Abstract: For patient weight estimation in a medical imaging system, a patient model, such as a mesh, is fit to a depth image. One or more feature values are extracted from the fit patient model, reducing the noise and clutter in the values. The weight estimation is regressed from the extracted features.

Abstract: A neural network-based method for quantifying a volume of a specimen. The method includes providing a specimen, capturing images of the specimen, and directly classifying to one of a plurality of volume classes or volumes using a trained neural network. Quality check modules and specimen testing apparatus adapted to carry out the volume quantification method are described, as are other aspects.

Abstract: A model-based method of classifying a specimen in a specimen container. The method includes capturing images of the specimen and container at multiple different exposures times, at multiple different spectra having different nominal wavelengths, and at different viewpoints by using multiple cameras. From the captured images, 2D data sets are generated. The 2D data sets are based upon selection of optimally-exposed pixels from the multiple different exposure images to generate optimally-exposed image data for each spectra. Based upon these 2D data sets, various components are classified using a multi-class classifier, such as serum or plasma portion, settled blood portion, gel separator (if present), tube, air, or label. From the classification data and 2D data sets, a 3D model can be generated. Specimen testing apparatus and quality check modules adapted to carry out the method are described, as are other aspects.

Abstract: A model-based method of inspecting a specimen for presence of one or more artifacts (e.g., a clot, bubble, and/or foam). The method includes capturing multiple images of the specimen at multiple different exposures and at multiple spectra having different nominal wavelengths, selection of optimally-exposed pixels from the captured images to generate optimally-exposed image data for each spectra, computing statistics of the optimally-exposed pixels to generate statistical data, identifying a serum or plasma portion of the specimen, and classifying, based on the statistical data, whether an artifact is present or absent within the serum or plasma portion. Testing apparatus and quality check modules adapted to carry out the method are described, as are other aspects.

Abstract: A method for training a learning-based medical scanner including (a) obtaining training data from demonstrations of scanning sequences, and (b) learning the medical scanner's control policies using deep reinforcement learning framework based on the training data.

Abstract: A method of tube slot localization is provided using a tray coordinate system and a camera coordinate system. The method includes receiving, a series of images from at least one camera of a tray comprising tube slots arranged in a matrix of rows and columns. Each tube slot is configured to receive a sample tube. The method also includes automatically detecting fiducial markers disposed on cross sectional areas between the tube slots on the tray and receiving an encoder value indicating when each row of the tray is substantially at the center of the camera's field of view. The method further includes determining calibration information to provide mapping of locations from the tray coordinate system to locations from the camera coordinate system and automatically aligning the tray based on the encoder value and calibration information.

Abstract: Embodiments provide a method of using image-based tube top circle detection based on multiple candidate selection to localize the tube top circle region in input images. According to embodiments provided herein, the multi-candidate selection enhances the robustness of tube circle detection by making use of multiple views of the same tube to improve the robustness of tube top circle detection. With multiple candidates extracted from images under different viewpoints of the same tube, the multi-candidate selection algorithm selects an optimal combination among the candidates and provides more precise measurement of tube characteristics. This information is invaluable in an IVD environment in which a sample handler is processing the tubes and moving the tubes to analyzers for testing and analysis.

Abstract: An automation method is disclosed for an X-ray tube scanner having an X-ray tube and an X-ray detector. The method allows the X-ray tube scanner to detect the X-ray detector's plane with an object to be imaged placed on the X-ray detector; determine a boundary box of the object to be imaged on the X-ray detector; determine the object's center position and orientation on the X-ray detector's plane; transfer the object's center position from the object's coordinate system to the X-ray tube's coordinate system; and estimate the X-ray tube control parameters for aligning the X-ray field emitted from the X-ray tube's collimator to the object's center position and the object's orientation on the X-ray detector.

Abstract: The present embodiments relate to localizing a mobile device in a complex, three-dimensional scene. By way of introduction, the present embodiments described below include apparatuses and methods for using multiple, independent pose estimations to increase the accuracy of a single, resulting pose estimation. The present embodiments increase the amount of input data by windowing a single depth image, using multiple depth images from the same sensor, and/or using multiple depth image from different sensors. The resulting pose estimation uses the input data with a multi-window model, a multi-shot model, a multi-sensor model, or a combination thereof to accurately estimate the pose of a mobile device.

Abstract: Systems and methods for image classification include receiving imaging data of in-vivo or excised tissue of a patient during a surgical procedure. Local image features are extracted from the imaging data. A vocabulary histogram for the imaging data is computed based on the extracted local image features. A classification of the in-vivo or excised tissue of the patient in the imaging data is determined based on the vocabulary histogram using a trained classifier, which is trained based on a set of sample images with confirmed tissue types.

Abstract: Methods for image-based detection of the tops of sample tubes used in an automated diagnostic analysis system may be based on a convolutional neural network to pre-process images of the sample tube tops to intensify the tube top circle edges while suppressing the edge response from other objects that may appear in the image. Edge maps generated by the methods may be used for various image-based sample tube analyses, categorizations, and/or characterizations of the sample tubes to control a robot in relationship to the sample tubes. Image processing and control apparatus configured to carry out the methods are also described, as are other aspects.

