Abstract: According to one embodiment, an anomaly detection device includes a processor that is configured to acquire input data. The processor derives a first anomaly degree corresponding to a difference between first feature data derived from the input data using a trained deep model and second feature data derived from the input data using a trained prediction model. The processor derives a second anomaly degree corresponding to an estimated relative positional relationship between a first and second region in the image data based on the second feature data. A total anomaly degree for the input data is then calculated from the first anomaly degree and the second anomaly degree.

Abstract: A system and method configured to better identify patient-specific anatomical landmarks, measure anatomical parameters and features, and predict the patient's need for surgery within a predetermined time period. In some embodiments, the system and method is configured to predict the likelihood or risk that a patient will require total hip arthroplasty. In some embodiments, the present invention includes machine learning technology Some embodiments of the present invention include a first ML machine configured to received medical images as inputs and identify anatomical landmarks as outputs; a measurement module to measure joint space width, hip dysplasia angles, and/or leg length differential; and a second ML machine configured to receive the anatomical measurements and patient demographic data as inputs and produce a risk or likelihood that the patient will require surgery within a certain time frame.

Abstract: The technology disclosed introduces two types of neural networks: “master” or “generalists” networks and “expert” or “specialists” networks. Both, master networks and expert networks, are fully connected neural networks that take a feature vector of an input hand image and produce a prediction of the hand pose. Master networks and expert networks differ from each other based on the data on which they are trained. In particular, master networks are trained on the entire data set. In contrast, expert networks are trained only on a subset of the entire dataset. In regards to the hand poses, master networks are trained on the input image data representing all available hand poses comprising the training data (including both real and simulated hand images).

Abstract: Various implementations disclosed herein include devices, systems, and methods that determine a wrist measurement or watch band size using depth data captured by a depth sensor from one or more rotational orientations of the wrist. In some implementations, depth data captured by a depth sensor including at least two depth map images of a wrist from different angles is obtained. In some implementations, an output is generated based on inputting the depth data into a machine learning model, the output corresponding to circumference of the wrist or a watch band size of the wrist. Then, a watch band size recommendation is provided based on the output.

Abstract: Systems and methods are disclosed for generating a specialized machine learning model by receiving a generalized machine learning model generated by processing a plurality of first training images to predict at least one cancer characteristic, receiving a plurality of second training images, the first training images and the second training images include images of tissue specimens and/or images algorithmically generated to replicate tissue specimens, receiving a plurality of target specialized attributes related to a respective second training image of the plurality of second training images, generating a specialized machine learning model by modifying the generalized machine learning model based on the plurality of second training images and the target specialized attributes, receiving a target image corresponding to a target specimen, applying the specialized machine learning model to the target image to determine at least one characteristic of the target image, and outputting the characteristic of the tar

Abstract: A method and a system for planning an orthodontic treatment are provided. The method comprises: acquiring an arch form 3D digital model including a representation of the given tooth in a current position thereof within a subject's gingiva; determining an initial crown reference point and an initial root reference point; obtaining a target position of the given tooth within the arch form 3D digital model; determining a number of steps for the given tooth to displace from the current position to the target position thereof in the course of the orthodontic treatment; and storing data indicative of the number of steps associated with the given tooth for use in the planning the orthodontic treatment.

Abstract: A method of operation of a compute system includes: qualifying a patient image for analyzing a suspected skin condition; detecting a skin area in the patient image; segmenting the skin area into a segmented image including the suspected skin condition; cropping the segmented image to form a cropped image including the suspected skin condition at a center of the cropped image; analyzing the suspected skin condition to identify a skin disease result and a disease subclass from the cropped image; and assembling a disease identification display including the patient image, a skin disease indication, an image match score, and the disease subclass for displaying on a device.

Abstract: Systems and methods are disclosed for using an integrated computing platform to view and transfer digital pathology slides using artificial intelligence, the method including receiving at least one whole slide image in a cloud computing environment located in a first geographic region, the whole slide image depicting a medical sample associated with a patient, the patient being located in the first geographic region; storing the received whole slide image in a first encrypted bucket; applying artificial intelligence to perform a classification of the at least one whole slide image, the classification comprising steps to determine whether portions of the medical sample depicted in the whole slide image are healthy or diseased; based on the classification of the at least one whole slide image, generating metadata associated with the whole slide image; and storing the metadata in a second encrypted bucket.

Abstract: A vascular image processing method performed by a processor includes extracting, from a vascular image, a first vascular region corresponding to an entire blood vessel included in the vascular image, a second vascular region corresponding to a target vessel and one or more branch vessels connected to the target vessel, and a third vascular region corresponding to the target vessel, and predicting a vascular structure in the vascular image, based on the first vascular region, the second vascular region, and the third vascular region, which are extracted from the vascular image.

Abstract: This disclosure describes, in part, techniques for identifying interactions and events associated with inventory locations. For instance, system(s) may receive image data representing a user interacting with an inventory location. The system(s) may then generate heatmap data indicating a first portion of the image data that represents the inventory location and feature data indicating a second portion of the image data that represents the user. Next, the system(s) may analyze the heatmap data with respect to the feature data to determine that the second portion of the image data corresponds to the first portion of the image data. As such, the system(s) may determine that the user is interacting with the inventory location. Based on the determination, the system(s) may analyze the first portion of the image data to identify an event that occurs at the inventory location, such as the user removing an item.

