression is based on the logistic function, also known as the sigmoid function. This S-shaped curve takes any real value and maps it between 0 and 1, gradually approaching these limits. The logistic model can be used to model the probability that a particular class or event exists, such as pass / fail, win / lose, survival / death, or healthy / sick. This can be extended to model events in several classes, such as determining whether an image contains a cat, dog, lion, etc. Each object detected in the image is assigned a probability between 0 and 1, and the sum of these probabilities is 1.

[0067] In the logistic model, the log-odds (logarithm of the odds) of the value labeled "1" is a linear combination of one or more independent variables ("predictors"), where each independent variable can be either a binary variable (two classes coded as indicator variables) or a continuous variable (any real value). Since the corresponding probability of the value labeled "1" can vary between 0 (certainly a value of "0") and 1 (certainly a value of "1"), it is labeled as such, and since the logistic function is the function that converts log-odds to probability, it is so named. The unit of measurement on the log-odds scale is called the logit, which is another name for the logistic unit. Similar models that use a different sigmoid function instead of the logistic function, such as the probit model, can also be used. A characteristic of the logistic model is that increasing one of the independent variables multiplicatively scales the odds of a given outcome by a certain ratio, and each independent variable has its own parameter. In the case of a binary dependent variable, this generalizes the odds ratio.

[0068] In a binary logistic regression model, the dependent variable has two levels (categories). An output with three or more values is modeled by multinomial logistic regression, and if multiple categories are ordered, it is modeled by ordinal logistic regression (e.g., the proportional odds ordinal logistic model). The logistic regression model itself simply models the probability of the output with respect to the input and does not perform statistical classification (it is not a classifier), but it can be used to create a classifier, for example, by selecting a cutoff value and classifying inputs with a probability greater than the cutoff as one class and inputs with a probability less than the cutoff as another class. This is a common way to create a binary classifier.

[0069] Figure 9 illustrates an exemplary logistic regression. This exemplary logistic regression enables prediction of results based on a set of variables. For example, based on an individual's grade point average, the result of being accepted into school can be predicted. The prediction can be made based on the past history of the grade point average and its relationship to passing. The logistic regression in Figure 9 enables the analysis of the grade point average variable 920 to predict the result 910 defined by 0 to 1. At the lower end 930 of the sigmoid curve, the grade point average 920 predicts the result 910 of not being accepted. At the upper end 940 of the sigmoid curve, the grade point average 920 predicts the result 910 of being accepted. Logistic regression can be used to predict housing values, the customer lifetime value in the insurance sector, and so on.

[0070] Using a support vector machine (SVM), data can be rearranged by separating the margin between two classes as far as possible. This is called margin maximization separation. Unlike linear regression, which uses the entire dataset for that purpose, the SVM can consider support vectors while plotting the hyperplane.

[0071] Figure 10 illustrates an exemplary support vector machine. In an exemplary SVM 1000, data can be classified into two different 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 (illustrated by line 1040) between the hyperplane 1030 and the closest data point 1050 from each class. The data point 1050 closest to the hyperplane 1030 is known as a support vector. The hyperplane 1030 is drawn based on these support vectors 1050, and the optimal hyperplane has the maximum distance from each of the support vectors 1050. The distance between the hyperplane 1030 and the support vectors 1050 is known as the margin.

[0072] The SVM 1000 can be used for data classification by using the hyperplane 1030 such that the distance between the hyperplane 1030 and the support vectors 1050 is maximized. Such an SVM 1000 can be used, for example, to predict heart disease.

[0073] The k-nearest neighbors (KNN) refers to a series of algorithms that generally do not assume a basic data distribution and perform a reasonably short training phase. Generally, KNN uses a large number of data points divided into multiple classes to predict the classification of a new sample point. In operation, KNN specifies an integer N for a new sample. The N entries in the system's model 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. KNN generally requires an increase in memory space as the training set increases. This also means that the estimation time increases proportionally to the number of training points.

[0074] In the regression algorithm, since the output is a continuous quantity, the regression algorithm 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 the true quality (such as housing cost, number of calls, all transactions, etc.) considering consistent variables. The connection between the variable and the result is obtained by fitting the optimal line (hence the name linear regression). This optimal line is known as the regression line and is directly represented by the condition Y = a × X + b. Linear regression is best used in approaches with a small number of dimensions.

[0075] Figure 11 illustrates an exemplary linear regression model. In this model, the predictor variable 1110 is modeled against the measured variable 1120. The clusters of instances of the predictor variable 1110 and the measured variable 1120 are plotted as data points 1130. Then, the data points 1130 are fitted to the optimal line 1140. Then, the optimal line 1140 is used for subsequent predictions. When the measured variable 1120 is given, in that example, the line 1140 is used to predict the predictor variable 1110. Linear regression can be used to model and predict financial portfolios, salary predictions, real estate, and the scheduled arrival times of transportation.

[0076] Clustering algorithms can be used for modeling and training a dataset. In clustering, the input is assigned to two or more clusters based on the similarity of features. Clustering algorithms usually learn patterns and useful insights from data without guidance. For example, unsupervised learning algorithms such as K-means clustering can be used to cluster viewers into similar groups based on interests, age, geography, etc.

[0077] K-means clustering is generally regarded as a simple unsupervised learning approach. In K-means clustering, similar data points can be clustered together and bound in the form of clusters. One way to bind data points together is by calculating the centroid of the group of data points. When determining effective clusters, in K-means clustering, the distance from the centroid of the cluster between each point is evaluated. Depending on the distance between the data point and the centroid, the data is assigned to the nearest cluster. The purpose of clustering is to determine the inherent grouping of a series of unlabeled data. "K" in K-means represents the number of clusters formed. The number of clusters (basically, the number of classes into which new instances of the data can be classified) can be determined by the user. This determination can be made, for example, by using feedback and looking at the size of the clusters during training.

[0078] K-means is mainly used when the dataset has distinct and well-separated points; otherwise, if the clusters are not separated, the modeling may render the clusters inaccurately. Also, K-means can be avoided when the dataset contains a large number of outliers or when the dataset is non-linear.

[0079] FIG. 12 illustrates K-means clustering. In K-means clustering, data points are plotted and a K value is assigned. For example, for K = 2 in FIG. 12, the data points are plotted as shown in depiction 1210. Then, in step 1220, points are assigned to like centers. The centroid of the cluster is identified as shown in 1230. Once the centroid is identified, the points are re-assigned to the clusters such that the distance between the data points and the centroid of each cluster is minimized, as illustrated in 1240. Then, a new centroid of the cluster can be determined as illustrated in depiction 1250. When the data points are re-assigned to the clusters, new cluster centroids are formed and an iteration or series of iterations occurs to minimize the size of the clusters and determine the centroid of the optimal centroid. Then, when new data points are measured, the new data points are compared to the centroids and clusters and can be identified within that cluster.

[0080] Ensemble learning algorithms may be used. These algorithms use multiple learning algorithms to achieve better prediction performance than can be obtained from any one of the constituent learning algorithms alone. Ensemble learning algorithms perform the task of searching the hypothesis space to find a suitable hypothesis for making good predictions about a particular problem. Even if the hypothesis space contains hypotheses that are very suitable for a particular problem, it may be very difficult to find the appropriate hypothesis. Ensemble algorithms combine multiple hypotheses to form a better hypothesis. The term ensemble is usually used in the context of a method of generating multiple hypotheses using the same basic learner. The broader concept of a multiple classifier system also encompasses the hybridization of hypotheses not induced by the same basic learner.

[0081] To evaluate the prediction of an ensemble, usually more calculations are required than to evaluate the prediction of a single model. Therefore, an ensemble can be considered as a way to complement weak learning algorithms by performing a lot of extra calculations. Fast algorithms such as decision trees are generally used in ensemble methods such as random forests, but even slower algorithms can benefit from ensemble methods.

[0082] Since an ensemble can be used to make predictions after training, it is itself a supervised learning algorithm. Therefore, a trained ensemble represents a single hypothesis. However, this hypothesis is not necessarily contained within the hypothesis space of the models it was constructed from. Thus, an ensemble can be shown to have more flexibility in the functions it can represent. This flexibility theoretically allows it to overfit the training data more than a single model, but in practice, some ensemble methods (especially bagging) tend to reduce problems related to overfitting of the training data.

[0083] Empirically, when there is significant diversity among models, ensemble algorithms tend to produce better results. Therefore, many ensemble methods aim to promote diversity among the models they combine. Counterintuitively, more random algorithms (such as random decision trees) can be used to generate stronger ensembles than very careful algorithms (such as entropy-reducing decision trees). However, it has been shown that using various powerful learning algorithms is more effective than using techniques that try to simplify the models to promote diversity.

