Abstract: A method including processing electrocardiograph (ECG) signals taken over a heartbeat received over a plurality of channels, selecting a subset of the ECG signals captured at a point in time with a window of interest (WOI) around reference annotations and having a morphology pattern within the WOI, storing the morphology patterns, receiving continuous sets of ECG signals taken over a plurality of heartbeats over the plurality of channels and having a morphology pattern within the WOI, performing a correlation between the stored morphology patterns of the ECH signals and the morphology patterns of the continuous sets of ECH signals for each heartbeat, generating a correlation coefficient that is a measure of a goodness of fit between geometries of the ECG signals and the continuous sets of ECG signals and identifying each heartbeat having a correlation coefficient that exceeds a threshold coefficient as having been caused by an arrhythmia.

Abstract: There is provided a method for training a ground truth generator machine learning model, comprising: creating a ground truth multi-record training dataset wherein a record comprises: a first image of a sample of tissue of a subject depicting a first group of biological objects, a second image of the sample depicting a second group of biological objects presenting at least one biomarker, and ground truth labels indicating a respective biological object category of a plurality of biological object categories for biological object members of the first group and the second group; and training the ground truth generator machine learning model on the ground truth multi-record training dataset for automatically generating ground truth labels selected from the plurality of biological object categories for biological objects depicted in an input set of images of a first type corresponding to the first image and a second type corresponding to the second image.

Abstract: Novel tools and techniques are provided for implementing augmented reality (AR)-based assistance within a work environment. In various embodiments, a computing system might receive, from a camera having a field of view of a work environment, first images of at least part of the work environment, the first images overlapping with a field of view of a user wearing an AR headset; might analyze the received first images to identify objects; might query a database(s) to determine a task associated with a first object(s) among the identified objects; might generate an image overlay providing at least one of graphical icon-based, text-based, image-based, and/or highlighting-based instruction(s) each indicative of instructions presented to the user to implement the task associated with the first object(s); and might display, to the user's eyes through the AR headset, the generated first image overlay that overlaps with the field of view of the user's eyes.

Abstract: A method includes inserting a catheter into an organ of a patient and selecting, in a three-dimensional (3D) image of the organ, a plane of interest (POI). A first image, which includes an endoscopic view of the 3D image from a direction facing the POI, is produced. A second image, which includes a sectional view of the 3D image that is clipped by the POI, is produced, and the first and second images are displayed to a user.

Abstract: Novel tools and techniques are provided for implementing digital microscopy imaging using deep learning-based segmentation and/or implementing instance segmentation based on partial annotations. In various embodiments, a computing system might receive first and second images, the first image comprising a field of view of a biological sample, while the second image comprises labeling of objects of interest in the biological sample. The computing system might encode, using an encoder, the second image to generate third and fourth encoded images (different from each other) that comprise proximity scores or maps. The computing system might train an AI system to predict objects of interest based at least in part on the third and fourth encoded images. The computing system might generate (using regression) and decode (using a decoder) two or more images based on a new image of a biological sample to predict labeling of objects in the new image.

Abstract: There is provided a computer implemented method of automatically creating a training dataset comprising a plurality of records, wherein a record includes: an image of a sample of an object, an indication of monitored manipulations by a user of a presentation of the sample, and a ground truth indication of a monitored gaze of the user viewing the sample on a display or via an optical device mapped to pixels of the image of the sample, wherein the monitored gaze comprises at least one location of the sample the user is viewing and an amount of time spent viewing the at least one location.

Abstract: A system and method for non-invasively detecting abnormal electrical propagation in the heart are disclosed. The system and method include an interface for receiving a pacing signal applied to a heart of a patient, the pacing signal comprising (i) a sequence of regular pacing stimuli shorter than the sinus-rate intervals, and (ii) one or more extra pacing stimuli at intervals that are shorter than the regular pacing stimuli, a processor to assess the envelope of a body-surface ECG component after the regular pacing stimuli, assess the envelope of a body surface ECG component after the one or more extra pacing stimuli, and compare the assessed component after the extra pacing stimuli to the assessed component after the regular pacing stimuli. The interface outputting the comparison as an indication of regions of arrhythmogenicity and ablation targets in the heart.

