A medical report natural language model includes an artificial neural network implemented via the processor and is trained based a plurality of medical reports wherein each of the medical reports is mapped to at least one medical code of a plurality of medical codes and further based a plurality of medical condition terms from a plurality of alias mapping pairs, wherein each of the plurality of medical condition terms are unique, wherein each of the plurality of medical condition terms indicates a corresponding medical condition and wherein each of the plurality of alias mapping pairs includes a one of the plurality of medical condition terms and a corresponding single one of the plurality of medical codes that is a deterministic function of the one of the plurality of medical condition terms and wherein the plurality of alias mapping pairs includes two or more alias mapping pairs that map a corresponding two or more of the plurality of medical condition terms to a same one of the plurality of medical codes. A system operates by determining a first medical code of the plurality of medical codes utilizing the medical report natural language model on the first medical report by: recognizing, via the artificial neural network of the medical report natural language model, a first medical condition term of the plurality of medical condition terms in the first medical report; and determining the first medical code of the plurality of medical codes, based on the plurality of alias mapping pairs, wherein the determining the first medical code includes identifying a first alias mapping pair of the plurality of alias mapping pairs that pairs the first medical condition term deterministically to the first medical code; and mapping the first medical code of the plurality of medical codes and the first medical condition term of the plurality of medical condition terms to a first medical scan corresponding to the first medical report.

Conventionally, systems and methods have been provided for manual annotation of anatomical landmarks in digital radiography (DR) images. Embodiments of the present disclosure provides system and method for anatomical landmark detection and identification from DR images containing severe skeletal deformations. More specifically, motion artefacts and exposure are filtered from an input DR image to obtain a pre-processed DR image and probable/candidate anatomical landmarks comprised therein are identified. These probable candidate anatomical landmarks are assigned a score. A subset of the candidate anatomical landmarks (CALs) is selected as accurate anatomical landmarks based on comparison of the score with a pre-defined threshold performed by a trained classifier. Position of remaining CALs may be fine-tuned for classification thereof as accurate anatomical landmarks or missing anatomical landmarks. The CALs may be further fed to the system for checking misalignment of any of the CALs and correcting the misaligned CALs.

A system operates by receiving a plurality of medical scans, a plurality of medical labels corresponding to the plurality of medical scans and a plurality of confidence scores corresponding to the plurality of medical labels, wherein each of the plurality of medical labels correspond to one of a set of abnormality classes and wherein the plurality of confidence scores indicate a quantified representation of uncertainty generated via natural language processing of a plurality of medical reports corresponding to the plurality of medical labels; generating a computer vision model by training on the plurality of medical scans and the plurality of medical labels, wherein a model confidence of the computer vision model is calibrated based on the plurality of confidence scores; receiving a new medical scan; generating inference data corresponding to the new medical scan utilizing the computer vision model, wherein the inference data indicates an inferred abnormality in the new medical scan and the model confidence corresponding to the inferred abnormality; and facilitating display of the inference data via an interactive interface.

Systems and methods are dislocated that include a method of determining the actual position and pose of a known three-dimensional construct utilizing at least four discrete shapes formed by fiducials of the construct shown in a two-dimensional (2D) image of the construct. The method includes identifying at least four fiducial shadows in the 2D image that correspond to the fiducials of the construct. The method also includes correlating the discovered at least four fiducial shadows with their respective locations on the construct. The method further includes determining a spatial relationship between the 2D image and the construct by determining a focal point of a source of the image relative to the 2D image via the discovered at least four fiducial shadows and pre-determined mutual separation distances between the fiducials of the construct corresponding thereto. The method also includes determining a spatial relationship between the 2D image and the construct.

This patent includes a method and apparatus for assigning a three-dimensional coordinate system to a segmented structure of a three-dimensional imaging examination of a tangible volume containing discrete structures. The boundary of a segmented structure, which corresponds to a discrete structure is determined through a segmentation algorithm. The assigned three-dimensional coordinate system has an origin and a set of coordinate axes. This allows each voxel within the segmented structure of the three-dimensional imaging examination to have a precise coordinate. This provides improved longitudinal analysis and more precise localization within the segmented structure.

A medical scan natural language analysis system is operable to generate a medical report natural language model based on a selected set of medical reports of the plurality of medical reports and the at least one medical code mapped to each of the selected set of medical reports. A medical report that is not included in the selected set is received via a network. A medical code is determined by utilizing the medical report natural language model on the first medical report. The medical code is mapped to a medical scan corresponding to the medical report.

The present invention discloses a method and a system of vertebral compression fracture detection. The method of vertebral compression fracture detection includes: recombining a plurality of anatomical images captured in at least a spine segment of a target individual into a 3D image; using a multi-planar reconstruction method to reformat the 3D image to obtain at least one sagittal reformatted image; using a classification model to determine whether the sagittal reformatted image covers the middle section of the vertebral column or not; using a vertebral detection method to detect each vertebral body in the sagittal reformatted image covering the middle section of the vertebral column; using a keypoint localization method to localize a plurality of keypoints of each vertebral body which was detected in the sagittal reformatted image; evaluating the compression fracture grade of each vertebral body in the sagittal reformatted image.

An abnormality detection system is a system that detects, based on a medical image, whether a prespecified anatomical structure included in the medical image is abnormal, and includes: an obtainer that obtains block information indicating one or more blocks that are each a region in which pixels representing the anatomical structure are connected in the medical image; and a determiner that determines and outputs whether the anatomical structure is abnormal, based on the number of blocks indicated by the block information obtained by the obtainer, wherein the determiner determines that the anatomical structure is not abnormal when the number of blocks indicated by the block information is 1, and determines that the anatomical structure is abnormal when the number of blocks indicated by the block information is 2 or more and the two or more blocks indicated by the block information satisfy a predetermined condition.

A radiographic image capturing apparatus includes: a two-dimensional array of radiation detecting elements that generate one or more electric charges in proportion to a dose of incident radiation; and a control unit that performs control, at least, to read the electric charges from the radiation detecting elements and generate image data. An image capturing mode is switchable between a controlled mode in which the radiographic image capturing apparatus is under control of an external console and a stand-alone mode in which the radiographic image capturing apparatus autonomously performs image capturing without the control of the console. When the image capturing mode is switched to the controlled mode or the stand-alone mode, the control unit changes its own process so as to meet the switched image capturing mode.

A non-transitory computer-readable storage medium storing a program causes a computer to perform an analysis process based on a radiation moving image in which a dynamic state of a specific site of a subject is captured. The program includes the analysis process in which, an analysis is performed based on the radiation moving image wherein when a plane in which the specific site is movable is to be a movable plane, the radiation moving image is obtained by irradiating radiation on the specific site in a state in which the radiation is orthogonal to the movable plane.