Abstract: Systems, methods, and computer programs disclosed herein relate to training a machine learning model to generate multimodal representations of objects, and to the use of said representations for predictive purposes.

Abstract: The present invention relates to the technical field of producing artificial contrast-enhanced radiological images by way of machine learning methods.

Abstract: A method for providing a prediction of a representation of an examination region that was generated using a medical image technique involving a contrast agent may include receiving a first representation in frequency space of an examination region of an examination object, receiving a second representation in the frequency space of the examination region of the examination object, providing the first representation and the second representation as an input to a predictive machine learning model that is configured to provide, as an output, a prediction of a representation in the frequency space of the examination region with an amount of the contrast agent administered during a medical imaging technique, receiving the output of the predictive machine learning model based on the input, and converting the output of the predictive machine learning model to a representation in real space of the examination region of the examination object.

Abstract: A system and method for training a machine learning module to provide classification and localization information for an image study. The method includes receiving a current image study. The method includes applying the machine learning module to the current image study to generate a classification result including a prediction for one or more class labels for the current image study using User Interface 104 a classification module of the machine learning module. The method includes receiving, via a user interface, a user input indicating a spatial location corresponding to a predicted class label. The method includes training a localization module of the machine learning module using the user input indicating the spatial location corresponding to the predicted class label.

Abstract: Presented are concepts for obtaining a confidence measure for a machine learning model. One such concept process input data with the machine learning model to generate a primary result. It also generate a plurality of modified instances of the input data and processes the plurality of modified instances of the input data with the machine learning model to generate a respective plurality of secondary results. A confidence measure relating to the primary result is determined based on the secondary results.

Abstract: A system and method for training a deep learning network with previously read image studies to provide a prioritized worklist of unread image studies. The method includes collecting training data including a plurality of previously read image studies, each of the previously read image studies including a classification of findings and radiologist-specific data. The method includes training the deep learning neural network with the training data to predict an urgency score for reading of an unread image study.

Abstract: A system and related method for image processing. The system comprises an input (IN) interface for receiving two segmentation maps for an input image. The two segmentation maps (11,12) obtained by respective segmentors, a first segmentor (SEG1) and a second segmentor (SEG2). The first segmentor (SEG1) implements a shape-prior-based segmentation algorithm. The second segmentor (SEG2) implements a segmentation algorithm that is not based on a shape-prior, or at least the second segmentor (SEG2) accounts for one or more shape priors at a lower weight as compared to the first segmentor (SEG1). A differentiator (DIF) configured to ascertain a difference between the two segmentation maps. The system may allow detection of abnormalities.

Abstract: Disclosed herein is a medical system comprising a memory storing machine executable instructions and at least one trained neural network. Each of the at least one neural network is configured for receiving a medical image as input. Each of the at least one trained neural network has been modified to provide hidden layer output. Execution of the machine executable instructions causes the computational system to: receive the medical image; receive the hidden layer output in response to inputting the medical image into each of the at least one trained neural network; provide an anonymized image fingerprint comprising the hidden layer output from each of the at least one trained neural network; and receive an image assessment of the medical image in response to querying a historical image database using the anonymized image fingerprint.

Abstract: The invention provides a method for determining a confidence value for an image segmentation. The method includes obtaining an image, wherein the image comprises a view of an anatomical structure and a model of the anatomical structure is obtained, wherein the model comprises a plurality of nodes. The image is processed to generate a plurality of image segmentation outputs, wherein each image segmentation output comprises a set of values for the view, wherein each value of the set of values is associated with a node of the plurality of nodes of the model. For each node of the model, a confidence value is determined based on the plurality of values corresponding to the node. A confidence map of the anatomical structure is generated based on the confidence value of each node.

Abstract: The present invention relates to a system and a method for automatic verification of a positioning of a medical device with respect to an anatomy of a patient in a medical image. A position of a plurality of reference points in a medical image is detected. Further, a presence and a position of a medical device in the medical image is detected. An expected position of the medical device is determined based on the position of the plurality of reference points, and a measure of a correctness of the positioning of the medical device is provided based on a proximity of the position of the medical device to the expected position of the medical device. The measure of the correctness of the positioning of the medical device is provided.

Abstract: A mechanism for determining a score indicative of a success of a segmentation of a 3D image, i.e. a success score. The mechanism proposes to obtain one or more 2D images of different target views of a target object in the 3D image, by processing a segmentation result of the 3D image. (A view of) each 2D image is classified using an automated classifier. The classification results are used to determine a success score, which may indicate, for example, whether, or how closely, the 3D segmentation result represents a ground truth segmentation result with sufficient accuracy, e.g. for clinical decision making.

