Computer-implemented electrocardiogram data processing method

Abstract

This specification describes a method for visually displaying electrocardiogram data in a compressed manner on a display screen, where rhythm information is visible, and a method for classifying intervals of electrocardiogram data.

Description

[0001] Divisional Application Statement

[0002] This application is a divisional application of a Chinese patent application with the invention name "Computer-Implemented Electrocardiogram Data Processing Method", application number 201780068050.5, which entered the Chinese national phase on April 30, 2019. The parent PCT international application has an international filing date of November 3, 2017 and an international application number of PCT / CA2017 / 051310. Background Art

[0003] Electrocardiography (ECG) is a process of recording the electrical activity of the heart over a period of time, typically accomplished by applying electrodes to a patient's skin. Computerized electrocardiography generates electrocardiogram data that has the amplitude of the recorded electrical activity varying over time. The electrocardiogram data can include absolute time values associated with each amplitude. Optionally, the time reference can be implicit. For example, if the time elapsed between successive amplitudes is constant and known, it may not be necessary to store absolute time values for each amplitude in order to be able to reconstruct and display the electrocardiogram.

[0004] For decades, electrocardiograms have been printed directly on paper in the form of two-dimensional graphs having amplitude plotted along the Y-axis and time plotted along the X-axis. Figure 1A Shows a given time period of an electrocardiogram presented in this way, including the start of a first heartbeat 10 and a second heartbeat 10'. A continuous sequence of normal heartbeats is called normal sinus rhythm, which has recognizable reference features. The accurate expression of these recognizable features can vary widely depending on the heartbeat, the person, and the recording method, but they generally appear in one form or another. These recognizable features include what are called "peaks" in the art, namely the P peak, Q peak, R peak, S peak, and T peak.

[0005] In the case where the normal resting heart rate is typically between 60 and 100 beats per minute, it can be understood from Figure 1A that electrocardiograms spanning any significant amount of time can require a large amount of paper.

[0006] In recent years, it has become increasingly common to store electrocardiogram data on a computer-readable storage medium and display the electrocardiogram data on a computer display screen rather than printing it on paper. Typically, only a relatively small portion of the electrocardiogram data is displayed at any given time, and the user can navigate the electrocardiogram data by providing inputs that can be used as commands to change the displayed portion of the electrocardiogram data.

[0007] There is a need to display electrocardiogram (ECG) data on a display screen for a longer period of time, such as ECG data for several minutes. A new method of displaying ECG data published in recent years has met this need to some extent. This new method involves compressing the displayed data by plotting the amplitude using a color scale instead of the Y-axis of a two-dimensional graph. Thus, the display of ECG data can be compressed onto a single axis. In addition, algorithms for automatically identifying individual heartbeats in ECG data have been developed. Therefore, single-axis graphs of the corresponding heartbeats can be displayed. By horizontally positioning and displaying these single-axis graphs side by side, a large number of heartbeats can be displayed simultaneously on the display screen.

[0008] Although the previous technologies are satisfactory to some extent, there is still room for improvement in the display, navigation, annotation, and computer processing of ECG data. Summary of the Invention

[0009] In one aspect, there is provided a method of displaying electrocardiogram (ECG) data on a computer display screen, the ECG data having a set of amplitudes that vary over time and representing consecutive heartbeats. The method includes: the computer displaying the ECG data within an elongated rectangular portion on the display screen, wherein a plurality of single-axis graphs extend horizontally with respect to the length of the rectangular portion and are positioned adjacent to each other along the length of the rectangular portion, each graph in the sequence representing the corresponding, consecutive, ECG data for a period of time, the amplitudes of the ECG data being displayed in corresponding colors or hues as a function of a color amplitude scale and / or a hue amplitude scale, corresponding time values being plotted along a single axis, wherein each of the corresponding heartbeats represented by the corresponding graphs among the plurality of graphs has a common alignment reference feature that is horizontally aligned at a common horizontal reference coordinate of the rectangular portion and ends at a common rhythm reference feature, the horizontal positions of the ends of the respective graphs varying from graph to graph as a function of the heart rhythm; and the computer further displaying a remaining area that extends horizontally between the end of the graph and the corresponding edge of the rectangular portion in a manner that contrasts with the color scale and / or hue scale of the graph.

[0010] In another aspect, correspondingly, there is provided a computer-implemented method for processing electrocardiogram (ECG) data. The method includes: the computer displaying ECG data on a display screen, the displayed ECG data having a plurality of amplitudes that vary over time and representing consecutive heartbeats; the computer classifying intervals of the ECG data, including: receiving a first user input and identifying a first time coordinate of the ECG data based on the first user input; receiving a second user input and identifying a second time coordinate of the ECG data based on the second user input; defining an interval of the ECG data as extending between the first time coordinate and the second time coordinate; receiving a third user input and associating a category with the defined interval based on the third user input.

[0011] After reading this invention, many further features and combinations thereof for improving this invention will be obvious to those skilled in the art.

