Abstract: Systems and method are described for medical image segmentation. A medical image of a patient in a first domain is received. The medical image comprises one or more anatomical structures. A synthesized image in a second domain is generated from the medical image of the patient in the first domain using a generator of a task driven generative adversarial network. The one or more anatomical structures are segmented from the synthesized image in the second domain using a dense image-to-image network of the task driven generative adversarial network. Results of the segmenting of the one or more anatomical structures from the synthesized image in the second domain represent a segmentation of the one or more anatomical structures in the medical image of the patient in the first domain.

Abstract: An image segmentation method includes generating a CTN (contour transformer network) model for image segmentation, where generating the CTN model includes providing an annotated image, the annotated image including an annotated contour, providing a plurality of unannotated images, pairing the annotated image to each of the plurality of unannotated images to obtain a plurality of image pairs, feeding the plurality of image pairs to an image encoder to obtain a plurality of first-processed image pairs, and feeding the plurality of first-processed image pairs to a contour tuner to obtain a plurality of second-processed image pairs.

Abstract: The present disclosure provides a computer-implemented method, a device, and a computer program product for radiographic bone mineral density (BMD) estimation. The method includes receiving a plain radiograph, detecting landmarks for a bone structure included in the plain radiograph, extracting an ROI from the plain radiograph based on the detected landmarks, estimating the BMD for the ROI extracted from the plain radiograph by using a deep neural network.

Abstract: A method for performing computer-aided diagnosis (CAD) based on a medical scan image includes: pre-processing the medical scan image to produce an input image, a flipped image, and a spatial alignment transformation corresponding to the input image and the flipped image; performing Siamese encoding on the input image to produce an encoded input feature map; performing Siamese encoding on the flipped image to produce an encoded flipped feature map; performing a feature alignment using the spatial alignment transformation on the encoded flipped feature map to produce an encoded symmetric feature map; and processing the encoded input feature map and the encoded symmetric feature map to generate a diagnostic result indicating presence and locations of anatomical abnormalities in the medical scan image.

Abstract: Systems and methods are provided for determining a set of imaging parameters for an imaging system. A selection of an image is received from a set of images. A modification of certain quality measures is received for the selected image. The modified selected image is mapped to a set of imaging parameters of an imaging system based on the certain quality measures using a trained Deep Reinforcement Learning (DRL) agent.

Abstract: For three-dimensional rendering, a machine-learnt model is trained to generate representation vectors for rendered images formed with different rendering parameter settings. The distances between representation vectors of the images to a reference are used to select the rendered image and corresponding rendering parameters that provides a consistency with the reference. In an additional or different embodiment, optimized pseudo-random sequences are used for physically-based rendering. The random number generator seed is selected to improve the convergence speed of the renderer and to provide higher quality images, such as providing images more rapidly for training compared to using non-optimized seed selection.

Abstract: A fracture detection method executed by an electronic device is provided. The fracture detection method includes obtaining a to-be-detected image; using a Fully Convolutional Networks (FCN) model to process the to-be-detected image to obtain a fracture probability map of the to-be-detected image; performing a maximum pooling process on the fracture probability map to obtain a first fracture probability; extracting Regions of Interests (ROIs) of the to-be-detected image based on the FCN model; inputting the ROIs into a classification model to obtain a second fracture probability; calculating a product of the first fracture probability and the second fracture probability as a probability of a fracture phenomenon in the to-be-detected image. The present disclosure combines the FCN model and the ROIs to realize an automatic fracture detection, and the accuracy is higher. A device employing the method is also disclosed.

Abstract: A fracture detection method executed by an electronic device is provided. The fracture detection method includes obtaining a to-be-detected image; using a Fully Convolutional Networks (FCN) model to process the to-be-detected image to obtain a fracture probability map of the to-be-detected image; performing a maximum pooling process on the fracture probability map to obtain a first fracture probability; extracting Regions of Interests (ROIs) of the to-be-detected image based on the FCN model; inputting the ROIs into a classification model to obtain a second fracture probability; calculating a product of the first fracture probability and the second fracture probability as a probability of a fracture phenomenon in the to-be-detected image. The present disclosure combines the FCN model and the ROIs to realize an automatic fracture detection, and the accuracy is higher. A device employing the method is also disclosed.

Abstract: The disclosure relates to a method of determining a transformation between coordinate frames of sets of image data. The method includes receiving a model of a structure extracted from first source image data, the first source image data being generated according to a first imaging modality and having a first data format, wherein the model has a second data format, different from the first data format. The method also includes determining, using an intelligent agent, a transformation between coordinate frames of the model and first target image data, the first target image data being generated according to a second imaging modality different to the first imaging modality.

Abstract: A correspondence between a source image and a reference image is determined. A generative model corresponds to a prior probability distribution of deformation fields, each deformation field corresponding to a respective coordinate transformation. A conditional model generates a style transfer probability distribution of reference images, given a source image and a deformation field. The first image data is the source image, and the second image data is the reference image. An initial first deformation field is determined. An update process is iteratively performed until convergence to update the first deformation field, to generate a converged deformation field representing the correspondence between the source image and the reference image.