[0084] The number of component classifiers in an ensemble has a significant impact on the accuracy of prediction. Determining a priori the size of the ensemble and the amount and speed of the big data stream makes this even more important for online ensemble classifiers. The theoretical framework suggests that there is an ideal number of component classifiers in the ensemble and that accuracy decreases if the number of classifiers is more or less than this number. The theoretical framework indicates that the highest accuracy can be obtained by using the same number of independent component classifiers as the number of class labels.

[0085] Some common types of ensembles include Bayesian optimal classifiers, bootstrap aggregation (bagging), boosting, Bayesian model averaging, combinations of Bayesian models, model buckets, and stacking. FIG. 13 illustrates an exemplary ensemble learning algorithm in which bagging is executed in parallel (1310) and boosting is executed sequentially (1320).

[0086] A neural network is a network or circuit of neurons, or in the modern sense, an artificial neural network composed of artificial neurons or nodes. The connections of biological neurons are modeled as weights. Positive weights reflect excitatory connections and negative values mean inhibitory connections. The inputs are modified by the weights and summed using a linear combination. An activation function can control the amplitude of the output. For example, the allowable range of the output is usually 0 to 1, but it can also be -1 to 1.

[0087] These artificial networks can be used for predictive modeling, adaptive control, and applications and can be trained via a dataset. Self-learning arising from experience may occur within the network, and conclusions can be drawn from complex and seemingly unrelated sets of information.

[0088] To be complete, a biological neural network consists of chemically connected neurons or groups of functionally associated neurons. One neuron can be connected to many other neurons, and the total number of neurons and connections within the network can vary widely. Connections, called synapses, are usually formed from axons to dendrites, but dendrodendritic synapses and other connections are also possible. Apart from electrical signaling, there are other forms of signaling that result from the diffusion of neurotransmitters.

[0089] Artificial intelligence, cognitive modeling, and neural networks are information - processing paradigms inspired by the data - processing methods of biological neural systems. Artificial intelligence and cognitive modeling attempt to simulate some of the characteristics of biological neural networks. In the field of artificial intelligence, artificial neural networks have been successfully applied to speech recognition, image analysis, and adaptive control, and software agents (in computer and video games) or autonomous robots have been constructed.

[0090] A neural network (NN) is an interconnected group of natural or artificial neurons that, in the case of artificial neurons called artificial neural networks (ANN) or simulated neural networks (SNN), uses a mathematical or computational model based on a connectionist approach to computing for information processing. 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, a neural network is a non - linear statistical data - modeling tool or a decision - making tool. These can be used to model the complex relationships between inputs and outputs and to find patterns in data.

[0091] An artificial neural network includes a network of simple processing elements (artificial neurons) that can exhibit complex and global behavior determined by connections between the processing elements and element parameters.

[0092] One classical type of artificial neural network is the recurrent Hopfield network. The usefulness of artificial neural network models lies in the fact that they can be used to estimate functions from observations and then use them. Unsupervised neural networks can also be used for learning representations of inputs that capture salient features of the input distribution and, more recently, deep learning algorithms that can implicitly learn the distribution function of observational data. Learning in neural networks is particularly useful in applications where it is not practical to design such functions manually due to the complexity of the data or tasks.

[0093] Neural networks can be used in various fields. The tasks to which artificial neural networks are applied tend to fall into the following broad categories: regression analysis including function approximation, or time series prediction and modeling, pattern and sequence recognition, classification including novelty detection and sequential decision-making, filtering, clustering, blind signal separation, and data processing including compression.

[0094] The application areas of ANN include identification and control of nonlinear systems (vehicle control, process control), playing games and making decisions (backgammon, chess, racing), pattern recognition (radar systems, face recognition, object recognition), sequence recognition (gesture, speech, handwritten text recognition), medical diagnosis, financial applications, data mining (or knowledge discovery in database, "KDD"), visualization, and email spam filtering. For example, it is possible to create a semantic profile of a user's interests arising from photos trained for object recognition.

[0095] Figure 14 illustrates an exemplary neural network. The neural network has an input layer represented by a plurality of inputs such as 14101 and 14102. The inputs 14101, 14102 are provided to a hidden layer depicted as including nodes 14201, 14202, 14203, 14204. These nodes 14201, 14202, 14203, 14204 are combined to create an output 1430 in the output layer. The neural network performs simple processing through a hidden layer of nodes 14201, 14202, 14203, 14204 which are simple processing elements, and these nodes can exhibit complex and global behavior determined by the connections between the processing elements and the element parameters.

[0096] The neural network of FIG. 14 can be implemented in hardware. As depicted in FIG. 15, a hardware-based neural network is shown.

[0097] The treatment of heart diseases such as arrhythmia often requires obtaining a detailed mapping of the heart tissue, heart chambers, veins, arteries, and / or electrical pathways. For example, a prerequisite for performing catheter ablation without problems is to accurately locate the cause of the arrhythmia within the heart chamber. Such location can be performed by an electrophysiological investigation, during which the potential is spatially resolved and detected by a mapping catheter introduced into the heart chamber. Thus, this electrophysiological examination, so-called electroanatomical mapping, provides 3D mapping data which can be displayed on a monitor. In many cases, the mapping function and the treatment function (e.g., ablation) are provided by a single catheter or a group of catheters, and thus the mapping catheter also operates as a treatment (e.g., ablation) catheter at the same time.

[0098] Mapping of cardiac regions such as cardiac sites, tissues, veins, arteries, and / or electrical pathways of the heart can lead to the identification of problem areas such as scar tissue, arrhythmia sources (e.g., electrical rotors), healthy regions, etc. The cardiac regions can be mapped such that a visual rendering of the mapped cardiac regions is provided using a display, as further disclosed herein. Further, cardiac mapping can include mapping based on one or more modalities including, 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 set values and / or medical expert preferences.

[0099] Cardiac mapping can be performed using one or more techniques. As an example of a first technique, cardiac mapping may be performed by sensing electrical properties of cardiac tissue, such as local excitation time, as a function of an exact location within the heart. The corresponding data can be acquired using one or more catheters advanced into the heart using a catheter having electrical and location sensors at its distal tip. As a specific example, location and electrical activity can first be measured at about 10 to about 20 points on the inner surface of the heart. These data points may generally be sufficient to generate a preliminary reconstruction or map of the heart surface with satisfactory quality. This preliminary map can be combined with data acquired at additional points to generate a more comprehensive map of the electrical activity of the heart. In a clinical setting, it is not uncommon to accumulate data at over 100 sites to generate a detailed and comprehensive map of the electrical activity of the cardiac chamber. The generated detailed map can then serve as a basis for making therapeutic decisions, such as decisions regarding tissue ablation, to modify the propagation of the electrical activity of the heart and restore normal cardiac rhythm.

[0100] Using a catheter that houses a position sensor, the trajectory of each point on the heart surface can be determined. Using these trajectories, motion characteristics such as the contractile force of the tissue can be inferred. A map showing such motion characteristics can be constructed when the trajectory information is sampled at a sufficient number of points within the heart.

[0101] Electrical activity at a point within the heart can typically be measured by advancing a catheter that houses an electrical sensor at or near its distal tip to that point within the heart, bringing the tissue into contact with the sensor, and acquiring data at that point. One drawback associated with mapping the ventricles using a catheter that houses only a single distal tip electrode is that it requires a long time to accumulate data point by point for the required number of points needed for a detailed map of the heart cavity as a whole. Therefore, multi-electrode catheters have been developed to simultaneously measure electrical activity at multiple points within the heart cavity.

[0102] The multi-electrode catheter can be implemented using any applicable shape such as a linear catheter having a plurality of electrodes, a balloon catheter including electrodes disposed on a plurality of struts that form a balloon, a lasso catheter or loop catheter having a plurality of electrodes, or any other applicable shape. FIG. 16A shows an embodiment of a linear catheter 1602 that includes a plurality of electrodes 1604, 1605, and 1606 that can be used to map a heart region. The linear catheter 1602 can be fully or partially elastic so that it can twist, bend, and / or otherwise change its shape based on the received signal and / or based on the application of an external force (e.g., heart tissue) to the linear catheter 1602.