[0012] It can be understood that the term "computer" used herein should not be construed in a restrictive manner. In a broad sense, it generally refers to a combination of one or more processing units in some form and a memory system accessible to the processing unit in some form. A computer can be a personal computer, a smart phone, a tablet computer, a home computer, etc.

[0013] It can be understood that various functions of a computer, or more specifically, various functions of a processing unit or a memory controller, can be performed by hardware, software, or a combination of both. For example, the hardware can include logic gates as part of a processor silicon chip. The software can be in the form of data, such as computer-readable instructions stored in a memory system. For a computer, a processing unit, a memory controller, or a processor chip, the term "configured to" relates to the presence of hardware, software, or a combination of hardware and software that can be run to perform the relevant functions. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In these figures:

[0015] Figure 1A is a part of a two-dimensional graph representing electrocardiogram data, where the time coordinate of the electrocardiogram data is plotted on the X-axis and the corresponding amplitude coordinate is plotted on the Y-axis;

[0016] Figure 1B is an example of a one-dimensional (single-axis) graph representing electrocardiogram data, which uses a color scale and / or a tone scale instead of Figure 1B the Y-axis, and the width of the single-axis graph is enlarged for illustration;

[0017] Figure 2 is a rectangular part of a display screen according to an embodiment, where multiple single-axis graphs as shown Figure 1A are placed continuously side by side;

[0018] Figure 3 shows an example of a graphical representation of Figure 2 applied to several superimposed rows on a display screen;

[0019] Figure 4 shows an example of a graphical representation of Figure 2 applied to the upper part of a display screen;

[0020] Figure 5 shows an example of the classification of intervals as atrial fibrillation; and

[0021] Figure 6Several intervals automatically classified as noise are shown, and one of the noise intervals is selected. Detailed Description

[0022] Referring to Figure 2 , an example of electrocardiogram data in graphical representation on a rectangular portion 12 of a display screen is shown. In the present embodiment, the individual uniaxial graphs 14 are displayed adjacent to each other along the length 16 of the rectangular portion 12. The individual uniaxial graphs 14 are aligned transversely to the length 16 of the rectangular portion 12, each graph representing a plurality of amplitudes using a color scale and / or a tone scale, and the associated time values as spatial coordinates along a single axis. The rectangular portion 12 of the display screen generally exceeds the duration of a single graph, and the remaining area 18 extends transversely between the end 20 of the graph 14 and the corresponding edge 22 of the rectangular portion 12 in a manner that visually contrasts with the color scale and / or tone scale used in the graph 14. As will be described in detail below, this remaining area 18, in combination with the method of graphically representing individual heartbeats of the electrocardiogram data, allows a technician to display rhythm (heartbeat frequency) information in addition to the amplitudes of a given heartbeat. The exact color scale and / or tone scale can be selected according to a particular application or alternatively adjusted by the user. It has been found that in at least some embodiments, it is advantageous for the color scale and / or tone scale to have a difference in the thickness interval 24 of the T wave. The displayed electrocardiogram data can be obtained in advance by any suitable means, including using a stationary electrocardiography recording device, as well as a wearable electrocardiography recording device, such as a traditional Holter monitor and more recent technologies such as the device. The electrocardiogram data can be preprocessed before being displayed, such as automatic identification of heartbeats. When a wearable monitor such as the device is designed to record electrocardiogram data for a long period of time (e.g., more than a few minutes, typically more than a few hours, preferably several days or longer), it may pose particular challenges. Such a large amount of resulting data cannot be processed in a traditional manner. As will be seen below, a computer can be utilized to provide software that allows processing of such a large amount of data and allows the user to quickly navigate between the data at relatively large time increments (such as minutes or even hours).

[0023] In the described embodiment, the automatic heartbeat recognition is implemented based on a QRS complex detection algorithm titled "An Efficient R-Peak Detection Based on New Nonlinear Transformation and First-Order Gaussian Differentiator" by P. Kathirvel et al., which was published online on October 12, 2011 in Volume 2, Number 4, December 2011, pages 408 - 425 of "Cardiovascular Engineering and Technology". This QRS complex detection algorithm generally includes the following steps:

[0024] · Band-pass filter the ECG signal between 0.5 and 40 Hz;

[0025] · Nonlinear transformation, including taking the power of 2 of each sample while preserving its sign;

[0026] · Estimate the amplitude using a regression low-pass filter;

[0027] · Add high-frequency components scaled by the amplitude estimate;

[0028] · Determine the number of zero-crossing events; and

[0029] · Detect the R peak by applying a threshold to the zero-crossing events.

[0030] Referring to Figure 1B , a single uniaxial graph 26 representing a portion of the electrocardiogram data with normal sinus rhythm is shown. Figure 1B The width 28 of the shown single uniaxial graph 26 is enlarged for illustration. In practice, there is a motivation to minimize its width 28 as much as possible, ideally to a single pixel, and position the individual graphs adjacent to each other as in the Figure 2 graphical representation shown, which allows more data to be compressed into the area of the rectangular portion 12 of the display screen. In practice, it may still be necessary to use more than one pixel to represent the width 28 of each graph. On the other hand, some graphs can be skipped to further compress the data without displaying all heartbeats. For example, only one graph is shown for each given number of adjacent heartbeats, such as one out of every three or one out of every five. In the present embodiment, the time coordinate across the uniaxial extends from the upper part 30 to the lower part 32 of the graph 26. As can be seen in the present embodiment, the graph 26 includes not only the electrocardiogram data corresponding to the respective heartbeats 10 (the first P peak, Q peak, R peak, S peak, and T peak extending from the upper part to the lower part of the graph), but also further extends to the common rhythm reference feature 34.

