Abstract: An imaging system and method for detecting a target in a sample. The imaging system includes a lens-free holographic microscope having a light source in a first plane spaced above an image sensor. The image sensor extends in a second plane. The system also includes a microfluidic chip positioned between the light source and the image sensor. The microfluidic chip extends in a third plane, which is parallel to the second plane. There is at least one chamber in the microfluidic chip configured to receive a sample solution with a target. The system also has a plurality of functionalized beads positioned within the at least one chamber in the microfluidic chip. Any two of the plurality of functionalized beads have an affinity for binding together when exposed to the target in the sample solution.

Abstract: A method and breast imaging system for processing breast tissue image data includes feeding image data of breast images to an image processor, identifying image portions depicting breast tissue and high density elements and executing different processing methods on input images. A first image processing method involves breast tissue enhancement and high density element suppression, whereas the second image processing method involves enhancing high density elements. Respective three-dimensional sets of image slices may be generated by respective image processing methods, and respective two-dimensional synthesized images are generated and combined to form a two-dimensional composite synthesized image which is presented through a display of the breast imaging system. First and second image processing may be executed on generated three-dimensional image sets or two-dimensional projection images acquired by an image acquisition component at respective angles relative to the patient's breast.

Abstract: An electronic apparatus is disclosed. The electronic apparatus includes a memory storing at least one instruction, and a processor, electrically connected to the memory, configured to, by executing the instruction, obtain, from an input image, a noise map corresponding to the input image; provide the input image to an input layer of a learning network model including a plurality of layers, the learning network model being an artificial intelligence (AI) model that is obtained by learning, through an AI algorithm, a relationship between a plurality of sample images, a respective noise map of each of the plurality of sample images, and an original image corresponding to the plurality of sample images; provide the noise map to at least one intermediate layer among the plurality of layers; and obtain an output image based on a result from providing the input image and the noise map to the learning network model.

Abstract: Systems and methods are provided for processing image data generated by a medical imaging system such as an ultrasound or echocardiogram system using artificial intelligence and machine learning to determine a presence of one or more congenital heart defects (CHDs) and/or other cardiovascular anomalies in the image data in a manner that is agnostic to the type of imaging system, software, and/or hardware. Image data from various types imaging systems, software, and/or hardware, having various styles of imaging data generated may be processed to determine image styles. Input image data for analysis may then be processed together with representative styles of image data to generate styled input images for each style. The styled input images may be processed by an image analyzer to detect one or more cardiovascular anomalies in the styled image data, for example. Alternatively, training data may be styled and used to train the image analyzer.

Abstract: This application relates to a medical image segmentation method, a computer device, and a storage medium. The method includes: obtaining medical image data; obtaining a target object and weakly supervised annotation information of the target object in the medical image data; determining a pseudo segmentation mask for the target object in the medical image data according to the weakly supervised annotation information; and performing mapping on the medical image data by using a preset mapping model based on the pseudo segmentation mask, to obtain a target segmentation result for the target object. Because the medical image data is segmented based on the weakly supervised annotation information, there is no need to annotate information by using much labor during training of the preset mapping model, thereby saving labor costs. The preset mapping model is a model used for mapping the medical image data based on the pseudo segmentation mask.

Abstract: A method and system for creating a dynamic self-learning medical image network system, wherein the method includes receiving, from a first node initial user interaction data pertaining to one or more user interactions with the one or more initially obtained medical images; training a deep learning algorithm based at least in part on the initial user interaction data received from the node; and transmitting an instance of the trained deep learning algorithm to the first node and/or to one or more additional nodes, wherein at each respective node to which the instance of the trained deep learning algorithm is transmitted, the trained deep learning algorithm is applied to respective one or more subsequently obtained medical images in order to obtain a result.

Abstract: A simulation platform may receive, from a plurality of image capture devices, a plurality of image streams that depict an event. The simulation platform may identify an object that is depicted in each of the plurality of image streams. The simulation platform may determine, for each of the plurality of image streams, respective image-based coordinates of a path associated with the object during the event. The simulation platform may determine, based on the respective image-based coordinates and timestamps of the plurality of image streams, simulation coordinates associated with a path of the object during the event. The simulation platform may detect, based on the simulation coordinates, that the object is involved in a collision during the event. The simulation platform may perform an action associated with detecting that the object is involved in the collision.

Abstract: Techniques for determining a classification probability of an object in an environment are discussed herein. Techniques may include analyzing sensor data associated with an environment from a perspective, such as a top-down perspective, using multi-channel data. From this perspective, techniques may determine channels of multi-channel input data and additional feature data. Channels corresponding to spatial features may be included in the multi-channel input data and data corresponding to non-spatial features may be included in the additional feature data. The multi-channel input data may be input to a first portion of a machine-learned (ML) model, and the additional feature data may be concatenated with intermediate output data from the first portion of the ML model, and input into a second portion of the ML model for subsequent processing and to determine the classification probabilities. Additionally, techniques may be performed on a multi-resolution voxel space representing the environment.