[0103] Figure 16B shows an embodiment of a balloon catheter 1612 that includes a plurality of splines (e.g., 12 splines in the specific example of FIG. 16B) including splines 1614, 1616, 1617, and a plurality of electrodes on each spline including electrodes 1621, 1622, 1623, 1624, 1625, and 1626 as shown. The balloon catheter 1612 may be designed such that when deployed within a patient's body, its electrodes can be held in intimate contact with the surface of the endocardium. By way of example, the balloon catheter can be inserted into a lumen such as a pulmonary vein (PV). The balloon catheter can be inserted into the PV in a collapsed state such that, as a result, the balloon catheter does not occupy the maximum volume of the PV while being inserted therein. The balloon catheter may expand while inside the PV such that the electrodes on the balloon catheter contact the entire circular portion of the PV. Such contact with the entire circular portion of the PV, or any other lumen, may enable efficient mapping and / or ablation.

[0104] Figure 16C shows an embodiment of a loop catheter 1630 (also referred to as a lasso catheter) that includes a plurality of electrodes 1632, 1634, and 1636 that can be used to map a cardiac region. The loop catheter 1630 can be fully or partially elastic such that it can twist, bend, and / or otherwise change its shape based on received signals and / or based on the application of an external force (e.g., cardiac tissue) on the loop catheter 1630.

[0105] According to one example, a multi-electrode catheter can be advanced into a heart chamber. Anteroposterior (AP) and lateral fluoroscopic images can be acquired to establish the position and orientation of each of the electrodes. An electrogram can be recorded from each of the electrodes that contact the heart surface relative to a time reference such as the onset of the P wave in sinus rhythm from a body surface ECG. The systems further disclosed herein can distinguish between electrodes that record electrical activity and electrodes that do not record electrical activity by virtue of not being in proximity to the endocardial wall. After an initial electrogram is recorded, the catheter can be repositioned and the fluoroscopic and electrogram images can be recorded again. An electrical map can then be constructed from the repetition of the above process.

[0106] According to one example, a cardiac mapping can be generated based on the detection of an intracardiac potential field. A non-contact method can be implemented to simultaneously acquire a large amount of cardiac electrical information. For example, a catheter having a distal end portion can include a series of sensor electrodes that are distributed over its entire surface and are connected to an insulated conductor for connection to signal sensing and processing means. The size and shape of the end portion can be such that the electrodes are disposed substantially spaced from the wall of the heart chamber. The intracardiac potential field can be detected during one heartbeat. According to one example, the sensor electrodes can be distributed on a series of circumferences located in a plane spaced from each other. These planes can be perpendicular to the long axis of the end of the catheter. At least two additional electrodes can be disposed adjacent to both ends of the long axis of the end portion. As a more specific example, the catheter can include four circumferences having eight electrodes equally angularly spaced on each circumference. Thus, in this particular implementation, the catheter can include at least 34 electrodes (32 circumferential electrodes and two end electrodes).

[0107] According to another example, an electrophysiological cardiac mapping system and technique based on a non-contact and non-expanding multi-electrode catheter can be implemented. An electrogram can be acquired using a catheter having a plurality of electrodes (e.g., 42 to 122 electrodes). According to this implementation, knowledge of the relative geometric shapes of the probe and the endocardium may be obtained by an independent imaging modality such as transesophageal echocardiography. After independent imaging, non-contact electrodes can be used to measure cardiac surface potentials and construct a map therefrom. This technique may include the following steps (after the independent imaging step), namely, (a) measuring potentials by a plurality of electrodes arranged 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 endocardial potentials based on the electrode potentials and the matrix of coefficients.

[0108] According to another example, techniques and apparatus for mapping the potential distribution of a cardiac chamber can be implemented. A cardiac multi-electrode mapping catheter assembly can be inserted into a patient's heart. This mapping catheter assembly can include a multi-electrode array having an integral reference electrode, or preferably, a companion reference catheter. These electrodes can be deployed in the form of a substantially spherical array. The electrode array can be spatially referenced to a point on the endocardial surface by a reference electrode or by a reference catheter that contacts the endocardial surface. A preferred electrode array catheter can have a number of individual electrode sites (e.g., at least 24). Additionally, this exemplary technique can be implemented with knowledge of the location of each of the electrode sites on the array and knowledge of the geometry of the heart. These locations are preferably determined by impedance plethysmography.

[0109] According to another example, a cardiac mapping catheter assembly can include an electrode array that defines a number of electrode sites. This mapping catheter assembly can also include a lumen for receiving a reference catheter having a distal tip electrode assembly that can be used to examine the heart wall. The mapping catheter can include a braid of insulated wires (e.g., having 24 to 64 wires in the braid), and each of the wires can be used to form an electrode site. The catheter can be easily positionable within the heart to collect electrical activity information from a first set of non-contact electrode sites and / or a second set of contact electrode sites.

[0110] According to another example, another catheter for mapping electrophysiological activity within the heart can be implemented. The catheter body can include a distal tip adapted to supply a stimulation pulse for pacing the heart, or an ablation electrode for ablating tissue in contact with its tip. The catheter may further include at least a pair of orthogonal electrodes, which generate a differential signal indicative of local cardiac electrical activity adjacent to the orthogonal electrodes.

[0111] According to another embodiment, a process for measuring electrophysiological data within the heart chamber can be implemented. The method can include, in part, positioning a set of active and passive electrodes on the heart, supplying 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 disposed on the inflatable balloon of a balloon catheter. In a preferred embodiment, the array is said to have 60 to 64 electrodes.

[0112] According to another example, cardiac mapping can be performed using one or more ultrasonic transducers. The ultrasonic transducer can be inserted into the patient's heart and can collect multiple ultrasonic slices (e.g., two-dimensional or three-dimensional slices) at various locations and orientations within the heart. If the location and orientation of a particular ultrasonic transducer are known, the collected ultrasonic slices can be stored so that they can be displayed later. One or more ultrasonic slices corresponding to the position of a probe (e.g., a therapeutic catheter) can be displayed later, and the probe can be overlaid on one or more ultrasonic slices.

[0113] According to other examples, a body patch and / or body surface electrodes can also be placed on or near the patient's body. A catheter having one or more electrodes can be positioned within the patient's body (e.g., within the patient's heart), and the position of the catheter can 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. Further, the catheter electrodes can sense biological data (e.g., LAT values) from within the patient's body (e.g., within the heart). The biological data can be associated with the determined position of the catheter, and as a result, a rendering of the patient's body part (e.g., the heart) can be displayed, showing the biological data overlaid on the shape of the body part as measured for each position of the catheter.

[0114] FIG. 17 is a diagram of an exemplary system 1720 in which one or more features of the subject matter of the present disclosure may be implemented. All or a portion of system 1720 may be used to collect information from a training data set and / or all or a portion of system 1720 may be used to implement a trained model or a trained learning system. System 1720 may include components such as a catheter 1740 configured to obtain biometric measurement data for cardiac mapping and / or configured to damage tissue regions of an internal organ. Thus, it will be understood that catheter 1740 may be a mapping catheter, a treatment (e.g., ablation) catheter, or both. The disclosure herein may refer to catheter 1740 operating as a mapping catheter, a treatment catheter, or both, but it will be understood that one or more catheters may be used to implement the subject matter disclosed herein. Although catheter 1740 is shown as a point catheter, it will be understood that embodiments disclosed herein may be implemented using a catheter of any shape that includes one or more elements (e.g., electrodes). System 1720 includes a probe 1721 having a shaft that can be navigated by a physician 1730 to a body part such as a heart 1726 of a patient 1728 lying horizontally on a table 1729. According to an embodiment, a plurality of probes may be provided, but for simplicity, a single probe 1721 is described herein, and it will be understood that probe 1721 may represent a plurality of probes. As shown in FIG. 17, physician 1730 can insert shaft 1722 through sheath 1723 while operating the distal end of shaft 1722 using a manipulator 1732 near the proximal end of catheter 1740 and / or deflection from sheath 1723. As shown in the inset FIG. 1725, catheter 1740 may be attached to the distal end of shaft 1722. Catheter 1740 may be inserted through sheath 1723 in a collapsed state and then expanded within heart 1726.As further disclosed herein, catheter 1740 may include at least one ablation electrode 1747 and a catheter needle 1748.

[0115] According to an embodiment, catheter 1740 may be configured to ablate a tissue region of a cardiac chamber of heart 1726. Insertion diagram 1745 shows catheter 1740 inside a cardiac chamber of heart 1726 in an enlarged view. As shown, catheter 1740 may include at least one ablation electrode 1747 coupled to the body of the catheter. According to other embodiments, a plurality of elements may be connected via a spline that forms the shape of catheter 1740. One or more other elements (not shown) may be provided, and they may be any element configured to perform ablation or acquire biological data, and may be electrodes, transducers, or one or more other elements.

[0116] According to embodiments disclosed herein, ablation electrodes such as electrode 1747 may be configured to supply energy to a tissue region of an internal organ such as heart 1726. The energy may be thermal energy and may cause damage to the tissue region starting from the surface of the tissue region and extending into the thickness of the tissue region.