[0031] The rhythm reference feature 34 can vary for different embodiments, but for a given instance of multiple graphs 14 that are simultaneously displayed adjacent to each other as Figure 2 shown, the rhythm reference features 34 of these graphs 14 are generally the same (common) to visually display rhythm information to the user. For example, the common rhythm reference feature 34 can be associated with the Q peak, R peak, or S peak of the next heartbeat. The common rhythm reference feature 34 can be automatically recognized by a computer through an algorithm, as will be elaborated below. In Figure 1B the embodiment shown, the common rhythm reference feature 34 is selected as the start of the R peak of the next heartbeat 10'. Correspondingly, the graph 26 can be seen as further extending along the P peak and Q peak of the next heartbeat 10′, and thus including the duration of the pause 36 between the T peak of the corresponding heartbeat 10 and the P peak of the next heartbeat 10′. Correspondingly, the uniaxial graph 26 ends at the common rhythm reference feature 34, and the length 38 of the graph 26 can vary, for example, according to the heart rate and the duration of the pause 36.

[0032] To construct Figure 2 the graphical representation shown, another common reference feature of each heartbeat is detected and used to horizontally align the graphs 14 with each other. This latter common reference feature will be referred to as the common alignment reference feature 40. The common alignment reference feature 40 can vary for different embodiments, but for a given instance of multiple graphs 14 that are simultaneously displayed adjacent to each other, the common alignment reference features 40 of these graphs 14 are generally the same (common), and a reference feature identical to the common rhythm reference feature 34 can be selected to visually display rhythm information to the user. By positioning the detected common alignment reference feature 40 at the common horizontal alignment coordinate 42 of the rectangular portion 12, each graph 14 in the rectangular portion 12 of the display screen can be horizontally aligned with the other graphs 14.

[0033] In as Figure 2In the illustrated embodiment, the common alignment reference feature 40 is the onset of the R-wave of the corresponding heartbeat, and the time period between the onset of the R-wave of the corresponding heartbeat and the onset of the R-wave of the next heartbeat characterizes the heart rate; along the corresponding edge 22 of the rectangular portion 12, this time period is graphically displayed in the form of the length of a part of a uniaxial graph, extending from the common transverse alignment coordinate 42 of the rectangular portion 12 to the end 20 of the graph (this end 20 is associated with the common rhythm reference feature 34). Accordingly, the transverse coordinate of the end of a given graph 14 in the graph 14 can be associated with the instantaneous heart rate between the corresponding heartbeat and the next heartbeat, or conversely, with the amount of time between the corresponding heartbeat and the next heartbeat. By displaying the portion of the rectangular portion 12 immediately following the end 20 in a color and / or hue that forms a sharp contrast with the color scale and / or hue scale of the graph 14 (more specifically, with the typical color scale and / or hue scale of the common rhythm reference feature 34), the transverse coordinate of the end 20 of the graph 14 can be made easily visible to the user. For example, in Figure 2 , the onset of the R-wave typically has an amplitude represented by white on the color scale and / or hue scale, while the remaining area 18 of the rectangular portion 12 beyond the end 20 of the graph 14 is all black. More generally, the remaining area 18 can be represented in a color and / or hue that forms a contrast with the color scale and / or hue scale used to display the amplitude of the electrocardiogram data in the graph 14. In one particular embodiment set forth, "warm" colors directed through yellow and red to white are used to represent increasing positive amplitudes, while "cool" colors leading to dark blue are used to represent negative amplitudes.

[0034] Still referring to Figure 2 , the graphs 14 in this embodiment are positioned adjacent to each other along the length 16 of the rectangular portion 12. Each graph 14 in the sequence represents corresponding, continuous electrocardiogram data over a period of time, where the amplitudes are displayed in corresponding colors or hues determined according to the color and / or hue amplitude scale. The time values associated with the amplitudes are then plotted along a single axis. Each heartbeat represented by a corresponding graph in the graph 14 has a common alignment reference feature 40 that is transversely aligned with the common transverse reference coordinate 42 of the rectangular portion 12. The graph 14 ends at the common rhythm reference feature 34. The transverse positions of the ends 20 of the individual graphs 14 vary from one graph 14 to another according to the heart rhythm, and in combination with the contrasting remaining area 18 of the rectangular portion 12, form an edge 44 that extends along the corresponding edge of the rectangular portion 12. The thickness of this edge 44 along the length 16 of the rectangular portion varies according to the heart rhythm.

