Abstract: An image analysis method is provided. In the image analysis method, a to-be analyzed image is inputted into a region-based convolutional neural network (RCNN) model to obtain a masked image outputted from the RCNN. The center of a masked object in the masked image is calculated. The center is regarded as an origin of coordinate and the farthest coordinate point from the origin of coordinate in each of the four quadrants relative to the origin of coordinate are searched. The image analysis block is generated for each of the farthest coordinate points. The post-processing is performed on the image analysis blocks to obtain an object range.

Abstract: A non-transitory computer-readable medium stores instructions readable and executable by a workstation (18) including at least one electronic processor (20) to perform an image reconstruction method (100). The method includes: determining a weighting parameter (13) of an edge-preserving regularization or penalty of a regularized image reconstruction of an image acquisition device (12) for an imaging data set obtained by the image acquisition device; determining an edge sensitivity parameter (?) of the edge-preserving algorithm for the imaging data set obtained by the image acquisition device; and reconstructing the imaging data set obtained by the image acquisition device to generate a reconstructed image by applying the regularized image reconstruction including the edge-preserving regularization or penalty with the determined weighting and edge sensitivity parameters to the imaging data set obtained by the image acquisition device.

Abstract: A device receives an input image of a portion of a patient's body, and applies the input image to a feature extraction model, the feature extraction model comprising a trained machine learning model that is configured to generate an output that comprises, for each respective location of a plurality of locations in the input image, an indication that the input image contains an object of interest that is indicative of a presence of a disease state at the respective location. The device applies the output of the feature extraction model to a diagnostic model, the diagnostic model comprising a trained machine learning model that is configured to output a diagnosis of a disease condition in the patient based on the output of the feature extraction model. The device outputs the determined diagnosis of a disease condition in the patient obtained from the diagnostic model.

Abstract: A method includes providing a target object and acquiring measured images of the target object. Each of the measured images is acquired by filtering radiation from the target object by a mask having multiple holes and detecting filtered radiation by a detector. The method further includes providing an estimated image of the target object and calculating an updating factor for each of the measured images. The calculating of the updating factor includes partitioning a mathematical representation of the mask into multiple first regions; for each of the first regions, deriving a separate forward projection from the estimated image of the target object and the respective first region; and comparing the respective measured image of the target object with the forward projections. The method further includes updating the estimated image of the target object based on the updating factors.

Abstract: A mechanism is provided for seed relabeling for seed-based slice-wise lesion segmentation. The mechanism receives a lesion mask for a three-dimensional medical image volume. The lesion mask corresponds to detected lesions in the medical image volume and wherein each detected lesion has a lesion contour. The mechanism generates a distance map for a given two-dimensional slice in the medical image volume based on the lesion mask. The distance map comprises a distance to a lesion contour for each voxel of the given two-dimensional slice. The mechanism performs local maxima identification to select a set of local maxima from the distance map such that each local maximum has a value greater than its immediate neighbor points. The mechanism performs seed relabeling based on the distance map and the set of local maxima to generate a set of seeds. Each seed represents a center of a distinct component of a lesion contour.

Abstract: Disclosed herein are systems and methods for enhancement of objects of interest in medical images.

Abstract: According to one aspect, a processor device in communication with an endoscopic device obtains a spectrum image of bleeding in an upper gastrointestinal (GI) area of a patient. A filter is applied to the spectrum image to generate a pre-enhanced image. The filter enhances the spectrum image at one or more light wavelengths in the light spectrum. The pre-enhanced image is analyzed to identify an area of interest that represents a portion of the upper GI area with an active bleed. A contrast enhancement technique is applied to the area of interest in the pre-enhanced image to generate an enhanced contrast image. Spatial filters are applied to the enhanced contrast image to produce a final colorized image with defined blood vessels in the upper GI area of the patient.

Abstract: A display device includes an active matrix substrate and a detection circuit. The active matrix substrate has pixels that are configured to display an image. The pixels include display and photo sensing pixels each including a display region and a photo sensing region. The display region is configured to emit light with a predetermined emitting frequency in a blank period of the display device. The photo sensing region is adjacent to the display region, and is configured to detect the light emitted by the display region and reflected from a biometric object and to convert the reflected light into photo detection signals. The detection circuit is coupled to the photo sensing circuit of each of the display and photo sensing pixels, and is configured to construct a biometric pattern corresponding to the biometric object based on the photo detection signals.

Abstract: A method for image reconstruction may include obtaining projection data D0 generated by an imaging device by scanning an object at first angles, wherein the first angles may be a subset of second angles. The second angles may include at least short-scan angles for a system to conduct image reconstruction. The method may also include generating an image F1 based on the projection data D0. The method may also include determining, based on the image F1, projection data D1 corresponding to third angles that are a subset of the second angles and different from the first angles. The method may also include generating a final image associated with the object by performing an iteration process including one or more iterations using initial data including the image F1, the projection data D0, and the projection data D1.