[0117] According to multiple embodiments disclosed in this specification, the biological data may include one or more of LAT, electrical activity, topology, bipolar mapping, dominant frequency, impedance, etc. The local activation time may be the time point of the threshold activity corresponding to local activation, calculated based on a normalized initial starting point. The electrical activity may be any applicable electrical signal that can be measured based on one or more thresholds, and can be sensed and / or enhanced based on the signal-to-noise ratio and / or other filters. The topology may correspond to the physical structure of a body part or a part of a body part, and may correspond to changes in the physical structure regarding different parts of the body part or regarding different body parts. The dominant frequency may be a frequency or a range of frequencies that pervades a part of a body part, and may be different in different parts of the same body part. For example, the dominant frequency of the pulmonary vein of the heart may be different from the dominant frequency of the right atrium of the same heart. The impedance may be a resistance measurement value in a given region of a body part.

[0118] As shown in FIG. 17, the probe 1721 and the catheter 1740 may be connected to the console 1724. The console 1724 may include a processor 1741 such as a general-purpose computer with a suitable front-end and interface circuit 1738 for transmitting and receiving signals to and from the catheter and for controlling other components of the system 1720. In some embodiments, the processor 1741 may be further configured to receive biological 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 and may be located, for example, within the catheter, within an external device, within a mobile device, within a cloud-based device, or may be a stand-alone processor.

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

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

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

[0122] Network 1762 can be any network or system generally known in the art, such as an intranet, a local area network (LAN), a wide area network (WAN), a metropolitan area network (MAN), a direct connection or series of connections, a cellular phone network, or any other network or medium that can facilitate communication between the mapping system 1720 and the server 1760. Network 1762 can be wired, wireless, or a combination of both. A wired connection can be implemented using Ethernet, a universal serial bus (USB), an RJ-11, or any other wired connection generally known in the art. A wireless connection can be implemented using Wi-Fi, WiMAX, and Bluetooth, infrared, a cellular network, a satellite, or any other wireless connection method generally known in the art. Additionally, some networks can operate alone or communicate with each other to facilitate communication within network 1762.

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

[0124] The control console 1724 can be connected to the body surface electrode 1743 by a cable 1739, and the body surface electrode can include an adhesive skin patch attached to the patient 1730. The processor can determine the position coordinates of the catheter 1740 within a body part of the patient (e.g., the heart 1726) in conjunction with the current tracking module. The position coordinates can be based on the impedance or electromagnetic field measured between the body surface electrode 1743 and the electrode 1748 or other electromagnetic components of the catheter 1740. Additionally or alternatively, the location-specific pad can be disposed on the surface of the bed 1729 and can be separated from the bed 1729.

[0125] The processor 1741 can include a real-time noise reduction circuit typically configured as a field programmable gate array (FPGA), and a subsequent analog-to-digital (A / D) electrocardiograph (ECG) or electromyogram (EMG) signal conversion integrated circuit. The processor 1741 can pass the signal from the A / D ECG or EMG circuit to another processor and / or can be programmed to perform one or more of the functions disclosed herein.

[0126] The control console 1724 can also include an input / output (I / O) communication interface that enables the control console to transfer signals from and / or to the electrode 1747.

[0127] During the procedure, the processor 1741 may facilitate presenting the body part rendering 1735 to the physician 1730 on the display 1727 and may also store data representing the body part rendering 1735 in the memory 1742. The memory 1742 may include any suitable volatile and / or non-volatile memory, such as random access memory or a hard disk drive. In some embodiments, the medical professional 1730 may be able to manipulate the body part rendering 1735 using one or more input devices such as a touchpad, mouse, keyboard, gesture recognition device, etc. For example, the position of the catheter 1740 may be changed using the input device so that the rendering 1735 is updated. In an alternative embodiment, the display 1727 may include a touch screen configured to receive input from the medical professional 1730 in addition to presenting the body part rendering 1735.

[0128] According to an exemplary embodiment of the disclosed subject matter, the new CPM matrix may be generated based on the historical CPM matrix with little or no corresponding catheter location information. Notably, the techniques disclosed herein may enable generating a CPM matrix without physically covering the entire heart chamber using an intracorporeal catheter and determining the ratio of each location. Advantageously, the new supplemental CPM matrix may be generated based on the sparse CPM matrix, and thus, the time required to generate a CPM matrix with increased resolution is dramatically reduced (such a process typically requires moving an intracorporeal catheter to locations within the heart and collecting location data and data associated with patient properties - e.g., electrical signals of the heart).

[0129] In the supplemental CPM matrix, the density of the mapping between patient properties (e.g., electrical signals) and location is higher than that of the sparse CPM matrix. In one example, to generate the sparse CPM matrix, electrical signals having corresponding locations are taken at each 3 mm of the heart 3It is collected in the voxel. However, the electrical signal having the corresponding location can be collected at more locations of the heart (e.g., each 1 mm 3 (in the voxel)). As described above, the new supplementary matrix having the higher mapping density can be generated only for the calculated sparse CPM matrix provided to the learning system.

[0130] The historical CPM matrix may be generated for each given catheter location (e.g., based on the corresponding cluster) based on the current distribution from the signals transmitted from the intravascular catheter electrode and received by a plurality of body surface (BS) patches on the patient's body.

[0131] FIG. 18a shows a process 1800 according to an embodiment of the present invention for determining a new augmented CPM matrix from a sparse CPM matrix. FIG. 18b shows an alternative or additional process 1800 for determining a catheter location based on a new electrical signal and generating a CPM matrix without new catheter location information corresponding to the new electrical signal. For clarity, a number of instances of a historical CPM matrix having corresponding patient characteristics (e.g., electrical signals, catheter orientation and / or position, cardiac anatomy, etc.) and catheter locations can be used as training data for generating a model as provided herein, according to FIGS. 4-15. To avoid misunderstanding, it should be understood that the "generated model" can refer to the same features as the "trained learning system". In other words, the "generated model" or "trained learning system" can refer to features trained to receive an input (e.g., a sparse CPM matrix) and generate an output when the input is provided (e.g., generate an augmented CPM matrix when a sparse CPM matrix is provided). The model / learning system is configured to be "trained" based on the training data, and as a result, the model / learning system can determine the relationship between the input and output of the training data. Based on the trained components of the model / learning system, an improved augmented CPM matrix may be identified, which can then be used to predict the catheter location assuming the received electrical signal. This implementation facilitates the generation of the CPM matrix as it reduces the need to capture location information and corresponding ratios for each catheter position.

[0132] According to some embodiments, a second model is provided that can be trained such that a new electrical signal transmitted by an electrode of a catheter can be captured, for example, by a BS patch and supplied to the model along with patient characteristics. Alternatively or additionally, the second model may be trained such that a new electrical signal is transmitted by the BS patch and received by an electrode of the catheter, and may be supplied to the model along with patient characteristics. Based on the trained components of the second model, the catheter location can be predicted and the CPM matrix can be identified. This implementation facilitates the generation of the CPM matrix because it reduces the need to capture location information and corresponding ratios for each catheter position.

[0133] As shown in flowchart 1800 of FIG. 18a, at step 1802, a historical sparse CPM matrix and a historical complementary CPM matrix can be collected. At step 1804 of process 1800 of FIG. 18a, the historical sparse CPM corresponding to the historical complementary CPM collected at step 1802 can be used as training data for the learning system. At step 1804, the training data can be used to train the learning system based on a given algorithm. The learning system may be trained such that when a new sparse CPM matrix is provided, the trained learning system is configured to provide a new complementary CPM matrix as an output having a minimum catheter location or no catheter location.

[0134] At step 1808 of process 1800 of FIG. 18a, new patient characteristics about the patient (e.g., electrical signals, catheter orientation and / or position, heart anatomy, etc.) can be received by the model generated at step 1806 or provided as an input to the model. At step 1810, the model may output a predicted catheter location and / or generate a new CPM matrix based on the input electrical signal.

[0135] As shown in flowchart 1800 of FIG. 18b, in step 1802, a historical CPM matrix, certain patient properties (e.g., electrical signals, catheter orientation and / or position, heart anatomical structure, etc.), and corresponding catheter locations may be collected. The corresponding catheter location may be the catheter location identified (e.g., via magnetic mapping electrodes) when the electrical signal of the corresponding CPM matrix is generated. In step 1804 of process 1800 of FIG. 18, the historical CPM, patient properties, and corresponding locations collected in step 1802 can be used as training data for a learning system. In step 1804, the training data can be used to train the learning system based on a given algorithm. In step 1806 of process 1800, a model can be generated using the trained learning system. The model may be generated such that when new patient properties (e.g., electrical signals, catheter orientation and / or position, heart anatomical structure, etc.) are provided, the model provides a new CPM matrix as an output having a minimum catheter location or no catheter location.