[0035] The horizontal coordinate of the edge 22 of the rectangular portion 12 corresponding to the edge of the edge 44 can be associated with a given time interval 46 starting from the common alignment reference feature 40, so as to be able to have a given and invariant spatial distance to the common horizontal alignment coordinate 42. This time interval 46 can be associated with the size of the rectangular portion 12 of the display screen and the maximum time interval allowed by the display configuration. This time interval 46 is usually selected in such a way as to be sufficient to include the maximum possible time interval that can be expected conventionally between heartbeats. In this embodiment, according to the start of the R peak, that is, the common rhythm reference feature 34, the time interval 46 is selected to be 2 seconds. In other embodiments, for example, the time interval 46 can be greater than 1.4 seconds, or greater than 1.8 seconds. In other embodiments, for example, the time interval 46 can also be adjusted according to user input. The scale of the time interval 46 can be displayed adjacent to the rectangular portion 12 for the convenience of user analysis, as Figure 4 shown. For example, the time interval 46 is presented with a time scale 48a in milliseconds on the right hand side of the display screen. Conversely, the time interval 46 can also represent the heart rate, and the heart rate scale 48b can be displayed adjacent to the rectangular portion 12. For example, in Figure 4 it, the heart rate scale 48b in beats per minute (BPM) is presented on the left hand side of the display screen.

[0036] In a display configuration such as Figure 3 shown, there may be a motivation to reduce the horizontal width 50 of the rectangular portion 12, because this can allow more rows of electrocardiogram data to be included, so that more electrocardiogram data can be displayed simultaneously on a given screen within a given time. For this purpose, in an alternative embodiment, a shorter time interval 46 can be preferably used, or a logarithmic scale can be used instead of the linear scale used in the illustrated drawings. In an embodiment using a logarithmic scale, for example, the values related to the logarithmic scale can be displayed adjacent to the display content.

[0037] In electrocardiogram data, there may be an event where the pause between a first heartbeat 10 and a second heartbeat 10' exceeds the time interval 46. This can be handled in various ways. In the illustrated embodiment, if the pause exceeds the time interval 46, the graph of the first heartbeat 10 continues until the edge of the rectangular portion 12, forming a horizontal line that extends completely through the rectangular portion 12, making the event easily perceptible by a trained technician querying the displayed electrocardiogram data. Additionally, if the pause lasts for more units of the time interval 46, subsequent graphs corresponding to the pauses of the first heartbeat 10 for different time interval units can be displayed in the electrocardiogram data. This can widen the thickness of the horizontal line that extends completely through the rectangular portion 12 in the direction of the length 16 of the rectangular portion 12 and can provide a very intuitive indication of the pause length to a skilled technician or doctor querying the displayed electrocardiogram data. When the second (subsequent) heartbeat 10' arrives, it can be displayed normally, and its common alignment reference feature 40 is aligned with the common horizontal alignment coordinate 42.

[0038] That is, the electrocardiogram data can be displayed according to the following conditions:

[0039] · If a heartbeat 10 is associated with a time period corresponding to a given graph, the amplitudes across the P, Q, R, S, and T peaks of the associated heartbeat are displayed in the given graph;

[0040] · If a heartbeat 10 is associated with a time period corresponding to a given graph and the next heartbeat 10' is further associated with the next time period of the electrocardiogram data, the amplitude from the T peak of the associated heartbeat to the common rhythm reference feature 34 of the next heartbeat 10' is displayed in the given graph, including at least one P peak of the next heartbeat 10';

[0041] · If a heartbeat 10 is associated with a time period corresponding to a given graph and the next heartbeat 10' is not associated with the next time period of the electrocardiogram data, the amplitude from the T peak of the associated heartbeat to the maximum duration of the given graph is displayed in the given graph (the maximum duration of the given graph corresponds to the spatial coordinate of the corresponding edge 22 of the rectangular portion 12); and

[0042] · If a heartbeat 10 is not associated with the corresponding time period of a given graph, the amplitude across the entire duration of the time period, leading to the maximum duration, is displayed in the given graph.

[0043] In Figure 3In the display configuration 52 shown, the time coordinate of the electrocardiogram data can be advanced from left to right and then continue on the left side of the lower rectangular portion 12, and so on. In addition, the application responsible for displaying the electrocardiogram data can include in its user interface means for enabling the user to conveniently navigate the electrocardiogram data (moving along the electrocardiogram data, continuously displaying new portions of the electrocardiogram data while hiding the previously displayed portions of the electrocardiogram data). In this embodiment, using the down arrow key on the keyboard allows the user to navigate the rectangular portions 12 of the lines forming the display content by moving one line "down" at a time. For example, keys such as the page down and page up keys allow navigating an entire number of display lines at once (5 lines in this example). As Figure 3 shown, in this embodiment, the "Go To" area 54 of the graphical interface is displayed in the upper left corner of the screen. By activating the "Go To" area 54 of the graphical interface, the user can access the calendar and select a specific date in the calendar corresponding to the portion of the electrocardiogram data that the user wishes to see displayed. For example, other keys such as the home key or the end key can be used as user input to allow the user to directly navigate to the start time period or the final time period of the electrocardiogram data.