Abstract: Novel tools and techniques are provided for implementing digital microscopy imaging using deep learning-based segmentation and/or implementing instance segmentation based on partial annotations. In various embodiments, a computing system might receive first and second images, the first image comprising a field of view of a biological sample, while the second image comprises labeling of objects of interest in the biological sample. The computing system might encode, using an encoder, the second image to generate third and fourth encoded images (different from each other) that comprise proximity scores or maps. The computing system might train an AI system to predict objects of interest based at least in part on the third and fourth encoded images. The computing system might generate (using regression) and decode (using a decoder) two or more images based on a new image of a biological sample to predict labeling of objects in the new image.

Abstract: Novel tools and techniques are provided for implementing augmented reality (AR)-based assistance within a work environment. In various embodiments, a computing system might receive, from a camera having a field of view of a work environment, first images of at least part of the work environment, the first images overlapping with a field of view of a user wearing an AR headset; might analyze the received first images to identify objects; might query a database(s) to determine a task associated with a first object(s) among the identified objects; might generate an image overlay providing at least one of graphical icon-based, text-based, image-based, and/or highlighting-based instruction(s) each indicative of instructions presented to the user to implement the task associated with the first object(s); and might display, to the user's eyes through the AR headset, the generated first image overlay that overlaps with the field of view of the user's eyes.

Abstract: Novel tools and techniques are provided for implementing augmented reality (AR)-based assistance within a work environment. In various embodiments, a computing system might receive, from a camera having a field of view of a work environment, first images of at least part of the work environment, the first images overlapping with a field of view of a user wearing an AR headset; might analyze the received first images to identify objects; might query a database(s) to determine a task associated with a first object(s) among the identified objects; might generate an image overlay providing at least one of graphical icon-based, text-based, image-based, and/or highlighting-based instruction(s) each indicative of instructions presented to the user to implement the task associated with the first object(s); and might display, to the user's eyes through the AR headset, the generated first image overlay that overlaps with the field of view of the user's eyes.

Abstract: Novel tools and techniques are provided for implementing digital microscopy imaging using deep learning-based segmentation and/or implementing instance segmentation based on partial annotations. In various embodiments, a computing system might receive first and second images, the first image comprising a field of view of a biological sample, while the second image comprises labeling of objects of interest in the biological sample. The computing system might encode, using an encoder, the second image to generate third and fourth encoded images (different from each other) that comprise proximity scores or maps. The computing system might train an AI system to predict objects of interest based at least in part on the third and fourth encoded images. The computing system might generate (using regression) and decode (using a decoder) two or more images based on a new image of a biological sample to predict labeling of objects in the new image.

Abstract: A method includes acquiring a bipolar signal from a first electrode and a second electrode contacting a first location and a second location, respectively, in a heart of a living subject. The method further includes acquiring a unipolar signal from the first electrode while in contact with the first location, and deriving from the bipolar signal and the unipolar signal a point in time at which the first location is generating the unipolar signal. The method also includes computing a metric for a conduction velocity of the unipolar signal at the first location based on a shape of the unipolar signal at the point in time.

Abstract: A method including processing electrocardiograph (ECG) signals taken over a heartbeat received over a plurality of channels, selecting a subset of the ECG signals captured at a point in time with a window of interest (WOI) around reference annotations and having a morphology pattern within the WOI, storing the morphology patterns, receiving continuous sets of ECG signals taken over a plurality of heartbeats over the plurality of channels and having a morphology pattern within the WOI, performing a correlation between the stored morphology patterns of the ECH signals and the morphology patterns of the continuous sets of ECH signals for each heartbeat, generating a correlation coefficient that is a measure of a goodness of fit between geometries of the ECG signals and the continuous sets of ECG signals and identifying each heartbeat having a correlation coefficient that exceeds a threshold coefficient as having been caused by an arrhythmia.