[0136] In step 1808 of process 1800 in FIG. 18b, new patient properties about a patient (e.g., electrical signals, catheter orientation and / or position, heart anatomical structure, etc.) can be received by the model generated in step 1806 or provided as input. In step 1810, the model may output a predicted catheter location and / or generate a new CPM matrix based on the input electrical signals. FIG. 19 shows an exemplary implementation for collecting electrical signals. As shown, catheter 1910 may be inserted into a heart chamber (e.g., a cluster) 1900 and may include electrodes configured to emit electrical signals. The electrical signals may be received by one or more BS patches 1920 such that a CPM matrix can be generated based on the received electrical signals emitted by catheter 1910. According to an alternative implementation, as shown in FIG. 19, catheter 1910 may be inserted into a heart chamber (e.g., a cluster) 1900 and may include electrodes configured to receive electrical signals. The electrical signals may be transmitted by one or more BS patches 1920 such that a CPM matrix can be generated based on the electrical signals received at catheter 1910 for transmission by BS patches 1920.

[0137] As shown in the flowchart 1800 of FIG. 18a or FIG. 18b, in step 1802, a historical CPM matrix and corresponding catheter locations can be received. The historical CPM matrix can be generated based on an adaptive CPM estimation process and a CPM application process. The CPM application process may further include a cluster selection module and a CPM application module.

[0138] The adaptive CPM estimation process is a vector obtained from any hybrid catheter having an EM sensor, such as a catheter, and electrodes associated therewith

[0139]

Number

[0140] The matrix constructed in the historical CPM estimation process is constructed over time, and therefore, there is an initialization period in the active current location (ACL) step, during which the processor receives initial data from the decomposition step. For a specific cluster, when the processor has collected sufficient data regarding that cluster, the processor can generate the matrix of the cluster.

[0141] In historical CPM application, the generated matrix is used together with the current measurement values of the cathode electrodes to calculate the location of each electrode in real time. This calculation is performed according to the following formula.

[0142]

Number

[0143]

Number

[0144]

Number

[0145] CPM vector

[0146]

Number

[0147]

Number

[0148]

Number

[0149] The processor may use the respiration indicator

[0150]

Number

[0151]

Number

[0152] When the breathing indicator is accumulated, the processor performs principal component analysis (PCA) on the indicator to find the direction between elements having the maximum eigenvalue as follows.

[0153]

Number

[0154] The direction given by the above equation is used to calculate the breathing descriptor value as follows.

[0155]

Number

[0156] The average and range of the RDi values are calculated as follows.

[0157]

Number

[0158] The average and range are used to calculate the normalized value RDni of RDi within the range of 0 to the maximum value ClNo. CLNo is a number that defines the resolution of the update holder described later and is typically about 5.

[0159]

Number

[0160] The value of RDni is stored in the memory.

[0161] Figure 20 is a process 2000 showing the steps performed by a processor to generate a CPM matrix according to an embodiment of the present invention. The steps of process 2000 are performed in an adaptive CPM estimation process when each measurement value is generated by a catheter.

[0162] In a first step 2002, as described above, measurement values are received from any hybrid catheter, and the processor processes those measurement values into a CPM vector, as described herein.

[0163]

Number

[0164] In a first update holder step 2004, an update holder index for the measurement value is calculated. In a first condition 2006, the processor checks whether the update holder index already exists by checking whether the index is stored in memory. If the index already exists, the measurement value is discarded and the process ends.

[0165] If the index does not exist, in a save step 2008, the index and the measurement value can be stored in a memory buffer. The measurement value is stored as a vector

[0166]

Number

[0167] In a cluster association step 2010, the measurement value is associated with a corresponding cluster. This association is performed by calculating a corresponding cluster index from the measurement value. The measurement value is associated with this cluster index.

[0168] Cluster origin

[0169]

Number

[0170]

Number

[0171] In the second update holder step 2012, the update holder index of the adjacent cluster is calculated using the measurement value received in step 2002. If the update index is not yet occupied, the measurement value is placed in the buffer and its index is saved. If the index is already occupied, no processing is performed.

[0172] In the second condition 2014, the number M of update indices in each cluster Clx is evaluated. If M is greater than a predetermined number, typically about 40, in the cluster matrix step 2016, the CPM matrix A of the cluster is calculated using the following formula,

[0173]

Number

[0174]

Number

[0175]

Number

[0176] The two CPM matrices A may be calculated for each cluster, one using measurements with a reference catheter and one without reference catheter measurements, and then process 2000 ends.

[0177] In condition 2014, if M is less than a predetermined number, process 2000 ends.

[0178] To confirm that the calculated results are not self - contradictory, the calculations in process 2000 are verified at various stages. For example, in the cluster association step 2010, if the number of existing adjacent clusters is less than a predetermined number, say less than 4, an error is assumed and the measurements in step 2002 are not admitted. Other self - consistency checks regarding the operation of the process will be apparent to those skilled in the art.

[0179] Accordingly, a historical CPM matrix based on the corresponding catheter locations may be generated according to process 2000 of FIG. 20. In step 1802 of process 1800 of FIGS. 18a and 18b, the historical CPM matrix may be received along with the corresponding patient properties (e.g., electrical signals, catheter orientation and / or position, heart anatomical structure, etc.) for each given historical CPM matrix. As further disclosed herein, historical CPM matrices may be provided for a number of patients. The number of patients for which historical CPM data is used may depend on one or more of the quality of the CPM data and the corresponding catheter locations, and this may be determined on a case - by - case basis.

[0180] According to an implementation form, the CPM matrix provided to a large number of patients can correspond to a treatment in which an accurate matrix is generated based on patient properties (e.g., electrical signals, catheter orientation and / or position, heart anatomical structure, etc.) and corresponding catheter locations. For example, as further disclosed herein, one set out of a set of 100 patients' CPM data may be available for training the system. From 100 available sets of CPM data, only 70 sets can correspond to treatments that meet the quality threshold. These 70 sets can be used as training data for the embodiments disclosed herein. In contrast, the 30 sets that do not meet the quality threshold may not be included as training data.

[0181] In step 1804 of process 1800 in FIG. 18a, the learning system can be trained using the historical sparse CPM matrix and the corresponding supplemental CPM matrix. This training can be performed using hardware, software, and / or firmware. The training can include the analysis and correlation of the CPM matrix and the corresponding supplemental CPM matrix received in step 1802. In particular, the learning system is configured to determine whether there is a correlation or link between each of the sparse CPM matrices and their respective supplemental CPM matrices.

[0182] In step 1804 of process 1800 in FIG. 18b, the learning system can be trained using the historical CPM matrix and the corresponding catheter location. This training can be performed using hardware, software, and / or firmware. The training may include the analysis and correlation of the CPM matrix, patient properties (e.g., electrical signals, catheter orientation and / or position, heart anatomical structure, etc.), and the corresponding catheter location received in step 1802. In particular, the CPM matrix value at a given catheter location can be used to determine whether there is a correlation or link between a given value or set of values and the corresponding catheter location (e.g., whether a given set of matrix values correlates to a given catheter location).

[0183] The CPM matrix, patient characteristics, and catheter location characteristics collected in step 1802 may be extracted, and may include patterns, relationships, correlations, electrical signals, catheter orientation and / or position, heart anatomical structures, and the like. In step 1804, a feature matrix may be generated based on the features extracted, and the learning system may be trained. According to an implementation form, the learning system may be trained based on a machine learning algorithm.

[0184] (In step 1806, in order to provide a trained learning system and / or generate a model), an algorithm may be used to train the learning system. This algorithm may be, for example, a classification algorithm, a regression algorithm, a clustering algorithm, or any applicable algorithm capable of generating a model for estimating the location of arrhythmia using ECG data.

[0185] As an example, FIG. 21 shows a logistic regression plot for predicting whether a particular patient property (e.g., electrical signal, catheter orientation and / or position, heart anatomy, etc.) is likely to correspond to a given catheter location within the heart. This logistic regression then enables prediction of the heart location and corresponding CPM matrix ratio based on the patient properties input into the model. For example, based on a plurality of patient properties such as a given heart anatomy and the initially emitted electrical signal, the outcome of the corresponding heart location can be provided by a logistic regression model. The past history of the combination of a given patient property and the relationship to a specific heart location enables the prediction to be made. The logistic regression of FIG. 21 enables the analysis of a combination of a given patient property (e.g., heart anatomy and received electrical signal) represented by variable 2120 to map to a given heart location, e.g., a cluster within the left atrium, and the probability 2110 is defined by 0 to 1. At the lower end 2130 of the sigmoid curve, a combination of a given patient property 2120 may correspond to an outcome 2110 of not being within the left atrium. While at the upper end 2140 of the sigmoid curve, the combination 2120 estimates an outcome 2110 of being in the left atrium.