[0044] Figure 4 The display configuration 52' of another embodiment is shown. In Figure 4 it, the compressed electrocardiogram data is displayed in the first rectangular portion 12a of the display screen as shown above, and a portion of the electrocardiogram data displayed in the first rectangular portion 12a is displayed in the second rectangular portion 12b of the display screen using a "classical" two-dimensional representation. In this embodiment, the application for displaying the electrocardiogram data can be further configured to be capable of receiving user input representing the spatial coordinates corresponding to at least one uniaxial graph displayed in the first rectangular portion 12a and displaying a two-dimensional representation of the portion of the electrocardiogram data selected by the user input in the second rectangular portion 12b. The user input can be received by left-clicking, for example, on the corresponding portion of the first rectangular portion 12a or the corresponding portion on the screen. A visualization indicator 56 can be displayed on the first rectangular portion 12a to indicate which portion of the electrocardiogram data is being displayed in the second rectangular portion 12b of the display screen. In this embodiment, the visualization indicator is a shaded or highlighted vertical column. If two screens are used, Figure 4 the display content can be displayed on the first display screen, Figure 3The displayed content can be shown on the second display screen. The two display screens can be linked in such a way that navigating to a new position in any of the rectangular sections automatically triggers an adjustment of the position of the markers on the other graphical representation. For example, visual indicators can also be presented in the corresponding areas of the second display screen. In this embodiment, a horizontal line is used as the cursor, which can be moved along the length of the two-dimensional graph by user input. In this embodiment, the exact time coordinates corresponding to the cursor are displayed in the rectangular box 58 in the first rectangular section 12a and, as Figure 3 is also shown in the upper left part of the screen. Moving the cursor along the length of the two-dimensional graph can also be used to navigate the data displayed in the first rectangular section 12a.

[0045] Now refer to Figure 5 and further explore the function of classifying the interval 60 in the electrocardiogram data. Two variations will be considered. The first variation is the manual classification of the interval, and the second variation is the automatic classification of the interval.

[0046] As Figure 5 shown, a computer can be used to classify the interval 60 in the displayed electrocardiogram data. This method can be used to classify the interval 60 of electrocardiogram data displayed as a series of uniaxial graphs, for example, for the upper rectangular section 12a of the display screen as Figure 5 shown. Alternatively, this method can also be used to classify the interval 60 of electrocardiogram data displayed as a two-dimensional graph, for example, as shown in the bottom rectangular section 12b of the display screen of Figure 5 .

[0047] The method may include a computer receiving a first user input and identifying a first time coordinate 62a of electrocardiogram data based on the first user input. For example, the first user input may be received by the user right - clicking or touching a given portion of the screen to indicate spatial coordinates associated with a graphical representation and a given time coordinate. As shown in the above example, the graphical feedback may be displayed in the form of a visual indicator. The method then includes the computer receiving a second user input and identifying a second time coordinate 62b of the electrocardiogram data based on the second user input. The computer may then define an interval 60 corresponding to a specific time interval of the electrocardiogram data, the interval 60 extending from the first time coordinate 62a to the second time coordinate 62b. The definition of the interval 60 may be stored by the computer in a computer - readable storage medium. The computer then may receive a third user input to associate a category to the defined interval 60. More specifically, the third user input is used to assign a category to the defined interval 60. For example, the user may right - click between the two visual indicators to trigger a window 64 in which the user may select a given category from among a plurality of possible categories 66. In the illustrated embodiment, there are four possible categories 66: noise, normal sinus rhythm, atrial fibrillation (paroxysmal (PAF) or chronic (CAF)), and atrial flutter, and the user selects atrial fibrillation. The category may be stored as data associated with the definition of the interval 60.

[0048] In this particular embodiment, the user may automatically adjust the selected interval 60 or the start of the "From" region by indicating the corresponding side outside the interval 60 in the displayed electrocardiogram data. For example, the "From" region may be automatically moved by clicking on the left - hand side of the previously selected "From" region, and the same applies to the "To" region. If the "From" region or the "To" region is to be adjusted to a position within the selected interval 60, then as Figure 5 shown, the corresponding "Update selection start" or "Update selection end" function may be selected from a pop - up window 64.

[0049] In another variation, an application may be used to automatically detect the category of electrocardiogram data, automatically define the interval 60, and automatically assign the corresponding category 66 to the corresponding interval 60. For example, in the illustrated embodiment, an automatic noise identification algorithm is also performed on the electrocardiogram data before the electrocardiogram data is displayed.

[0050] Alternatively, automatic interval identification can be based on deep learning and / or heart rate variability algorithms. For example, deep learning algorithms can be trained using a dataset of previously manually annotated ECG records. Such deep learning algorithms can be used to automatically identify atrial fibrillation, atrial flutter, ectopic beats, atrioventricular block, bundle branch block, and / or any other suitable form of arrhythmia. For atrial fibrillation and atrial flutter, more classical heart rate variability algorithms can be used to refine the deep learning identification to more accurately identify the start and end of each interval or segment. The proposed two-dimensional color scale method can not only provide a highly compressed and expressive ECG representation for technicians in their analysis tasks, but can also be more suitable for deep learning algorithms than traditional two-dimensional graphical data representations.