Abstract: Acquiring ECG signals from electrodes positioned in a heart taken over a single heartbeat, selecting a morphology pattern within a window of interest around time of occurrence annotations for the signals; computing a weighted cross-correlation between each morphology pattern of the signals and a stored template morphology pattern, to generate a weighted correlation coefficient of a match between the morphology patterns of the acquired signals and the stored morphology pattern; iteratively changing a phase of the signals relative to the phase of the morphology pattern and repeating the step of generating the weighted correlation coefficient at each iteration; determining a maximum value of the weighted correlation coefficient based on the iterations; comparing the maximum value to a threshold; and when the maximum value exceeds the threshold, accepting the heartbeat as having been caused by the arrhythmia and incorporating a location of the arrhythmia into a local activation map.

Abstract: Novel tools and techniques are provided for implementing digital microscopy imaging using deep learning-based segmentation via multiple regression layers, implementing instance segmentation based on partial annotations, and/or implementing user interface configured to facilitate user annotation for instance segmentation. In various embodiments, a computing system might generate a user interface configured to collect training data for predicting instance segmentation within biological samples, and might display, within a display portion of the user interface, the first image comprising a field of view of a biological sample. The computing system might receive, from a user via the user interface, first user input indicating a centroid for each of a first plurality of objects of interest and second user input indicating a border around each of the first plurality of objects of interest.

Abstract: Novel tools and techniques are provided for implementing digital microscopy imaging using deep learning-based segmentation via multiple regression layers, implementing instance segmentation based on partial annotations, and/or implementing user interface configured to facilitate user annotation for instance segmentation. In various embodiments, a computing system might generate a user interface configured to collect training data for predicting instance segmentation within biological samples, and might display, within a display portion of the user interface, the first image comprising a field of view of a biological sample. The computing system might receive, from a user via the user interface, first user input indicating a centroid for each of a first plurality of objects of interest and second user input indicating a border around each of the first plurality of objects of interest.

Abstract: Novel tools and techniques are provided for implementing digital microscopy imaging using deep learning-based segmentation and/or implementing instance segmentation based on partial annotations. In various embodiments, a computing system might receive first and second images, the first image comprising a field of view of a biological sample, while the second image comprises labeling of objects of interest in the biological sample. The computing system might encode, using an encoder, the second image to generate third and fourth encoded images (different from each other) that comprise proximity scores or maps. The computing system might train an AI system to predict objects of interest based at least in part on the third and fourth encoded images. The computing system might generate (using regression) and decode (using a decoder) two or more images based on a new image of a biological sample to predict labeling of objects in the new image.

Abstract: A method includes acquiring a bipolar signal from a first electrode and a second electrode contacting a first location and a second location, respectively, in a heart of a living subject. The method further includes acquiring a unipolar signal from the first electrode while in contact with the first location, and deriving from the bipolar signal and the unipolar signal a point in time at which the first location is generating the unipolar signal. The method also includes computing a metric for a conduction velocity of the unipolar signal at the first location based on a shape of the unipolar signal at the point in time.

Abstract: A method includes acquiring a bipolar signal from a first electrode and a second electrode contacting a first location and a second location, respectively, in a heart of a living subject. The method further includes acquiring a unipolar signal from the first electrode while in contact with the first location, and deriving from the bipolar signal and the unipolar signal a point in time at which the first location is generating the unipolar signal. The method also includes computing a metric for a conduction velocity of the unipolar signal at the first location based on a shape of the unipolar signal at the point in time.

Abstract: Acquiring ECG signals from electrodes positioned in a heart taken over a single heartbeat, selecting a morphology pattern within a window of interest around time of occurrence annotations for the signals; computing a weighted cross-correlation between each morphology pattern of the signals and a stored template morphology pattern, to generate a weighted correlation coefficient of a match between the morphology patterns of the acquired signals and the stored morphology pattern; iteratively changing a phase of the signals relative to the phase of the morphology pattern and repeating the step of generating the weighted correlation coefficient at each iteration; determining a maximum value of the weighted correlation coefficient based on the iterations; comparing the maximum value to a threshold; and when the maximum value exceeds the threshold, accepting the heartbeat as having been caused by the arrhythmia and incorporating a location of the arrhythmia into a local activation map.