[0186] As described above, a number of different combinations of such features (e.g., patient properties) can be used to generate a plurality of logistic regression-based outcomes, each predicting the likelihood of a heart location and its respective CPM matrix. Using such a combination of logistic regression-based outcomes, a logistic regression model generated based on the training of the learning system in step 1904 can be generated in step 1906.

[0187] Although a logistic regression-based model is provided as an example, it will be understood that any applicable algorithm (e.g., classification, regression clustering, etc.) can be used to generate each model in step 1906.

[0188] When a learning system is trained using sufficient training data (e.g., features) to generate an applicable model, that model can be used with new data. In step 1908, new patient properties (e.g., electrical signals, catheter orientation and / or position, heart anatomical structure, etc.) may be applied to the model training learning system, and the model training learning system may extract features from the new patient properties. A feature vector can be generated and input into the model generated in 1906. The model may use the feature vector as an input (e.g., as shown in FIG. 15) and may predict a complementary CPM matrix from a sparse CPM matrix. An alternative model / trained learning system may use the feature vector as an input and may predict a CPM matrix and / or the location of the heart based on patient properties. According to this alternative form, a generated CPM matrix and a subset of the corresponding known heart locations may be provided to the model / trained learning system such that the model / trained learning system outputs the remainder of the CPM matrix corresponding to the unknown heart location.

[0189] A system and method for creating an estimated location of each electrode of a catheter based on measurement of current by a chest patch are disclosed. The system and method estimate the location of the electrodes of the catheter when given measurements of current recorded by a patch (e.g., 6) placed on a patient's chest. The catheter contains electrodes (e.g., 2 in the main stem and 4 each on 5 arms for a total of 22, as illustrated in FIG. 16B). Each electrode on the catheter emits a current of a different frequency. The current travels through the body until it reaches sensors on the patch, and the sensors create readings of the current. When an electrode is located at a particular position, the patch creates one or more vector measurements that describe the current distribution and a respiration indicator. This vector may be referred to as VEC. The system and method may create an estimated location of the electrodes (in 3D space) when given VEC.

[0190] One way to estimate the location of the electrode is to utilize an electromagnetic (EM) sensor located on the main stem of the catheter. The position of the EM sensor in 3D space can be known. One of the electrodes on the catheter, referred to as the mapping electrode, is an electrode placed immediately adjacent to the EM sensor at a given distance on a non-flexible catheter stem. Thus, its position in 3D space can be accurately known assuming the position based on the EM sensor.

[0191] When the physician moves the catheter around the heart, the system collects, at each point, measurements of the known position (based on the EM sensor) of the mapping electrode, along with the VEC of the mapping electrode. The system and method can construct a Current to Position Mapping (CPM). The CPM includes a mapping from VEC values to the voxels visited by the mapping electrode, leading to the VEC values at that position.

[0192] The CPM can include estimates for other nearby voxels not yet visited by the mapping electrode. The estimate at a particular voxel is currently created by linear interpolation of the VEC values from nearby voxels where both the VEC value and the position value are reliably known.

[0193] The present system and method can estimate the 3D position of the electrode using the CPM. That is, when a VEC value for the electrode is given, the CPM is utilized to identify which voxel contains this VEC value. This voxel indicates the 3D position of the electrode.

[0194] First, the CPM is very small, so not all VEC values of the electrode can be found in the CPM, and thus the location of the electrode cannot be estimated. As the catheter visits more distinct positions within the heart, the CPM can become more comprehensive, and the electrode location estimation becomes available for more locations. The location estimation can become more accurate as more measurements are collected.

[0195] FIG. 22 illustrates the configuration of a system 2200 for predicting electrode positions. In FIG. 22, "M.E." or "ME" indicates a mapping electrode. The system 2200 includes a neural network (NN) 2260 that can output an estimated value of the electrode position 2280 when a VEC value 2230.3 for this electrode is given. The NN 2260 is trained "on the fly" 2200.3 for a new patient 2275 by using measurement values from the EM sensor and the ME 2250.3, and is then applied to VEC measurement values 2230.3 for other electrodes to estimate their locations. For a given patient, the NN is not trained "from tabula rasa" because there are not enough measurement values for the data of the given patient. Instead, the NN2260vis initializes the offline 2200.1 by training the NN2260.1 on data collected from previous patients 2205.1 and data collected from simulated models. Next, in a new patient, transfer learning 2260.2 is utilized to update the NN model 2260 based on measurement values from a specific patient 2205.2. Since different heart structures vary, the training of the NN 2260 cannot be performed on raw measurement values collected from previous patients 2205.1. Registration (alignment), that is, converting the raw data into a universal coordinate system 2240.1 is necessary.

[0196] Next, refer also to FIG. 23, which illustrates the configuration of the core of NN 2300 that can function as NN 2260. FIG. 23 will be described with reference also to FIG. 14. NN 2300 that functions as NN 2260 may include several layers. The input layer 2310 (represented as five nodes 2310.1, 2310.2, 2310.3, 2310.4, 2310.5) includes vectors representing VEC measurement values. The output layer 2340 (represented as three nodes 2340.1, 2340.2, 2340.3) includes a vector of three values representing a position in 3D space. The number and size of the intermediate layers 2320, 2330, and their architecture can be changed. As illustrated in FIG. 23, there is an intermediate layer 1 2320 (represented as nine nodes 2320.1, 2320.2, 2320.3, 2320.4, 2320.5, 2320.6, 2320.7, 2320.8, 2320.9) and an intermediate layer 2 2330 (represented as seven nodes 2330.1, 2330.2, 2330.3, 2330.4, 2330.5, 2330.6, 2330.7).

[0197] Various input signals are supplied to the input layer 2310 on the left side. Additional hidden layers are used to represent various features of the input and non-linear combinations of the input. The output layer 2340 combines the signals into a vector representing a position in 3D space. Depending on the nature of the data and the required output accuracy, more (or fewer) intermediate layers 2320, 2330, and intermediate layers with different numbers of nodes or different connection topologies may be used.

[0198] This type of topology may be able to represent any function (universal approximation theorem) that maps an input to an output, but in practice, depending on the nature of the data, some deformed forms or other topologies of it may be required. For example, another NN component that classifies the input into a specific class may be added, and this classification value is supplied to other layers.

[0199] As will be understood, any number of layers and nodes within the layers may be utilized, and the depiction in FIG. 23 is for illustration and understanding only.

[0200] To provide a vector representing the VEC measurement value to the input layer 2310, data from previous patients 2205.1 is used to train the system 2200.1 and to train the NN 2260.1. For each patient, the data contains measurement values consisting of pairs of the position 2220.1 of the ME (based on the EM sensor position) and the VEC 2230.1 of this electrode. Additionally, the data from previous patients 2205.1 may include a Carto 3D model 2210.1 of the heart tissue.

[0201] This type of data 2205.2 is also required for the online transfer learning phase 2200.2. To provide a vector representing the VEC measurement value to the input layer 2310, data from the current case 2205.2 is used to fine-tune the system 2200.2 and to train the NN 2260.2. For each patient, the data contains measurement values consisting of pairs of the position 2220.2 of the ME (based on the EM sensor position) and the VEC 2230.2 of this electrode. Additionally, the data from previous patients 2205.2 may include a Carto 3D model 2210.2 of the heart tissue.

[0202] Since different heart structures vary and the location where the patches are placed on the patient also varies, the training of the NN 2260 cannot be performed on the raw measurement values collected from previous patients 2205.1. Registration (alignment) is required, that is, converting the raw data into a universal coordinate system (UCS) (2240.1 for previous patients and 2240.2 for the current patient). Given the Carto 3D models 2210.1, 2210.2 of the heart tissue, the UCS can be constructed. For example, in each heart chamber, the origin of the UCS may be accurately located at the center of mass of the heart chamber. The x-axis can be aligned with particularly relevant heart structures. To convert to this UCS, the 3D locations within the available data can be shifted, rotated, and stretched.

[0203] The transformed locations 2240.1, 2240.2 are used as the desired output values of NN 2260.1, 2260.2 together with the ME positions within UCS 2250.1, 2250.2. Therefore, the raw data 2205.1, 2205.2 is first transformed to UCS 2240.1, 2240.2 in order to train 2260.1 the system 2200 and fine-tune 2260.2 the system 2200, and the output of the system 2200 including the ME position of UCS 2250.3 can be transformed back to the original coordinate system 2240.3 as needed during the inference stage 2200.3 of a new patient.