[0051] An example of a noise identification algorithm that is executed after the QRS complex detection algorithm can be summarized as follows:

[0052] · High-pass filter the resulting signal to remove the typical low-frequency components in normal P waves and T waves;

[0053] · For each detected R peak, define a masking window to enclose the QRS complex (i.e., from 50 ms before the R peak to 100 ms after the R peak);

[0054] · For each R-R interval, the algorithm calculates the signal-to-noise ratio and compares the result with a threshold to determine whether a given R-R interval is identified as noise, where the threshold is set to a value at which the artifact content of the R-R interval will become too high for a technician to visually identify the PQRST complex when querying the displayed electrocardiogram data.

[0055] Interval 60 in the electrocardiogram data corresponds to a specific time interval in which signals classified as noise can be automatically defined by a computer, and the categories corresponding to these intervals can be automatically stored in memory in association with the definitions of these intervals.

[0056] In Figure 5 the illustrated embodiment, a visual indicator 68 in the form of a horizontal bar extending along the edge 22 of the upper rectangular portion 12a is used to indicate that a given interval 60 has been classified. For example, the category can be visually represented to the user by assigning a corresponding color to the horizontal bar. Figure 3 All of the electrocardiogram data shown in

[0057] Figure 6Shows the intervals 60' in the electrocardiogram data that are automatically classified as noise by the software. In the illustrated embodiment, the application provides a function for automatic interval selection: for example, when the user indicates a given part of the display screen corresponding to the classified interval 60, the interval 60 can be automatically selected, a visual indicator 70 can be used to mark the start and end of the interval, and a pop-up window 64 can be triggered to allow the user to remove the classification of the interval 60 or change the classification of the interval 60. For example, the automatic selection of the interval 60 can be triggered by clicking on the corresponding horizontal bar 72.

[0058] In the illustrated embodiment, the percentage of all classified electrocardiogram data (including both displayed and non-displayed parts) is tracked. Additionally, in Figure 3 the display, a progress bar 74 is used in the upper rectangular part of the screen to visually represent the percentage of all classified electrocardiogram data. Correspondingly, in Figure 3 it shows that 41% of the entire electrocardiogram data has been classified, which indicates to the user that the non-displayed part of the electrocardiogram data is currently unclassified.

[0059] In the illustrated embodiment, the application guides a skilled technician to perform the general workflow of classification, including guiding the skilled technician to assign a category to each part of the entire electrocardiogram data so as to reach 100% on the progress bar 74. Then, the application generates a report (not shown), which summarizes the main features of a given electrocardiogram data set and presents some excerpts from the electrocardiogram data in the form of two-dimensional graphs, and these excerpts are typical representatives of the given electrocardiogram data set. In the illustrated embodiment, the user is only allowed to use the report generation function when 100% progress is reached. Thus, as long as some parts of the electrocardiogram data are unclassified, the report cannot be finalized. The generation of the report depends on the complete classification of the electrocardiogram data.

[0060] The user can start by evaluating the first 24 hours using, for example, the detailed view as shown in Figure 4 and then quickly view the remaining days using the main view as shown in Figure 3 .

[0061] In this particular embodiment, the electrocardiogram data can be encoded with patient events (PEV). For example, the patient event marker can be associated with the time coordinates of the electrocardiogram data. For example, during the recording of the electrocardiogram data, the addition of a patient event marker can be triggered by receiving user input through a wearable monitor. The device has buttons to achieve this purpose. The application can provide visual markers on the displayed electrocardiogram data for each patient event, or can be adjusted to provide information about the patient event in another way. For example, the user can be instructed to check for the presence of a patient event while linking it to notes in the patient log and / or abnormalities in the records.

[0062] The application software provides a feature of addable deduction to avoid small gaps inadvertently remaining between the classified intervals in the electrocardiogram data. This addable deduction feature will be described below.

[0063] The addable deduction feature employs a predetermined time interval 76. When an interval in the "from" or "to" region is identified from adjacent intervals within the predetermined time interval 76, the "from" or "to" region is automatically adjusted to exactly coincide with the corresponding end of the adjacent interval 60. The predetermined time interval 76 can be defined, for example, in units of time or in units of heartbeats. In this embodiment, the predetermined time interval 76 is defined in units of heartbeats. When the "from" or "to" selection is made by selecting coordinates in a compressed graph, such as Figure 5 the upper rectangular portion 12a shown, the predetermined time interval 76 can be set to a given number of heartbeats, such as 8 heartbeats, 10 heartbeats, or 20 heartbeats. When the "from" or "to" selection is made by selecting coordinates in a two-dimensional graph, such as Figure 5 the lower rectangular portion 12b shown, and the software can infer that the user expects additional precision, the predetermined time interval 76 can be smaller, such as a single heartbeat. Correspondingly, if the user clicks on a region corresponding to the predetermined time interval 76 in the compressed graph, for example, 10 heartbeats of the adjacent interval 60, the corresponding ends of the adjacent interval 60 can be automatically selected instead of the actual coordinates indicated by the user. This function can help avoid the occurrence of small gaps between the intervals 60, which would otherwise be difficult to detect on the main view as shown in Figure 3 the main view shown.

