Abstract: A method for implementing a dynamic three-dimensional lung map view for navigating a probe inside a patient's lungs includes loading a navigation plan into a navigation system, the navigation plan including a planned pathway shown in a 3D model generated from a plurality of CT images, inserting the probe into a patient's airways, registering a sensed location of the probe with the planned pathway, selecting a target in the navigation plan, presenting a view of the 3D model showing the planned pathway and indicating the sensed location of the probe, navigating the probe through the airways of the patient's lungs toward the target, iteratively adjusting the presented view of the 3D model showing the planned pathway based on the sensed location of the probe, and updating the presented view by removing at least a part of an object forming part of the 3D model.

Abstract: A system and/or method for imaging system calibration, including: determining a dense correspondence map matching features in a first image to features in a second image, and determining updated calibration parameters associated with the image acquisition system based on the matching features.

Abstract: Certain aspects of the disclosure provide systems and methods for training an information extraction transformer model architecture directed to pre-training a first multimodal transformer model on an unlabeled dataset, training a second multimodal transformer model on a first labeled dataset to perform a key information extraction task processing the unlabeled dataset with the second multimodal transformer model to generate pseudo-labels for the unlabeled dataset, training the first multimodal transformer model based on a second labeled dataset comprising one or more labels, the pseudo-labels generated, or combinations thereof to generate a third multimodal transformer model, generating updated pseudo-labels based on label completion predictions from the third multimodal transformer model, and training the third multimodal transformer model using a noise-aware loss function and the updated pseudo-labels to generate an updated third multimodal transformer model.

Abstract: A method of performing convolution in a neural network with variable dilation rate is provided. The method includes receiving a size of a first kernel and a dilation rate, determining at least one of size of one or more disintegrated kernels based on the size of the first kernel, a baseline architecture of a memory and the dilation rate, determining an address of one or more blocks of an input image based on the dilation rate, and one or more parameters associated with a size of the input image and the memory. Thereafter, the one or more blocks of the input image and the one or more disintegrated kernels are fetched from the memory, and an output image is obtained based on convolution of each of the one or more disintegrated kernels and the one or more blocks of the input image.

Abstract: An image processing apparatus includes: a blurring amount estimation unit configured to estimate a blurring amount of a medical image including a biological body motion of a subject; and a blurring correction processing unit configured to perform blurring correction processing so as not to remove a screen motion caused by the biological body motion, on a basis of the blurring amount. This configuration makes it possible to correct blurring without removing a screen motion caused by a biological body motion. Thus, an optimum image for observing a biological body motion can be obtained.

Abstract: Method and system of deploying a machine learning neural network (MLNN). The method comprises receiving a set of input features associated with data representative of a patient medical state at input layers of a trained MLNN, the trained MLNN comprising an output layer interconnected to the input layers via intermediate layers configured in accordance with an initial matrix of weights, a subset of the input features being activated responsive to a data sufficiency threshold reached in conjunction with deactivating, from the intermediate layers, a remainder of the input layers, the trained MLNN produced in accordance with adjusting the initial matrix of weights in diminishment of false positives in providing, at the output layer, a patient state diagnosis, and generating, at the output layer, a medical state diagnosis in accordance with the diminishment of false positives.

Abstract: This document discloses system, method, and computer program product embodiments for controlling a vehicle.

Abstract: One embodiment provides for a method for evaluation of an artificial intelligence (AI) service, the method includes partitioning, by a processor, data into in-domain data and out-of-domain data. The processor defines held-out data from both of the in-domain data and the out-of-domain data for evaluation by each of domain and sub-domain based on building a taxonomy of both domains and sub-domains for the AI service. The held-out data is excluded from training data used for training the AI service. The processor further determines distribution underlying performance metrics for the held-out data using bootstrap validation processing. The processor also determines performance guarantees for multiple settings conditioned on multiple characteristics of an application scenario for the held-out data of the taxonomy based on the underlying performance metrics.

Abstract: Systems and methods are provided for generating and using statistical data which is indicative of a difference in shape of a type of anatomical structure between images acquired by a first imaging modality and images acquired by a second imaging modality. This statistical data may then be used to modify a first segmentation of the anatomical structure which is obtained from an image acquired by the first imaging modality so as to predict the shape of the anatomical structure in the second imaging modality, or in general, to generate a second segmentation of the anatomical structure as it may appear in the second imaging modality based on the statistical data and the first segmentation.

Abstract: The present disclosure relates to a scene detection method, a chip, an electronic device, and a storage medium, the present disclosure detects periodic scenes, which helps to improve the accuracy of motion estimation.

Abstract: Automated image analysis used in vascular state modeling. Coronary vasculature in particular is modeled in some embodiments. Methods of “virtual revascularization” of a presently stenotic vasculature are described; useful, for example, as a reference in disease state determinations. Structure and uses of a model which relates records comprising acquired images or other structured data to a vascular tree representation are described.