Abstract: A method and system for 3D/3D medical image registration. A digitally reconstructed radiograph (DRR) is rendered from a 3D medical volume based on current transformation parameters. A trained multi-agent deep neural network (DNN) is applied to a plurality of regions of interest (ROIs) in the DRR and a 2D medical image. The trained multi-agent DNN applies a respective agent to each ROI to calculate a respective set of action-values from each ROI. A maximum action-value and a proposed action associated with the maximum action value are determined for each agent. A subset of agents is selected based on the maximum action-values determined for the agents. The proposed actions determined for the selected subset of agents are aggregated to determine an optimal adjustment to the transformation parameters and the transformation parameters are adjusted by the determined optimal adjustment.

Abstract: A method and system for 3D/3D medical image registration. A digitally reconstructed radiograph (DRR) is rendered from a 3D medical volume based on current transformation parameters. A trained multi-agent deep neural network (DNN) is applied to a plurality of regions of interest (ROIs) in the DRR and a 2D medical image. The trained multi-agent DNN applies a respective agent to each ROI to calculate a respective set of action-values from each ROI. A maximum action-value and a proposed action associated with the maximum action value are determined for each agent. A subset of agents is selected based on the maximum action-values determined for the agents. The proposed actions determined for the selected subset of agents are aggregated to determine an optimal adjustment to the transformation parameters and the transformation parameters are adjusted by the determined optimal adjustment.

Abstract: Methods and systems for image registration using an intelligent artificial agent are disclosed. In an intelligent artificial agent based registration method, a current state observation of an artificial agent is determined based on the medical images to be registered and current transformation parameters. Action-values are calculated for a plurality of actions available to the artificial agent based on the current state observation using a machine learning based model, such as a trained deep neural network (DNN). The actions correspond to predetermined adjustments of the transformation parameters. An action having a highest action-value is selected from the plurality of actions and the transformation parameters are adjusted by the predetermined adjustment corresponding to the selected action. The determining, calculating, and selecting steps are repeated for a plurality of iterations, and the medical images are registered using final transformation parameters resulting from the plurality of iterations.

Abstract: Systems and methods are provided for identifying pathological changes in follow up medical images. Reference image data is acquired. Follow up image data is acquired. A deformation field is generated for the reference image data and the follow up data using a machine-learned network trained to generate deformation fields describing healthy, anatomical deformation between input reference image data and input follow up image data. The reference image data and the follow up image data are aligned using the deformation field. The co-aligned reference image data and follow up image data are analyzed for changes due to pathological phenomena.

Abstract: For three-dimensional rendering, a machine-learnt model is trained to generate representation vectors for rendered images formed with different rendering parameter settings. The distances between representation vectors of the images to a reference are used to select the rendered image and corresponding rendering parameters that provides a consistency with the reference. In an additional or different embodiment, optimized pseudo-random sequences are used for physically-based rendering. The random number generator seed is selected to improve the convergence speed of the renderer and to provide higher quality images, such as providing images more rapidly for training compared to using non-optimized seed selection.

Abstract: A method for performing 2D/3D registration includes acquiring a 3D image. A pre-contrast 2D image is acquired. A sequence of post-contrast 2D images is acquired. A 2D image is acquired from a second view. The first view pre-contrast 2D image is subtracted from each of the first view post-contrast 2D images to produce a set of subtraction images. An MO image is generated from the subtraction images. A 2D/3D registration result is generated by optimizing a measure of similarity between a first synthetic 2D image and the MO image and a measure of similarity between a second synthetic image and the intra-operative 2D image from the second view by iteratively adjusting an approximation of the pose of the patient in the synthetic images and iterating the synthetic images using the adjusted approximation of the pose.

Abstract: For three-dimensional rendering, a machine-learnt model is trained to generate representation vectors for rendered images formed with different rendering parameter settings. The distances between representation vectors of the images to a reference are used to select the rendered image and corresponding rendering parameters that provides a consistency with the reference. In an additional or different embodiment, optimized pseudo-random sequences are used for physically-based rendering. The random number generator seed is selected to improve the convergence speed of the renderer and to provide higher quality images, such as providing images more rapidly for training compared to using non-optimized seed selection.

Abstract: A method for performing 2D/3D registration includes acquiring a 3D image. A pre-contrast 2D image is acquired. A sequence of post-contrast 2D images is acquired. A 2D image is acquired from a second view. The first view pre-contrast 2D image is subtracted from each of the first view post-contrast 2D images to produce a set of subtraction images. An MO image is generated from the subtraction images. A 2D/3D registration result is generated by optimizing a measure of similarity between a first synthetic 2D image and the MO image and a measure of similarity between a second synthetic image and the intra-operative 2D image from the second view by iteratively adjusting an approximation of the pose of the patient in the synthetic images and iterating the synthetic images using the adjusted approximation of the pose.

Abstract: A correspondence between a source image and a reference image is determined. A generative model corresponds to a prior probability distribution of deformation fields, each deformation field corresponding to a respective co-ordinate transformation. A conditional model generates a style transfer probability distribution of reference images, given a source image and a deformation field. The first image data is the source image, and the second image data is the reference image. An initial first deformation field is determined. An update process is iteratively performed until convergence to update the first deformation field, to generate a converged deformation field representing the correspondence between the source image and the reference image.

Abstract: Pose of a probe is detected in x-ray medical imaging. Since the TEE probe is inserted through the esophagus of a patient, the pose is limited to being within the esophagus. The path of the esophagus is determined from medical imaging prior to the intervention. During the intervention, the location in 2D is found from one x-ray image at a given time. The 3D probe location is provided by assigning the depth of the esophagus at that 2D location to be the depth of the probe. A single x-ray image may be used to determine the probe location in 3D, allowing for real-time pose determination without requiring space to rotate a C-arm during the intervention.