[0204] In the offline stage 2200.1, NN 2260 is trained 2260.1 on data from previous cases 2205.1. As described above, this data may include data consisting of pairs of the known 3D positions of the mapping electrodes (after conversion to UCS) 2250.1 and VEC values 2230.1. The ME positions within UCS 2250.1 can be achieved from the Carto 3D model 2210.1 and ME positions 2220.1 transformed to UCS 2240.1. The VEC values 2230.1 function as inputs to NN 2260, and the 3D positions function as the desired outputs of NN 2260.

[0205] The dataset of past data 2205.1 may be split into a training set, a validation set, and a test set. The first two sets will be used during system development, and the test set will be used only to evaluate the accuracy of the system. Also, cross-validation may be used to improve performance. This incorporates knowledge from past patients 2205.1 into NN 2260.

[0206] After the system 2200 is deployed, additional data may be accumulated and added to the training dataset, and it is allowed for the system 2200 to re-learn and continuously improve its accuracy.

[0207] Historical data 2205.1 may be insufficient to achieve the desired results in the training 2260.1 of NN 2260. A standard technique in machine learning is to augment real data with artificial data created by simulation. A system trained on a combination of real data 2205.1 and simulated data can learn better and be more robust than a system trained only on real data (under the condition that the simulation is close enough to reality).

[0208] There are existing systems that create high-quality simulations of the anatomical structure of the heart, including the current and how it spreads and changes as it passes through heart tissue and liquid. These systems can be used to simulate readings from a catheter as it moves through the heart. The resulting simulated dataset can be added to the real data 2205.1.

[0209] After training the NN 2260 in 2260.1, the trained NN 2270 may undergo online fine-tuning 2200.2. The trained NN 2270 is fine-tuned 2260.2 using data from the current case 2205.2. As described above, this data may include data consisting of pairs of known 3D positions of the mapping electrodes (after conversion to the UCS) 2250.2 and VEC values 2230.2. The ME position within the UCS 2250.2 can be achieved from the Carto 3D model 2210.2 and the ME position 2220.2 converted to the UCS 2240.2. The VEC value 2230.2 functions as an input to the NN 2260, and the 3D position functions as the desired output of the NN 2260.

[0210] The NN 2270 trained on past data 2205.1 may be used "as is" for online inference on a new patient 2205.2. To improve accuracy, fine-tuning 2260.2 may be utilized. Fine-tuning 2260.2, also known as transfer learning, may enable knowledge from a previously learned task to be reused for learning a new task, and thus the accuracy of the outcome is improved. This means that as the catheter visits new positions in the patient's heart, the NN 2260 is continuously trained and updated on data collected from the new patient 2205.2. Due to time constraints, not all layers 2320, 2330 of the NN are relearned, only the last one or two layers are relearned. This provides "fine-tuning" 2260.2 of the NN 2260 for a specific patient's heart structure and measurements. Fine-tuning 2260.2 may be performed after a certain amount of time required for a physician to move the catheter within the heart to create at least a part of the Carto model 2210.2 of the current patient's heart, since this model is required to convert raw input to UCS 2240.2. As more measurements are collected to create the Carto model 2210.2, the conversion to UCS 2240.2 becomes more accurate, and thus the output of the NN 2260 can be improved.

[0211] When the NN 2260 is fine-tuned 2260.2 for patient data 2205.2, the trained NN 2275 can be used to enter the inference stage 2200.3. The NN 2260 may be provided with the VEC values of the electrodes 2230.3 to obtain the positions of the electrodes within the UCS 2250.3. This position then needs to be inverse-transformed to the patient's specific coordinate system by converting from the UCS 2240.3 to obtain the actual position relative to the predicted electrode position 2280. In parallel, more data continues to be collected for the current patient 2205.2 and is used to maintain the fine-tuning 2260.2 of the NN 2260.

[0212] The various chambers within the heart are distinct (left atrium / right atrium, left ventricle / right ventricle, etc.) and have different structures, so separate NN 2260s may be utilized for each such chamber.

[0213] Each entry in the past dataset 2205.1 includes an indication of which chamber the entry was recorded in. Also, during online use, this indication may be provided manually by a physician or automatically based on the classification of catheter readings using a separate system.

[0214] The overall process is performed separately for each chamber. However, transfer learning 2260.2 may be performed on data across chambers to improve both the accuracy and performance of the NN 2260. A multi-chamber approach may be utilized during the learning of the NN 2260.1, 2260.2 as the lower layers automatically learn to detect basic features, which are relevant to all chambers.

[0215] Although the features and elements have been described above in specific 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, the methods described herein can be implemented in a computer program, software, or firmware incorporated in a computer-readable medium for execution by a computer or processor. 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, read-only memory (ROM), random access memory (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 digital versatile disks (DVD). A processor associated with the software can be used to implement a radio frequency transceiver for use in a WTRU, UE, terminal, base station, RNC, or any host computer.

[0216] The following non-exhaustive listing of embodiments also forms part of the present disclosure.

[0217] [[ID=—7]] Embodiment 1. A system for automatically generating a CPM matrix, comprising: a plurality of body surface electrodes configured to sense electrical signals; a processor comprising a neural network, the processor being configured to: receive a plurality of historical CPM matrices, patient characteristics, and corresponding catheter locations; train a learning system based on the plurality of CPM matrices, patient characteristics, and catheter locations; generate a model based on the learning system; receive at least partially new patient characteristics from the plurality of body surface electrodes; A system comprising a processor configured to generate a new CPM matrix based at least in part on new patient properties from a plurality of body surface electrodes.

[0218] Embodiment 2. The system of Embodiment 1, wherein the plurality of received historical CPM matrices are based on known corresponding catheter locations.

[0219] Embodiment 3. The processor of the system of Embodiment 1 is further configured to: Receive a subset of patient properties having known catheter locations; Generate a new CPM matrix based also on the subset of patient properties having known catheter locations.

[0220] Embodiment 4. The system of Embodiment 1, wherein the patient properties are selected from one or more of electrical signals, catheter position, catheter orientation, and heart anatomy.

[0221] Embodiment 5. The system of Embodiment 4, wherein the patient properties include data from previous patients.

[0222] Embodiment 6. The system of Embodiment 5, wherein the data from previous patients includes at least one of a Carto 3D model, a ME position, and a VEC.

[0223] Embodiment 7. The system of Embodiment 5, wherein the processor further transforms the Carto 3D model and the ME position into the UCS.

[0224] Embodiment 8. The system of Embodiment 1, wherein the learning system is trained using at least one of classification, regression, and clustering algorithms.

[0225] Embodiment 9. The system of Embodiment 1, wherein the processor is further configured to transform the ME position in the UCS to a predicted electrode position.

[0226] Embodiment 10. The system of Embodiment 1, wherein the processor is further configured to provide the ME position in the UCS based on training for the electrodes and the VEC.

[0227] Embodiment 11. A method for generating an arrhythmia estimation model, the method comprising: receiving a plurality of historical CPM matrices, patient characteristics, and corresponding catheter locations; training a learning system such that combinations of attributes from the historical CPM matrices and patient characteristics correlate with a first set of corresponding catheter locations based on the first set of the plurality of historical CPM matrices, patient characteristics, and corresponding catheter locations; updating the learning system such that combinations of attributes from the historical CPM matrices and patient characteristics correlate with a second set of corresponding catheter locations based on a second set of the plurality of historical CPM matrices, patient characteristics, and corresponding catheter locations; generating a model based on the first set of corresponding catheter locations and the second set of corresponding catheter locations.

[0228] Embodiment 12. The method of Embodiment 11, wherein the second set of corresponding catheter locations is refined based on a first set of corresponding arrhythmia locations.

[0229] Embodiment 13. The method of Embodiment 11, wherein the plurality of received historical CPM matrices are based on known corresponding catheter locations.

[0230] Embodiment 14. receiving a subset of patient characteristics having known catheter locations; further comprising generating a new CPM matrix based on the subset of patient characteristics having known catheter locations.

[0231] Embodiment 15. The method of Embodiment 11, wherein the patient property is selected from one or more of an electrical signal, a catheter position, a catheter orientation, and a cardiac anatomical structure.

[0232] Embodiment 16. The method of Embodiment 15, wherein the patient property includes data from a previous patient.

[0233] Embodiment 17. The method of Embodiment 16, wherein the data from the previous patient includes at least one of a Carto 3D model, a ME position, and a VEC.

[0234] Embodiment 18. The method of Embodiment 17, further including converting the Carto 3D model and the ME position into a UCS.