Abstract: The present disclosure relates generally to data hiding for retail product packaging and other printed objects such as substrates. One embodiment embeds an information signal in a spot color for printing on various substrates. The spot color is screened, and overprinted with process color tint. The tint is modulated prior to overprinting with optimized signal tweaks. The optimization can include consideration of a detector spectral dependency (e.g., red and/or green illumination). Many other embodiments and combinations are described in the subject patent document.

Abstract: Described are systems and methods for self-learned label refinement of a training set. In on example, a system includes a processor and a memory having a training set generation module that causes the processor to train a model using an image as an input to the model and 2D bounding based on 3D bounding boxes as ground truths, select a first subset from predicted 2D bounding boxes previously outputted by the model, retrain the model using the image as the input and the first subset as ground truths, select a second set of predicted 2D bounding boxes previously outputted by the model, and generate the training set by selecting the 3D bounding boxes from a master set of 3D bounding boxes that have corresponding 2D bounding boxes that form the second subset.

Abstract: An example method to control a light source in structured light imaging associated with a target includes obtaining two-dimensional image information and depth information of an image including the target and a scene, associating the depth information with pixels of the image, selecting a first set of pixels of the pixels based on a first set of the depth information and a second set of pixels of the pixels based on a second set of the depth information, calculating a first average pixel luminance (APL) of the first set of pixels and a total APL of the first and the second sets of pixels, obtaining a weighted APL based on the first APL and the total APL and controlling the light source based on the weighted APL.

Abstract: Systems and methods for image tagging are described. In some embodiments, images with problematic tags are identified after applying an auto-tagger. The images with problematic tags are then sent to an object detection network. In some cases, the object detection network is trained using a training set selected to improve detection of objects associated with the problematic tags. The output of the object detection network can be merged with the output of the auto-tagger to provide a combined image tagging output. In some cases, the output of the object detection network also includes a bounding box, which can be used to crop the image around a relevant object so that the auto-tagger can be reapplied to a portion of the image.

Abstract: A multi-model medical scan analysis system is operable to generate a generic model by performing a training step on image data of a plurality of medical scans and corresponding labeling data. A plurality of fine-tuned models are generated by performing a fine-tuning step on the generic model. Abnormality detection data is generated for a new medical scan by utilizing the generic model. A first one of the plurality of abnormality types that is detected in the new medical scan is determined based on a corresponding one of the plurality of probability values. Additional abnormality data is generated by performing a fine-tuned inference function on the image data of the new medical scan that utilizes one of the plurality of fine-tuned models that corresponds to the first one of the plurality of abnormality types. The additional abnormality data is transmitted for display.

Abstract: The method includes: obtaining a plurality of pieces of feature data; automatically performing two different types of nonlinear combination processing operations on the plurality of pieces of feature data to obtain two groups of processed data, where the two groups of processed data include a group of higher-order data and a group of lower-order data, the higher-order data is related to a nonlinear combination of m pieces of feature data in the plurality of pieces of feature data, and the lower-order data is related to a nonlinear combination of n pieces of feature data in the plurality of pieces of feature data, where m?3, and m>n?2; and determining prediction data based on a plurality of pieces of target data, where the plurality of pieces of target data include the two groups of processed data.

Abstract: A depth camera assembly (DCA) determines depth information. The DCA projects a dynamic structured light pattern into a local area and captures images including a portion of the dynamic structured light pattern. The DCA determines regions of interest in which it may be beneficial to increase or decrease an amount of texture added to the region of interest using the dynamic structured light pattern. For example, the DCA may identify the regions of interest based on contrast values calculated using a contrast algorithm, or based on the parameters received from a mapping server including a virtual model of the local area. The DCA may selectively increase or decrease an amount of texture added by the dynamic structured light pattern in portions of the local area. By selectively controlling portions of the dynamic structured light pattern, the DCA may decrease power consumption and/or increase the accuracy of depth sensing measurements.

Abstract: Systems and methods for analyzing perfusion-weighted medical imaging using deep neural networks are provided. In some aspects, a method includes receiving perfusion-weighted imaging data acquired from a subject using a magnetic resonance (“MR”) imaging system and modeling at least one voxel associated with the perfusion-weighted imaging data using a four-dimensional (“4D”) convolutional neural network. The method also includes extracting spatio-temporal features for each modeled voxel and estimating at least one perfusion parameter for each modeled voxel based on the extracted spatio-temporal features. The method further includes generating a report using the at least one perfusion parameter indicating perfusion in the subject.

Abstract: A biological information detection method including: causing a light source to emit first light in a first area; detecting a presence of a first living body in the first area based on a second light which is resulted in the emission of the first light; detecting body motion of the first living body based on the second light; determining whether or not the body motion is within a low level; notifying, in response to the determination that the body motion is within the low level, a first alert to the first living body, where the first alert includes an information to draw attention of the first living body.