Abstract: Disclosed herein are systems, methods, devices, and media for carrying out detection of ophthalmic and systemic diseases and disorders. Deep learning algorithms enable the automated analysis of ophthalmic images such as retinal scans to generate accurate detection of various diseases and disorders. Point-of-care implementations allow for rapid and efficient detection outside of the clinical setting.

Abstract: The disclosure relates to a wind turbine blade inspection system and method based on an unmanned aerial vehicle. The system includes: a collection module, used for collecting blade data and surrounding environment data of blades to be inspected, determining feature inspection points of the unmanned aerial vehicle according to the blade data and the surrounding environment data, and generating inspection paths; an inspection module, used for shooting corresponding blade at the feature inspection points according to the inspection paths to obtain a first inspection image and a second inspection image; an analysis module, used for receiving the first inspection image and the second inspection image, analyzing the first inspection image and the second inspection image to obtain a health state of the corresponding blade, and making a maintenance plan according to the health state of each of the blades.

Abstract: Systems and methods for augmenting a training data set with annotated pseudo images for training machine learning models. The pseudo images are generated from corresponding images of the training data set and provide a realistic model of the interaction of image generating signals with the patient, while also providing a realistic patient model. The pseudo images are of a target imaging modality, which is different than the imaging modality of the training data set, and are generated using algorithms that account for artifacts of the target imaging modality. The pseudo images may include therein the contours and/or features of the anatomical structures contained in corresponding medical images of the training data set. The trained models can be used to generate contours in medical images of a patient of the target imaging modality or to predict an anatomical condition that may be indicative of a disease.

Abstract: A defect detection method includes obtaining a first image corresponding to a first area on an object revealing an apparent defect, the first area bounding the area showing the apparent defect; selecting a detection model according to a size of the first image, the detection model selected being one of a group of three models based on dimensions of image, the image being in one of portrait, landscape, or regular square proportions; detecting and confirming or denying a defect on the first image according to the detection model and outputting a detection result, the use of only three possible AI models reduces the likelihood of mis-determinations. An electronic device and a non-volatile storage medium therein, for performing the above-described method, are also disclosed.

Abstract: Embodiments relate to a front-end scaler circuit configured to receive and process demosaiced image data in different modes depending on if the demosaiced image data was demosaiced from Bayer or Quad Bayer raw image data. The front-end scaler circuit shares memory with a demosaicing circuit, and is configured to perform different operations that use different amounts of the shared memory based on the original image format of the demosaiced image data being processed, to compensate for additional memory utilized by the demosaicing circuit when demosaicing certain types of image data. For example, when processing image data demosaiced from Quad Bayer image data, the front-end scaler circuit discards a portion of the chrominance component data for the received image data before performing chromatic suppression, compared to when processing image data demosaiced from Bayer image data.

Abstract: An object identification method is disclosed. The method includes obtaining images of a target geographical area and telemetry information of an image-collection vehicle at a time of capture, analyzing each image to identify objects, and determining a position of the objects. The method further includes determining an image capture height, determining a position of the image using the capture height and the telemetry information, performing a transform on the image based on the capture height and the telemetry information, identifying the objects in the transformed image, determining first pixel locations of the objects within the transformed image, performing a reverse transform on the first pixel locations to determine second pixel locations in the image, and determining positions of the objects within the area based on the second pixel locations within the captured image and the determined image position.

Abstract: Presented herein are systems and methods that provide for improved detection and characterization of lesions within a subject via automated analysis of nuclear medicine images, such as positron emission tomography (PET) and single photon emission computed tomography (SPECT) images. In particular, in certain embodiments, the approaches described herein leverage artificial intelligence (AI) to detect regions of 3D nuclear medicine images corresponding to hotspots that represent potential cancerous lesions in the subject. The machine learning modules may be used not only to detect presence and locations of such regions within an image, but also to segment the region corresponding to the lesion and/or classify such hotspots based on the likelihood that they are indicative of a true, underlying cancerous lesion. This AI-based lesion detection, segmentation, and classification can provide a basis for further characterization of lesions, overall tumor burden, and estimation of disease severity and risk.

Abstract: The present invention relates to a method for predicting neurodegenerative decline and/or its severity for a patient, especially of cognitive impairment (CI). Strokes and Parkinson's disease are frequently associated with occurrence of long-term cognitive impairment or dementia with still incompletely resolved mechanisms. The discovery of diagnostic and predictive biomarkers thus remains a major challenge. The method of the invention uses radiomics corresponding to texture features extracted from a plurality of previously-acquired medical brain images and correlated with previously-acquired clinical and/or biological data. A classifier is trained beforehand for learning these radiomics, and then operated on radiomics computed from at least one brain image of a patient to generate a score representative of its risks of neurodegenerative decline. By applying this method on a cohort of 160 MCI and non-MCI patients, the inventors show that MCI patients could be early predicted with a mean accuracy of 88%.

Abstract: As an aspect of the present disclosure, a method of processing an image may be proposed. The method is a method of processing an image, which is performed by an electronic device including one or more processors and one or more memories in which instructions to be executed by the one or more processors are stored, and may include obtaining a first mixed image of a sample including a first biomolecule labeled with a first fluorescent material and a second biomolecule that has not been labeled, obtaining a second mixed image of the sample including the first biomolecule labeled with the first fluorescent material and the second biomolecule labeled with a second fluorescent material, and generating an unmixed image of the second biomolecule based on the first mixed image and the second mixed image.