[0064] Correspondingly, when classifying another interval before classifying a given interval 60, if one or both of the first time coordinate 62a and the second time coordinate 62b in the given interval 60 are adjacent to the first time coordinate and the second time coordinate of another interval within the buffer time interval, then the corresponding one or both of the first time coordinate and the second time coordinate are identified as the corresponding time coordinates of the other interval.

[0065] The feature of meta-zone classification can also be provided. A meta-zone can be defined as an interval extending between "from" (start) and "to" (end), which spans at least two sub-intervals that either have different categories or have at least one category and an unclassified part of the electrocardiogram data. For example, when it is determined that a meta-zone has been defined, additional functions can be used to present different pop-up menus to the user. To give a few examples, the menu can allow the user to assign a given category to all unclassified parts, remove the classification of all intervals with a given category, or remove all classifications within the meta-zone entirely.

[0066] The software can have additional functions beyond the above-mentioned functions. For example, additional automatic category recognition functions can be provided, such as premature atrial contraction (PAC) detection, premature ventricular contraction (PVC) detection, and PVC morphology classification. In the described embodiments, these additional algorithms are not used to automatically assign a category to the corresponding interval of the electrocardiogram data, but are displayed as indicators aimed at attracting the attention of skilled users of the software.

[0067] In the described embodiments, when a technician labels a new ECG interval as a normal interval, the PAC detection algorithm will be executed. PACs are physiologically not applicable to atrial fibrillation or flutter intervals, and based on the above noise recognition process, intervals marked as noise will also be excluded. The technician can enable automatic PAC detection and use a slider widget through the graphical user interface to define the sensitivity of PAC detection. An example of a PAC detection algorithm that can be enabled can be summarized as follows:

[0068] · Based on the normal intervals classified by the user, a list of detected QRS complexes is first defined;

[0069] · Apply a sliding window from the first QRS complex to the last QRS complex in the normal interval as the R-R history guiding each QRS complex;

[0070] · If the R-R interval of the last QRS complex in the sliding history window is less than the threshold determined based on the R-R history guiding that QRS complex, the last QRS complex in the sliding history window will be marked as a PAC;

[0071] · The threshold is parameterized so that the user can adjust the threshold using the slider widget;

[0072] ·Detected PACs can be marked with the blue letter S in the upper area of the two-dimensional graphical view, and short vertical red lines can also be added at the bottom of the compressed graphical representation, as Figure 4 shown.

[0073] In the described embodiment, when a technician classifies a new ECG interval as normal sinus rhythm, atrial fibrillation, or flutter, a premature ventricular contraction (PVC) detection algorithm is executed. Based on the above noise identification process, only intervals marked as noise are excluded. The technician can enable automatic PVC detection and use two independent slider widgets in the graphical user interface to define the importance of complex premature beats and the sensitivity of PVC detection. When enabled, the PVC detection algorithm can be summarized as follows:

[0074] ·Based on the (one or more) intervals delimited by the user, a list of detected QRS complexes is first defined;

[0075] ·For each QRS complex in the list, the following five factors are calculated, where each factor quantifies a specific characteristic of PVC:

[0076] Complex premature beats quantified in a manner similar to the PAC algorithm;

[0077] QRS complex width (Q-S interval), which is estimated based on the QRS complex morphology and normalized to the median width of all QRSs falling within the selected (one or more) intervals. PVCs typically have a larger QRS complex duration;

[0078] QRS complex amplitude, which is estimated using a subset of the QRS detection algorithm itself and also normalized using the median amplitude of all QRSs falling within the selected (one or more) intervals;

[0079] QRS complex maximum (usually the amplitude of the positive R peak), which is also normalized using the median maximum of all QRSs falling within the selected (one or more) intervals;

[0080] QRS complex minimum (usually the amplitude of the negative S peak), which is also normalized using the median minimum of all QRSs falling within the selected (one or more) intervals;

[0081] If the premature beat of the QRS complex exceeds a threshold proportional to the value of the user-defined premature beat slider widget and the weighted sum of the other four factors is higher than the value of the user-defined PVC sensitivity slider widget, then the QRS complex is marked as a PVC; and

[0082] The detected PVCs can be labeled with the brown letter V in the upper area of the two-dimensional graphic view, and short vertical red lines can also be added at the bottom of the compressed graphic, just below the space reserved for the PVC labeling.

[0083] In the described embodiment, after a new PVC is detected, the PVC morphology classification algorithm is executed. A technician can define the sensitivity of the classification using the slider widget in the graphical user interface. The PVC morphology classification algorithm can be summarized as follows:

[0084] · Compare each detected but unclassified PVC morphology with all previously classified PVC morphologies; this comparison is made by calculating the distance between two PVCs; if the measured distance value between one PVC and another is higher than the adjustable threshold defined by the value of the morphological classification sensitivity slider widget, then the two PVCs are considered similar and thus they belong to the same morphological family;

[0085] · When an unclassified PVC does not meet the classification criteria of any previously classified PVC morphology, a new morphological family is created with this PVC as the first and only morphology; and

[0086] · The PVC morphology can be labeled on the two-dimensional graphic view after the letter V with a family ID (e.g., a numerical value between 1 and 32).