Abstract: The present disclosure provides user interface methods of and systems for displaying at least one available action overlaid on an image, comprising displaying an image; selecting at least one action and assigning a ranking weight thereto based on at least one of (1) image content, (2) current device location, (3) location at which the image was taken, (4) date of capturing the image; (5) time of capturing the image; and (6) a user preference signature representing prior actions chosen by a user and content preferences learned about the user; and ranking the at least one action based on its assigned ranking weight.

Abstract: Systems and methods for generative adversarial networks (GANs) to remove artifacts from undersampled magnetic resonance (MR) images are described. The process of training the GAN can include providing undersampled 3D MR images to the generator model, providing the generated example and a real example to the discriminator model, applying adversarial loss, L2 loss, and structural similarity index measure loss to the generator model based on a classification output by the discriminator model, and repeating until the generator model has been trained to remove the artifacts from the undersampled 3D MR images. At runtime, the trained generator model of the GAN can be generate artifact-free images or parameter maps from undersampled MRI data of a patient.

Abstract: A method of operation of a compute system includes: detecting a skin area in a patient image; segmenting the skin area into a segmented image having an acne pimple at the center; generating a target pixel array from the segmented image includes separating a plurality of the acne pimples that are adjacent in the segmented image; identifying an acne characterization of the acne pimples including an area of each acne and an acne score; and assembling a user interface display from the acne characterization for displaying on a device.

Abstract: According to the present application, a computer-implemented method of predicting thyroid eye disease is disclosed. The method comprising: preparing a conjunctival hyperemia prediction model, a conjunctival edema prediction model, a lacrimal edema prediction model, an eyelid redness prediction model, and an eyelid edema prediction model, obtaining a facial image of an object, obtaining a first processed image and a second processed image from the facial image, wherein the first processed image is different from the second processed image, obtaining predicted values for each of a conjunctival hyperemia, a conjunctival edema and a lacrimal edema by applying the first processed image to the conjunctival hyperemia prediction model, the conjunctival edema prediction model, and the lacrimal edema prediction model, and obtaining predicted values for each of an eyelid redness and an eyelid edema by applying the second processed image to the eyelid redness prediction model and the eyelid edema prediction model.

Abstract: Approaches for analyzing an input image and providing one or more outputs related to the input image are provided. In accordance with an exemplary embodiment, an input image may be received and analyzed, using a trained machine learning model, to generate an inference related to the image. Based, at least in part, upon the generated inference, one or more reports related to the inference can be generated and provided for presentation on a user device. A user can interact with the report in a conversational manner with the computer system to generate additional reports or insights related to the input image.

Abstract: A method for determining a disease state prediction, relating to a potential disease or medical condition of a subject, includes accessing a set of subject images, the subject images capturing a part of a subject's body, and accessing a set of clinical factors from the subject. The clinical factors are collected by a device or a medical practitioner substantially contemporaneously with the capture of the subject images. The subject images are inputted into an image data model to generate disease metrics for disease prediction for the subject. The disease metrics generated by the image data model and the clinical factors are inputted into a classifier to determine the disease state prediction, and the disease state prediction is returned.

Abstract: In an method for training artificial intelligence entities (AIE) for abnormality detection, medical imaging data of the human organ is provided as training data having training samples, the medical imaging data including imaging results from different types of imaging techniques for each training sample of the training data, a pre-trained or randomly initialized AIE is provided, and the AIE is trained using the provided training samples. The training may include, for at least one training sample, a first loss function for a sub-structure of the AIE is calculated independently of a first spatial region of the human organ, and, for a training sample, a second loss function for a sub-structure of the AIE is calculated independently of a second spatial region of the human organ. The AIE may be trained using the calculated first loss function and the calculated second loss function.

Abstract: Systems and methods for designing and implementing patient-specific surgical procedures and/or medical devices are disclosed. In some embodiments, a method includes receiving a patient data set of a patient. The patient data set is compared to a plurality of reference patient data sets, wherein each of the plurality of reference patient data sets is associated with a corresponding reference patient. A subset of the plurality of reference patient data sets is selected based, at least partly, on similarity to the patient data set and treatment outcome of the corresponding reference patient. Based on the selected subset, at least one surgical procedure or medical device design for treating the patient is generated.

Abstract: This application relates to an image recognition technology in the field of computer vision of artificial intelligence, and provides an image classification method and apparatus. An example method includes obtaining an input feature map of a to-be-processed image, and then performing feature extraction processing on the input feature map based on a feature extraction kernel of a neural network to obtain an output feature map, where each of a plurality of output sub-feature maps is determined based on the corresponding input sub-feature map and the feature extraction kernel, at least one of the output sub-feature maps is determined based on a target matrix obtained after an absolute value is taken, and a difference between the target matrix and the input sub-feature map corresponding to the target matrix is the feature extraction kernel. The to-be-processed image is classified based on the output feature map to obtain a classification result of the to-be-processed image.

Abstract: Systems and methods are disclosed for grouping cells in a slide image that share a similar target, comprising receiving a digital pathology image corresponding to a tissue specimen, applying a trained machine learning system to the digital pathology image, the trained machine learning system being trained to predict at least one target difference across the tissue specimen, and determining, using the trained machine learning system, one or more predicted clusters, each of the predicted clusters corresponding to a subportion of the tissue specimen associated with a target.