Abstract: A method of characterizing a serum and plasma portion of a specimen in regions occluded by one or more labels. The characterization may be used for Hemolysis, Icterus, and/or Lipemia, or Normal detection. The method captures one or more images of a labeled specimen container including a serum or plasma portion, processes the one or more images to provide segmentation data and identification of a label-containing region, and classifying the label-containing region with a convolutional neural network (CNN) to provide a pixel-by-pixel (or patch-by-patch) characterization of the label thickness count, which may be used to adjust intensities of regions of a serum or plasma portion having label occlusion. Optionally, the CNN can characterize the label-containing region as one of multiple pre-defined label configurations. Quality check modules and specimen testing apparatus adapted to carry out the method are described, as are other aspects.

Abstract: A method of characterizing a serum and plasma portion of a specimen in regions occluded by one or more labels. The characterization may be used for determining Hemolysis (H), Icterus (I), and/or Lipemia (L), or Normal (N) of a serum or plasma portion of a specimen. The method includes capturing one or more images of a labeled specimen container including a serum or plasma portion, processing the one or more images with a convolutional neural network to provide a determination of Hemolysis (H), Icterus (I), and/or Lipemia (L), or Normal (N). In further embodiments, the convolutional neural network can provide N-Class segmentation information. Quality check modules and testing apparatus adapted to carry out the method are described, as are other aspects.

Abstract: A method of characterizing a serum and plasma portion of a specimen in regions occluded by one or more labels. The characterization method may be used to provide input to an HILN (H, I, and/or L, or N) detection method. The characterization method includes capturing one or more images of a labeled specimen container including a serum or plasma portion from multiple viewpoints, processing the one or more images to provide segmentation data including identification of a label-containing region, determining a closest label match of the label-containing region to a reference label configuration selected from a reference label configuration database, and generating a combined representation based on the segmentation information and the closest label match. Using the combined representation allows for compensation of the light blocking effects of the label-containing region. Quality check modules and testing apparatus and adapted to carry out the method are described, as are other aspects.

Abstract: Systems and methods for image classification include receiving imaging data of in-vivo or excised tissue of a patient during a surgical procedure. Local image features are extracted from the imaging data. A vocabulary histogram for the imaging data is computed based on the extracted local image features. A classification of the in-vivo or excised tissue of the patient in the imaging data is determined based on the vocabulary histogram using a trained classifier, which is trained based on a set of sample images with confirmed tissue types.

Abstract: Embodiments include a medical imaging device and a method controlling one or more parameters of a medical imaging device. In one embodiment, a method includes receiving image data representing a first image of an object to be imaged using the radiation source and detecting a plurality of positions of respective predetermined features in the first image. Based upon the detected positions, a boundary of an imaging area of the object to be imaged is determined. Based on the determined boundary, one or more parameters of the radiation source unit are controlled.

Abstract: Robust calcification tracking is provided in fluoroscopic imagery. A patient with an inserted catheter is scanned over time. A processor detects the catheter in the patient from the scanned image data. The processor tracks the movement of the catheter. The processor also detects a structure represented in the data. The structure is detected as a function of movement with a catheter. The processor tracks the movement of the structure using sampling based on a previous location of the structure in the patient. The processor may output an image of the structure.

Abstract: A trained recurrent neural network having a set of control policies learned from application of a template dataset and one or more corresponding template deep network architectures may generate a deep network architecture for performing a task on an application dataset. The template deep network architectures may have an established level or performance in executing the task. A deep network based on the deep network architecture may trained to perform the task on the application dataset. The control policies of the recurrent neural network may be updated based on the performance of the trained deep network.

Abstract: In pose estimation from a depth sensor (12), depth information is matched (70) with 3D information. Depending on the shape captured in depth image information, different objects may benefit from more or less pose density from different perspectives. The database (48) is created by bootstrap aggregation (64). Possible additional poses are tested (70) for nearest neighbors already in the database (48). Where the nearest neighbor is far, then the additional pose is added (72). Where the nearest neighbor is not far, then the additional pose is not added. The resulting database (48) includes entries for poses to distinguish the pose without overpopulation. The database (48) is indexed and used to efficiently determine pose from a depth camera (12) of a given captured image.

Abstract: The present embodiments relate to generating synthetic depth data. By way of introduction, the present embodiments described below include apparatuses and methods for modeling the characteristics of a real-world light sensor and generating realistic synthetic depth data accurately representing depth data as if captured by the real-world light sensor. To generate accurate depth data, a sequence of procedures are applied to depth images rendered from a three-dimensional model. The sequence of procedures simulate the underlying mechanism of the real-world sensor. By simulating the real-world sensor, parameters relating to the projection and capture of the sensor, environmental illuminations, image processing and motion are accurately modeled for generating depth data.