Abstract: The present application discloses a method, device, and system for processing a medical image. The method includes obtaining a source spinal image, identifying one or more vertebral bodies and one or more intervertebral discs comprised in the source spinal image, determining the vertebral body recognition results corresponding to the one or more vertebral bodies and the intervertebral disc recognition results corresponding to the one or more intervertebral discs, and determining target recognition results corresponding to the source spinal image based at least in part one on one or more of the vertebral body recognition results and the intervertebral disc recognition results.

Abstract: In some embodiments, a method can include executing a first machine learning model to detect at least one lane in each image from a first set of images. The method can further include determining an estimate location of a vehicle for each image, based on localization data captured using at least one localization sensor disposed at the vehicle. The method can further include selecting lane geometry data for each image, from a map and based on the estimate location of the vehicle. The method can further include executing a localization model to generate a set of offset values for the first set of images based on the lane geometry data and the at least one lane in each image. The method can further include selecting a second set of images from the first set of images based on the set of offset values and a previously-determined offset threshold.

Abstract: An MPI reconstruction method, device, and system based on a RecNet model include obtaining a one-dimensional (1D) MPI signal on which imaging reconstruction is to be performed, taking the 1D MPI signal as an input signal, and inputting the input signal and a velocity signal of an FFP corresponding to the input signal into a trained magnetic particle reconstruction model RecNet for image reconstruction to obtain a two-dimensional (2D) MPI image, where the magnetic particle reconstruction model RecNet is constructed based on a domain conversion network and an improved UNet network. The MPI reconstruction method, device, and system obtain a high-quality and clear magnetic particle distribution image without obtaining the system matrix.

Abstract: A method for control of data transmission in a wireless communication system includes receiving, by an application, from a modem, information indicative of a radio coverage condition, where a user equipment (UE) includes the application and the modem; based on the received information, determining that the UE is in an enhanced coverage state; and in response to the determining, controlling uplink data transmission by the modem to reduce power consumption of the UE.

Abstract: Systems and methods are described for segmenting medical images, such as magnetic resonance images, using a deep learning model that has been trained using random dropped inputs, standardized inputs, or both. Medical images can be segmented based on anatomy, physiology, pathology, other properties or characteristics represented in the medical images, or combinations thereof. As one example, multi-contrast magnetic resonance images are input to the trained deep learning model in order to generate multiple segmented medical images, each representing a different segmentation class.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for processing fundus images using fundus image processing machine learning models.

Abstract: The present application discloses a detection model training method and apparatus. The method includes determining an initial training model; determining a training sample; determining whether a target object is present in a first image through the initial detection model according to a feature of a first image, to obtain a detection result; and determining a domain that an image in the training sample belongs to through the adaptive model according to a feature of an image, to obtain a domain classification result; calculating, a loss function value related to the initial training model according to the detection result, the domain classification result, a first identifier, a second identifier, and a third identifier; and adjusting a parameter value in the initial training model according to the loss function value, to obtain a final detection model.

Abstract: Techniques are described for producing machine learning models to generate findings associated with user experiences with products and/or services. In some embodiments, a training process receives a set of findings from one or more user experience tests, where a finding includes a summary and a set of one or more references supporting the summary. The training process further identifies a supplemental set of one or more references that were not included in the initial finding to support the summary. The training process trains a machine learning model, such as a neural or generative language model, based on the first set of one or more references and the second set of one or more references to generate summaries from a subset of sampled references based at least in part on the first set of one or more references and the second set of one or more references.

Abstract: Provided is an artificial intelligence (AI) system that mimics functions, such as recognition and determination by human brains, by utilizing a machine learning algorithm, such as deep learning, and applications of the AI system.

Abstract: A radio frequency (RF) switch arrangement that improves the voltage handling capacity of a stack of switching elements (e.g., field-effect transistors (FETs)). The RF switch arrangement can include a ground plane and a stack arranged in relation to the ground plane, the stack including a plurality of switching elements coupled in series with one another. The RF switch arrangement can also include a plurality of capacitive elements, each of the plurality of capacitive elements providing a capacitive path across respective terminals of a corresponding one of the plurality of switching elements.

Abstract: According to one embodiment, a data compression apparatus includes processing circuitry. The processing circuitry generates reconstructed data by performing reconstruction processing on data. The processing circuitry generates decompressed reconstructed data by performing the reconstruction processing on decompressed data obtained by decompressing compressed data that is generated by performing compression processing on the data. The processing circuitry determines a parameter relating to a compression ratio of the data based on comparison between the reconstructed data and the decompressed reconstructed data.

Abstract: A system for searching for a pathological image includes: an autoencoder having an encoder for receiving an original pathological image and extracting a feature of the original pathological image, and a decoder for receiving the feature of the original pathological image extracted by the encoder and generating a reconstructed pathological image corresponding to the original pathological image; a diagnostic neural network for receiving the reconstructed pathological image generated by the autoencoder that has received the original pathological image, and outputting a diagnosis result of a predetermined disease; and a training module for training the autoencoder and the diagnostic neural network by inputting a plurality of training pathological images, each labeled with a diagnosis result, into the autoencoder. The autoencoder is trained by reflecting the diagnosis result of the reconstructed pathological image output from the diagnostic neural network.