[0087] From the above, it can be understood that the technical tool provided herein allows the user to visualize a large number of QRS complexes - tens of thousands - representing several hours of an ECG recording in a static view. The number of QRS complexes and the ECG duration that can be displayed in a single screen view vary depending on the recorded heart rate and the screen resolution.

[0088] From the above, it can be understood that the classical ECG recorded signal can first be divided into heartbeat segments, where one segment represents an R-R interval. In this embodiment, a segment starts 400 milliseconds before a given heartbeat and ends at the detection point of the next heartbeat. For example, in a color scale and / or hue scale, the isoelectric reference (0mv) can be represented by yellowish-green, where positive values go from yellowish-green to red to white (warm colors), and negative values go from yellowish-green to blue to black (cold colors), although this is just one possible example. Figure 4An example of a detailed view is shown, in which the compressed color-coded graph in the upper rectangular portion corresponds to approximately 25 minutes of continuous ECG data, while the lower two-dimensional graph represents a 12-second ECG band. In this view, the user can navigate the ECG data recording by clicking the left mouse button to move the yellow vertical cursor in any of the three views. The cursor position will be updated to the new click position, and the other two views will recenter around the selected time point. Each view displays a navigation cursor, the position of which indicates the same time point in the ECG recording. At the bottom of the detailed view, a navigation toolbar allows the user to quickly synchronize the three ECG views to a specific ECG event or arrhythmia.

[0089] The various functions described above can be provided in the form of one or more computer program products (applications) stored in a computer-readable storage medium.

[0090] It can be understood that the embodiments described and illustrated above are only examples. For example, in an alternative embodiment, the elongated rectangular portion may be vertical instead of horizontal, and the comparison edges may be shown above the graph instead of below. The scope of protection of this application shall be subject to the appended claims.

Claims

1. A computer-implemented method for processing electrocardiogram data having multiple amplitudes that vary over time and representing consecutive heartbeats, the method comprising: The computer displays the electrocardiogram data on a display screen, the displayed electrocardiogram data including a plurality of uniaxial graphs; The computer classifies a first interval of the electrocardiogram data, including: Receiving a first user input and identifying a first time coordinate of the electrocardiogram data based on the first user input; Receiving a second user input and identifying a second time coordinate of the electrocardiogram data based on the second user input; Defining the first interval of the electrocardiogram data as extending between the first time coordinate and the second time coordinate; and Receiving a third user input indicating a class selected by the user on the computer, and the computer assigning the class to the first interval according to the third user input.

2. The method according to claim 1, wherein The first user input and the second user input are spatial coordinates of the displayed electrocardiogram data.

3. The method according to claim 2, wherein A second interval is classified before the step of classifying the first interval, wherein when at least one of the first time coordinate of the first user input and the second time coordinate of the second user input is adjacent to a corresponding one of the first time coordinate and the second time coordinate of the second interval within a buffer time interval, at least one of the first time coordinate and the second time coordinate of the first interval is identified as the corresponding time coordinate of the second interval.

4. The method according to claim 3, wherein The buffer time interval is defined according to the heart rate.

5. The method according to claim 1, wherein It further includes repeating the steps from classifying a first interval of the electrocardiogram data to classifying other intervals of the electrocardiogram data, and each different interval is assigned a corresponding class selected from a group of classes.

6. The method according to claim 5, wherein It further includes generating a report that provides data on class attribution, wherein the generation of the report depends on the condition that all parts of the electrocardiogram data have attributed classes.

7. The method according to claim 5, wherein It further includes tracking the percentage of the electrocardiogram data that has been assigned a class and displaying a visual marker of the percentage on the display screen.

8. The method according to claim 1, characterized in that The step of displaying the electrocardiogram data includes displaying the electrocardiogram data within at least one elongated rectangular portion of the display screen, wherein the plurality of uniaxial graphs extend laterally with respect to the length of the at least one rectangular portion and are positioned adjacent to each other side by side along the length of the at least one rectangular portion, and each graph in the sequence represents corresponding, consecutive, electrocardiogram data for a period of time, wherein the corresponding time values are plotted as corresponding spatial coordinates along a single axis.

9. The method according to claim 1, characterized in that It further includes changing the displayed portion of the electrocardiogram data based on user input to navigate to other parts of the electrocardiogram data.

10. The method according to claim 1, wherein It further includes displaying a first visual indicator marking the first time coordinate on the display screen and displaying a second visual indicator marking the second time coordinate on the display screen.

11. The method according to claim 8, wherein The amplitudes of the corresponding, consecutive, electrocardiogram data for a period of time associated with each graph are displayed in corresponding colors and / or hues as a function of a color amplitude scale and / or a hue amplitude scale, and the remaining area contrasts with the color amplitude scale and / or the hue amplitude scale.

12. A computer program product stored in a computer memory and readable by the computer for performing the method according to any one of claims 1 to 11.

Images