A method includes obtaining, at a local imaging system, projection data for an object representing an intensity of radiation detected along a plurality of rays through the object using a first set of imaging parameters; transmitting an image quality dataset related to the obtained projection data to a remote server; generating, via the remote server, localized restoration information based on the received image quality dataset; transferring the localized restoration information from the remote server to the local imaging system; and updating the local imaging system using the localized restoration information.

Hybrid deep learning systems and methods allow for detecting anomalies in objects, such as electrical printed circuit board (PCB) components, based on image data. In one or more embodiments, a hybrid deep learning model comprises a Graph Attention Network (GAT) that uses spatial properties of the PCB components to extract latent semantic information and generate an output set of hidden representations. The GAT treats each of the electrical components as a node and each connection between them as edges in a graph. The hybrid system further comprises a Convolutional Neural Network (CNN) that uses pixel data to obtain its own output set of hidden representations. The hybrid deep learning model concatenates both sets to detect anomalies that may be present on the PCB.

An interactive 3D cursor facilitates selection and manipulation of a three-dimensional volume from a three-dimensional image. The selected volume image may be transparency-adjusted and filtered to remove selected tissues from view. Qualitative and quantitative analysis of tissues in a selected volume may be performed. Location indicators, annotations, and registration markers may be overlaid on selected volume images.

A reconstruction unit generates a plurality of tomographic images representing a plurality of tomographic planes of a subject by reconstructing a plurality of projection images acquired by performing tomosynthesis imaging. A synthesis unit synthesizes the plurality of tomographic images to generate a composite two-dimensional image. A display control unit displays the composite two-dimensional image on a display, and in a case where one tissue of a first tissue and a second tissue that are present in the subject in association with each other is selected in the displayed composite two-dimensional image, emphasizes and displays the selected one tissue and the other tissue associated with the selected one tissue.

A text generation unit (14) generates a plurality of texts which describe properties of feature portions and are different from each other for at least one feature portion included in an image. A display control unit (15) performs control such that each of the plurality of texts is displayed on a display unit.

Methods and systems for training computer-aided condition detection systems. One method includes receiving a plurality of images for a plurality of patients, some of the images including an annotation associated with a condition; iteratively applying a first deep learning network to each of the images to produce an attention map, a feature map, and an image-level probability of the condition for each of the images; iteratively applying a second deep learning network to each feature map produced by the first network to produce a plurality of outputs; training the first network based on the attention map produced for each image; and training the second network based on the output produced for each of the patients. The second network includes a plurality of convolution layers and a plurality of convolutional long short-term memory (LSTM) layers. Each of the outputs includes a patient-level probability of the condition for one of the patients.

A system is configured to perform operations includes accessing a set of model points of a model of an anatomic structure of a patient, the model points being associated with a model space. A set of measured points of the anatomic structure of the patient are collected, the measured points being associated with a patient space. The set of model points are registered to the set of measured points using a first set of initial parameters to generate a first transformation. One or more sets of perturbed initial parameters are generated based on the first set of initial parameters. One or more perturbed registration processes are performed to register the set of model points to the set of measured points using the one or more sets of perturbed initial parameters respectively to generate corresponding perturbed transformations. A registration quality indicator is generated based on the first transformation and the one or more perturbed transformations.

Disclosed herein are systems and methods for vertebral bone segmentation and vertebral bone enhancement in medical images.

An analysis unit (13) specifies properties of a feature portion included in an image for each of a plurality of predetermined property items. A text generation unit (14) generates a plurality of texts such that a combination of the property items is different between the plurality of texts. In a case where any one of the plurality of texts is selected, an association data generation unit (16) generates association data in which a selection item, which is a property item corresponding to the property described in the selected text, and a property specifying result are associated with each other. In a case where the specified property and the property specifying result included in the association data match, the text generation unit (14) generates a priority text describing the property specified for the same property item as the selection item associated with the property specifying result as one of the plurality of texts.

Systems and methods for computed tomography imaging are provided. In one embodiment, a method includes acquiring an image, inputting the image to a machine learning model to generate a denoised image, the machine learning model trained with a loss function that weights variance differently from bias, and outputting the denoised image. In this way, structural details in denoised CT images may be improved while maintaining textural information in the denoised images.