Abstract: The invention provides a method for refining a mapped surface mesh of a cardiac chamber. The method includes obtaining a mapped surface mesh of the cardiac chamber anatomy, wherein the mapped surface mesh comprises a central region representing a cardiac chamber and an outer region representing a peripheral cardiac structure connected to the cardiac chamber, and wherein the mapped surface mesh comprises a first view of an anatomical landmark within the cardiac chamber, and obtaining image data of a cardiac chamber anatomy of a subject. The central region of the mapped surface mesh is deformed based on a first segmentation algorithm configured according to one or more predetermined shape- constraints and the outer region of the mapped surface mesh is deformed based on a second segmentation algorithm configured according to the image data, thereby generating a deformed outer region. The deformed central region and the deformed outer region are then combined, thereby generating a refined mapped surface mesh.

Abstract: An ultrasound imaging system includes a processor circuit in communication with an ultrasound transducer configured to receive three-dimensional ultrasound data of an anatomy, and generate a target image corresponding to a target image plane of the anatomy, and a plurality of adjacent images corresponding to image planes adjacent to the target image plane along a simulated motion path. The processor circuit is further configured to display the target image, receive a user input representative of a direction of motion along the simulated motion path, and display an adjacent image of the plurality of adjacent images corresponding to the direction of motion. Accordingly, the user can observe the target image in its spatial context by scanning through the target image and one or more adjacent images on the display as if the ultrasound transducer were being scanned along the simulated motion path.

Abstract: Presented are concepts for initialising a model for model-based segmentation of an image which use specific landmarks (e.g. detected using other techniques) to initialize the segmentation mesh. Using such an approach, embodiments need not be limited to predefined model transformations, but can initialise a segmentation mesh with arbitrary shape. In this way, embodiments may provide for an image segmentation algorithm that not only delivers a robust surface-based segmentation result but also does so for strongly varying target structure variations (in terms of shape).

Abstract: Some embodiments are directed to a segmentation of medical images. For example, a medical image may be registering to multiple atlas images after which a segmentation function may be applied. Multiple segmentation may be fused into a final overall segmentation. The atlas images may be selected on the basis of high segmentation quality or low registration quality.

Abstract: A mechanism for creating/synthesizing realistic training data, for training a machine-learning model, using anatomical knowledge. An anatomical model can be obtained. Information from annotated training data entries (i.e. “ground truth” information), can be used to model the anatomical variation, from the obtained model, in the population of the training data. This anatomical model can then be modified, e.g. incorporating some random factors, in order to generate one or more augmented models of realistic anatomies. The augmented anatomy is then transferred from the model domain to the data entry domain to thereby generate a new data entry or data entries for training a machine-learning model. This latter process can be achieved in various ways, e.g. using GANs, such as CycleGANs and label images, or deformation vector fields.

Abstract: In a method of segmenting a feature in an image, an image product related to the image is provided (102) to a model trained using a machine learning process. An indication of a shape descriptor for the feature in the image is received (104) from the model, based on the image product. The indicated shape descriptor is then used (106) in a model based segmentation, MBS, to initialize the MBS and segment the feature.

Abstract: A system for recording ultrasound images comprises a memory comprising instruction data representing a set of instructions and a processor configured to communicate with the memory and to execute the set of instructions. The set of instructions, when executed by the processor, cause the processor to receive a data stream of two dimensional images taken using an ultrasound transducer and determine from the data stream that a feature of interest is in view of the transducer. The set of instructions further cause the processor to trigger an alert to be sent to a user to indicate that the feature of interest is in view of the transducer, and send an instruction to the transducer to trigger the transducer to capture a three dimensional ultrasound image after a predetermined time interval.

Abstract: The invention provides for a method for switching between fields of view of an ultrasound probe. The method begins by obtaining an anatomical model representing a region of interest of a subject and establishing a first field of view relative to an ultrasonic probe, wherein the first field of view comprises an initial portion of the region of interest. Ultrasound data is then obtained from the first field of view by way of the ultrasonic probe and a first anatomical feature is identified within the first field of view based on the ultrasound data. A location in digital space of the first field of view relative to the anatomical model is determined based on the first anatomical feature. A second field of view is then established based on the anatomical model and the first field of view, wherein the first field of view functions as a reference field of view. The field of view is then switched from the first field of view to the second field of view.

Abstract: Presented are concepts for obtaining a confidence measure for a machine learning model. One such concept process input data with the machine learning model to generate a primary result. It also generate a plurality of modified instances of the input data and processes the plurality of modified instances of the input data with the machine learning model to generate a respective plurality of secondary results. A confidence measure relating to the primary result is determined based on the secondary results.