Abstract: Disclosed in an intravascular ultrasound (IVUS) image analysis method, comprising the steps of: allowing a computer to acquire an IVUS image of an object; segmenting a constituent element included in the IVUS image; and determining the constituent parts and the degree of risk of plaque included in the IVUS image.

Abstract: Techniques are disclosed for correcting noisy clamped image sources while preserving original luma, chroma, or other channel value. A computing device can receive an image that has a plurality of pixels associated with respective channel values. The computing device can then determine a region of the image for correction and generate a histogram of the respective channel values in the region. Based on the histogram, the computing device can generate a corrected histogram of corrected channel values. The computing device can then denoise the corrected channel values.

Abstract: In an object identification apparatus, a foreground extraction unit performs a foreground extraction with respect to input images, and generates a foreground extraction result. A state extraction unit extracts a state for each foreground based on the foreground extraction result. An identification model selection unit selects one or more identification models based on the extracted state for each foreground by using a selection model. An identification unit identifies a moving object included in the input images using the selected identification model.

Abstract: An image processing method is disclosed. The image processing method includes setting n reference pixels around a pixel of interest, where n is an integer, sequentially comparing a pixel value of the pixel of interest with a pixel value of each reference pixel, counting a number of the reference pixels whose pixel values are less than or equal to the pixel value of the pixel of interest, incrementing a histogram value of each reference pixel by 1 to compare with a preset inclination limit value, counting a number of the reference pixels satisfying both a condition that the reference pixel value is less than or equal to the pixel value of the pixel of interest and a condition that the histogram value is less than or equal to the inclination limit value so as to obtain a true-in-both count value, and proportionally distributing the true-in-both count value for output.

Abstract: There is provided herein an apparatus and method for roadside asset tracking and maintenance monitoring having a mobile unit with data capture devices for capturing roadside asset imagery, global positioning system (GPS) receivers and data interfaces for communicating with an asset management server. As such, the apparatus may take roadside imagery for automated asset identification which may include utilising an asset type image recognition technique for automating the identification of the roadside assets.

Abstract: The present disclosure describes imaging systems configured to generate adaptive scanning protocols based on anatomical features and conditions identified during a prenatal scan of an object. Systems may include an ultrasound transducer configured to acquire echo signals responsive to ultrasound pulses transmitted toward a target region. Processors coupled with the transducer can generate an image frame from the echoes and provide the image frame to a first neural network. The first neural network may be configured to identify an anatomical feature in the image frame. An indication of the anatomical feature may be provided a second neural network. The second neural network may then determine an anatomical measurement to be obtained based, in part, on the feature identified. The processors may be further configured to cause an indicator of the anatomical measurement to be obtained to be displayed on a user interface.

Abstract: A method for processing breast tissue image data includes processing image data of a patient's breast tissue to generate a high-dimensional grid depicting one or more high-dimensional objects in the patient's breast tissue; determining a probability or confidence of each of the one or more high-dimensional objects depicted in the high-dimensional grid; and modifying one or more aspects of at least one of the one or more high-dimensional objects based at least in part on its respective determined probability or confidence to thereby generate a lower-dimensional format version of the one or more high-dimensional objects. The method may further include displaying the lower-dimensional format version of the one or more high-dimensional objects in a synthesized image of the patient's breast tissue.

Abstract: The present disclosure provides a method, apparatus, medium and device for extracting a river drying-up region and frequency, and belongs to the technical field of remote sensing. The method for extracting a river drying-up region and frequency can be applied to efficiently acquiring river drying-up information for a long time within the large spatial region. With the inverse normalized difference water index (iNDWI) and the maximum value composite (MVC) method for the remote sensing images, the present disclosure omits the troublesome step of separately extracting a river range for each image in the conventional method. In addition, the present disclosure quickly obtains the drying-up frequency by counting the total number of available images and the number of non-water images at the same pixel position, and yields a greater efficiency for extracting the river drying-up region and frequency.

Abstract: Systems and methods for planning a surgical procedure are provided. Information corresponding to an examination of a patient and an image of the patient may be received. The information and the image may be inputted into an analytical model configured to identify a pathology location. A needed surgical modification of the patient anatomy may be automatically identified. At least a portion of an anatomical element in a three-dimensional model of the patient anatomy may be automatically labeled.

Abstract: For reconstruction in medical imaging using phase correction, a machine learning model is trained for reconstruction of an image. The reconstruction may be for a sequence without repetitions or may be for a sequence with repetitions. Where repetitions are used, rather than using just a loss for that repetition in training, the loss based on an aggregation of images reconstructed from multiple repetitions may used to train the machine learning model. In either approach, a phase correction is applied in machine training. A phase map is extracted from output of the model in training or extracted from the ground truth of the training data. The phase correction, based on the phase map, is applied to the ground truth and/or the output of the model in training. The resulting machine-learned model may better reconstruct an image as a result of having been trained using phase correction.

Abstract: A method and a device for localizing and lateralizing a brain functional region, an electronic apparatus, and a storage medium are provided herein, wherein the method for localizing and lateralizing the brain functional region comprises: acquiring brain structural magnetic resonance imaging data and brain functional magnetic resonance imaging data of a subject (201) based on the brain structural magnetic resonance imaging data and the brain functional magnetic resonance imaging data, determining a brain functional map of the subject (202), wherein the brain functional map comprises brain functional region identifiers of at least two brain functional regions and a corresponding set of voxels; for each of to-be-clinically-intervened voxels in a set of the to-be-clinically-intervened voxels, determining, based on the to-be-clinically-intervened voxel and the brain functional map, the brain functional region identifier corresponding to the to-be-clinically-intervened voxel (203); and determining a brain functiona