Abstract: A learning model provided in a segmentation device is a learning model which is generated using training data such that segmentation data of a feature region is output when at least one of projection data and reconfiguration data acquired by an imaging device or data derived from the at least one of projection data and reconfiguration data is input.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for segmenting a medical image. In one aspect, a method comprises: receiving a medical image that is captured using a medical imaging modality and that depicts a region of tissue in a body; and processing the medical image using a segmentation neural network to generate a segmentation output. The segmentation neural network can include a sequence of multiple encoder blocks and a decoder subnetwork. Training the segmentation neural network can include determining a set of error values for a segmentation channel; identifying the highest error values from the set of error values for the segmentation channel; and determining a segmentation loss based on the highest error values identified for the segmentation channel.

Abstract: In accordance with implementations of the subject matter described herein, a solution for fine-grained image recognition is proposed. This solution includes extracting a global feature of an image using a first sub-network of a first learning network; determining a first attention region of the image based on the global feature using a second sub-network of the first learning network, the first attention region including a discriminative portion of an object in the image; extracting a first local feature of the first attention region using a first sub-network of a second learning network; and determining a category of the object in the image based at least in part on the first local feature. Through this solution, it is possible to localize an image region at a finer scale accurately such that a local feature at a fine scale can be obtained for object recognition.

Abstract: A method for tracing pollution at a drainage outlet of a culvert. (1) The main line of the culvert is fully detected through an inspection robot to obtain the image data, water quality data, coordinates and other basic data of the outlet. (2) The type of the outlet is determined according to the data information obtained in step (1). (3) Pollutants in the drainage outlet with suspected rainwater-sewage connection are traced, and sources of the pollutants flowing to the outlet are all located.

Abstract: A method for analysis of yeast includes: receiving a microscopic image of yeast by a cloud server (2901), the microscopic image including a scaling pattern for determining a magnification; determining the magnification by the cloud server based on the scaling pattern (2902); and analyzing, by the cloud server, the microscopic image based on the magnification to obtain an analysis result (2903).

Abstract: A navigation method based on ground texture images, an electronic device and storage medium. The method includes: performing transform domain based image registration on an acquired image of a current frame and an image of a previous frame, and determining a first pose of the image of the current frame; determining whether the image of the current frame meets a preset condition, and if so, inserting the image of the current frame as the key-frame image into a map, and performing loop closure detection and determining a loop key-frame image; performing transform domain based image registration on the image of the current frame and the loop key-frame image, and determining a second pose of the image of the current frame; and determining an accumulated error according to the first pose and the second pose, and correcting the map according to the accumulated error.

Abstract: In a data processing unit, alignment is performed by appropriately deforming one image among MS imaging images acquired from different samples so that positions and sizes on the MS imaging image are matched (S1 to S5). When the aligned image is displayed on a screen of a display unit and a user sets a region of interest on the image serving as a reference (S6), a micro region including a center point within a range of the set region of interest is extracted in each of an image serving as the reference and an image not serving as the reference (S7).

Abstract: An approach for improving object detection performance by using bilateral organ information disclosed. The approach comprises introducing a penalty term to an object localization model. The penalty term encourages the model to identify same physical regions (actual same physical location in the patient's body) in multiple images, which create an image of the same organ. The approach includes a sub-component of the model that can output an entity embedding, in additional to the conventional classification and localization prediction. During optimization process, similar entity embedding for the same logical entity are encouraged, and similar entity embedding for different logical entities are discouraged.

Abstract: A system and a method for monitoring a brain CT scan image using ASPECTS score. The method includes receiving the brain CT scan image of a patient. Further, a basal ganglia region and a corona radiata level are identified in a plurality of slices in the brain CT scan image. Furthermore, a plurality of anatomical regions, a plurality of infarcts and a plurality of black regions are segmented using deep learning. Subsequently, an overlapping region across the plurality of slices is determined based on the plurality of anatomical regions, the plurality of infarcts, and the plurality of black regions. The overlapping region and a predefined threshold are used to compute an ASPECTS score. The ASPECTS score is further used to recommend a course of action to the patient.

Abstract: A three-dimensional data encoding method includes: dividing three-dimensional points included in three-dimensional data into three-dimensional point sub-clouds including a first three-dimensional point sub-cloud and a second three-dimensional point sub-cloud; appending first information indicating a space of the first three-dimensional point sub-cloud to a header of the first three-dimensional point sub-cloud, and appending second information indicating a space of the second three-dimensional point sub-cloud to a header of the second three-dimensional point sub-cloud; and encoding the first three-dimensional point sub-cloud and the second three-dimensional point sub-cloud so that the first three-dimensional point sub-cloud and the second three-dimensional point sub-cloud are decodable independently of each other.

Abstract: A method of harvesting lesion annotations includes conditioning a lesion proposal generator (LPG) based on a first two-dimensional (2D) image set to obtain a conditioned LPG, including adding lesion annotations to the first 2D image set to obtain a revised first 2D image set, forming a three-dimensional (3D) composite image according to the revised first 2D image set, reducing false-positive lesion annotations from the revised first 2D image set according to the 3D composite image to obtain a second-revised first 2D image set, and feeding the second-revised first 2D image set to the LPG to obtain the conditioned LPG, and applying the conditioned LPG to a second 2D image set different than the first 2D image set to harvest lesion annotations.