[0235] Embodiment 19. The method of Embodiment 11, wherein the learning system is trained using at least one of a classification, a regression, and a clustering algorithm.

[0236] Embodiment 20. Converting the ME position in the UCS to a predicted electrode position, and Providing the ME position within the UCS based on training for the electrode and the VEC, further including the method of Embodiment 11.

[0237] The following enumeration of aspects also forms part of the present disclosure.

[0238] Aspect 1. A system for generating an improved current-to-position mapping (CPM) matrix, comprising a processor, the processor being receiving a plurality of historical sparse CPM matrices and a plurality of historical complementary CPM matrices, each sparse CPM matrix being associated with a respective complementary CPM matrix, receiving Training a learning system based on a plurality of historical sparse CPM matrices and a plurality of historical supplementary CPM matrices, wherein the learning system is trained to generate a supplementary CPM matrix when a sparse CPM matrix is provided, the training, and receiving, by the trained learning system, a new sparse CPM matrix A system configured to generate a new supplementary CPM matrix using the trained learning system.

[0239] Aspect 2. The system of Aspect 1, wherein the processor is further configured to convert the predicted electrode position of the ME position in the UCS.

[0240] Aspect 3. The system of Aspect 1, wherein the processor is further configured to provide the ME position in the UCS based on training and VEC for the electrodes.

[0241] Aspect 4. The system of Aspect 1, wherein the patient characteristics are selected from one or more of electrical signals, catheter position, catheter orientation, and cardiac anatomy

[0242] Aspect 5. The system of Aspect 4, wherein the data from previous patients includes at least one of a Carto 3D model, ME position, and VEC.

[0243] Aspect 6. The system of Aspect 5, wherein the processor further converts the Carto 3D model and ME position to the UCS.

[0244] 〔Embodiment〕 (1) A system for generating an improved current-to-position mapping (CPM) matrix, Comprising a processor, the processor Receiving a plurality of historical sparse CPM matrices and a plurality of historical supplementary CPM matrices, each sparse CPM matrix being associated with a respective supplementary CPM matrix, the receiving Training a learning system based on the plurality of historical sparse CPM matrices and the plurality of historical supplementary CPM matrices, wherein the learning system is trained to generate a supplementary CPM matrix when a sparse CPM matrix is given, and the training; Receiving, by the trained learning system, a new sparse CPM matrix; Generating, using the trained learning system, a new supplementary CPM matrix, a system configured to perform. (2) The system according to embodiment 1, further comprising a plurality of body surface electrodes configured to sense electrical signals. (3) Receiving the new sparse CPM matrix is Receiving a subset of patient properties having known catheter locations and Generating the new sparse CPM matrix from the subset of patient properties and the known catheter locations, the system according to embodiment 1. (4) The processor is Receiving a subset of patient properties having no known catheter locations and Based on the subset of patient properties having no known catheter locations and the new supplementary CPM matrix, configured to determine a catheter location for each of the subset of patient properties having no known catheter locations, the system according to embodiment 1. (5) The system according to embodiment 3, wherein each catheter location includes the location of each electrode of the catheter.

[0245] (6) The system according to embodiment 3, wherein both the subset of patient properties having known catheter locations and the subset of patient properties having no known catheter locations include current signals associated with each electrode of the catheter. The system according to embodiment 6, when dependent on embodiment 2, wherein the current signals associated with the respective electrodes of the catheter are sensed by the plurality of body surface electrodes. (8) The system according to embodiment 3, wherein the patient properties of a subset of patient properties having a known catheter location include data from a treatment in the history. (9) The system according to embodiment 1, wherein the learning system is trained using at least one of classification, regression, and clustering algorithms. (10) The system according to embodiment 1, wherein the learning system comprises a neural network.

[0246] (11) A computer-implemented method for generating a current-to-position mapping (CPM) matrix, the method comprising: receiving a plurality of historical sparse CPM matrices and a plurality of historical complementary CPM matrices, each sparse CPM matrix being associated with a respective complementary CPM matrix; training a learning system based on the plurality of historical CPM matrices and the plurality of historical complementary CPM matrices, wherein the learning system is trained to generate a complementary CPM matrix when a sparse CPM matrix is provided; receiving a new sparse CPM matrix by the trained learning system; generating a new complementary CPM matrix using the trained learning system. (12) Receiving a new sparse CPM matrix comprises: receiving a subset of patient properties having a known catheter location; generating the new sparse CPM matrix from the subset of patient properties and the known catheter location. ((13) Receiving a subset of patient properties having no known catheter location; Based on the subset of patient characteristics without the known catheter locations and the new supplemental CPM matrix, determining a catheter location for each of the subset of patient characteristics without the known catheter locations, the method according to embodiment 11 further comprising. (14) The method according to embodiment 12, wherein each catheter location includes the location of each electrode of the catheter. (15) The method according to embodiment 12, wherein the subset of patient characteristics with the known catheter locations and the subset of patient characteristics without the known catheter locations both include current signals recorded at each electrode of the catheter.

[0247] (16) The method according to embodiment 12, wherein the patient characteristics of the subset of patient characteristics with the known catheter locations include data from past treatments in the history. (17) The method according to embodiment 11, wherein the learning system is trained using at least one of classification, regression, and clustering algorithms. (18) The method according to embodiment 14, wherein the learning system comprises a neural network.

Claims

1. A system for generating an improved current-to-position mapping (CPM) matrix, comprising a processor, the processor being configured to: receive a plurality of historical sparse CPM matrices and a plurality of historical complementary CPM matrices, each sparse CPM matrix being associated with a respective complementary CPM matrix; train a learning system based on the plurality of historical sparse CPM matrices and the plurality of historical complementary CPM matrices, the learning system being trained to generate a complementary CPM matrix when a sparse CPM matrix is provided; receive a new sparse CPM matrix by the trained learning system; and generate a new complementary CPM matrix using the trained learning system.

2. The system of claim 1, further comprising a plurality of body surface electrodes configured to sense electrical signals.

3. Receiving the new sparse CPM matrix includes: receiving a subset of patient properties having known catheter locations; and generating the new sparse CPM matrix from the subset of patient properties and the known catheter locations.

4. The processor is configured to: receive a subset of patient properties without known catheter locations; and determine a catheter location for each of the subset of patient properties without known catheter locations based on the subset of patient properties without known catheter locations and the new complementary CPM matrix.

5. The system of claim 3, wherein each catheter location includes the location of respective electrodes of the catheter.

6. The system of claim 3, wherein the subset of patient properties with known catheter locations and the subset of patient properties without known catheter locations both include current signals associated with respective electrodes of the catheter.

7. The system of claim 6, when dependent on claim 2, wherein the current signals associated with respective electrodes of the catheter are sensed by the plurality of body surface electrodes.

8. ​ The system of claim 3, wherein the patient characteristics of the subset of patient characteristics having the known catheter location include data from procedures in the history.

9. The system of claim 1, wherein the learning system is trained using at least one of classification, regression, and clustering algorithms.

10. The system of claim 1, wherein the learning system comprises a neural network.

11. A computer-implemented method for generating a current-to-position mapping (CPM) matrix, the method comprising: Receiving a plurality of historical sparse CPM matrices and a plurality of historical complementary CPM matrices, each sparse CPM matrix being associated with a respective complementary CPM matrix; Training a learning system based on the plurality of historical CPM matrices and the plurality of historical complementary CPM matrices, the learning system being trained to generate a complementary CPM matrix when a sparse CPM matrix is provided; Receiving a new sparse CPM matrix by the trained learning system; Generating a new complementary CPM matrix using the trained learning system.

12. Receiving a new sparse CPM matrix comprises: Receiving a subset of patient characteristics having a known catheter location; Generating the new sparse CPM matrix from the subset of patient characteristics and the known catheter location. The method of claim 11.

13. Receiving a subset of patient characteristics having no known catheter location; Determining a catheter location for each of the subset of patient characteristics having no known catheter location based on the subset of patient characteristics having no known catheter location and the new complementary CPM matrix. The method of claim 11.

14. The method of claim 12, wherein each catheter location includes the location of each electrode of the catheter.

15. The method of claim 12, wherein both the subset of patient characteristics having a known catheter location and the subset of patient characteristics having no known catheter location include current signals recorded at each electrode of the catheter. Claim 16 The method of claim 12, wherein the patient characteristics of the subset of patient characteristics having the known catheter location include data from procedures in the history. Claim 17 The method of claim 11, wherein the learning system is trained using at least one of classification, regression, and clustering algorithms. Claim 18 The method of claim 14, wherein the learning system comprises a neural network.

Images