International Business Machines Corporation

Merative US L.p.

Sti Medical Systems, Inc.

Sun Young Park

Dustin Michael Sargent

Mark D Bronkalla

Murray A Reicher

Marwan Sati

Maria Victoria Sainz De Cea

Dusty Sargent

David Joseph Richmond

Amin Katouzian

James G Thompson

Ulf Peter Gustafsson

David Richmond

Kourosh Jafari-Khouzani

Srivenkata Laksh Gantikota

Puja Gupta

Hui Shen

Dale Seegmiller Maudlin

Methods and systems for optimizing user interaction with a software application. One system includes an electronic processor configured to receive a collection of interaction data including a plurality of interaction sequences and data associated with each of the plurality of interaction sequences, determine a performance metric for each of the plurality of interaction sequences, and train, with the collection of interaction data and the performance metric determined for each of the plurality of interaction sequences, an artificial intelligence ('AI') model using supervised learning. The electronic processor is also configured to receive a current interaction sequence of a user for the software application and generate, via the AI model as applied to the interaction pattern of the user, a recommendation, for display within a user interface, for a modified user interaction pattern of the user for the software application.

User performance analysis and correction for S/W

An embodiment for learning-based automatic selection of artificial intelligence applications. The embodiment may receive a user request for an exam, the user request including exam information. The embodiment may automatically identify an exam type cluster corresponding to the received exam information. The embodiment may automatically detect applicable AI applications corresponding to the identified exam type cluster. The embodiment may automatically run each applicable AI application on a series of relevant test sets to generate a score for each applicable AI application. The embodiment may automatically recommend to a user a highest-scoring applicable AI application.

Learning-based automatic selection of AI applications

A computer system identifies a medical condition in a patient. A trained machine learning image generator is used to generate a set of training images based on three-dimensional patient imaging data, wherein each training image is labeled with a projection angle of the corresponding two-dimensional projection. Using the set of training images, a machine learning image classifier model is trained to identify patient rotation angles in x-ray images. X-ray images are processed with the machine learning image classifier model to identify patient rotation angles. A machine learning medical condition classifier model is trained to identify a medical condition using the labeled x-ray images. The machine learning medical condition classifier model determines an indication of the medical condition in a patient's x-ray image. Embodiments of the present invention further include a method and program product for identifying a medical condition in a patient in substantially the same manner described above.

Classification of medical images using machine learning to account for body orientation

A system, method, and computer program product for treatment planning are disclosed. The system includes at least one processing component, at least one memory component, a training module, a retrieval module, and a plan generator. The training module generates hash codes by hashing features from data sources with data source-specific hash functions, and generates superclass hash codes by hashing the generated hash codes with at least one superclass hash function. The retrieval module extracts features from case data, and locates features from the data sources that are similar to the extracted features. The plan generator calculates outcome probabilities for the case data based on known outcomes associated with the located features.

Treatment planning based on multimodal case similarity

Methods and systems and computer readable media are provided for processing digitized radiology images of a patient's tissue to perform a tissue density assessment. A set of radiology images is analyzed with a tissue density image classifier, wherein the tissue density classifier includes a plurality of classifiers. A features vector is generated, for each processed image in the set of radiology images, based on the processing. The features vector is provided, for each processed image in the set of radiology images, to a tissue density classifier to obtain an output. A final tissue density score is assigned based on the output of the tissue density classifier.

Identifying high tissue density in images using machine learning

A method for preemptively generating and rendering a view of a computerized image is provided. The method may include determining rendering parameters for the computerized image by predicting an action to be performed on the computerized image, wherein the action modifies a view of the computerized image, and wherein the determined rendering parameters are based on the modified view of the computerized image. The method may further include preemptively rendering the view of the computerized image that is based on the determined rendering parameters before the action is performed on the computerized image.

Predictive rendering of an image

One or more processors may identify a missing image in the set of multi-view images. Each of the images is associated with a particular view type. The one or more processors may generate, utilizing a replacement AI model, a replacement image for the missing image in the set of multi-view images. The replacement image is generated utilizing an AI model trained to generate a replacement image using training images from two or more time-adjacent sets of images. The one or more processors may identify a duplicate image in the set of multi-view images. The one or more processors may generate, utilizing a quality generative AI model, a characteristic improved image based on the duplicate image for the set of multi-view images. The one or more processors may output the replacement image and the characteristic improved image.

Study-level multi-view processing system

Methods and systems are provided for managing identifying information for an entity. The identifying information of the entity embedded in or associated with a digital image is detected, wherein the identifying information is selected from the group consisting of: text information and image information corresponding to one or more features of an entity. The text information may be removed from the digital image. The image information may be replaced with one or more computer generated synthetic images, wherein the computer generated synthetic images are based on a natural appearance of the digital image. The synthetic content, which may be generated by a GAN, is based on a natural appearance of the image. The medical image may also contain PHI in text-based fields associated with private tags/fields, which are automatically identified and removed using the systems and methods provided herein.

Automatic detection and replacement of identifying information in images using machine learning

Topological evolution of a lesion within a time series of medical imagery is provided. In various embodiments, a time series of medical images is read. Each of the images depicts a subject anatomy and a lesion. The lesion has a size and a contour within each of the medical images. At least one anatomical label is read for the subject anatomy within each of the plurality of images. Based upon the contour of the lesion within each of the medical images and based on the at least one anatomical label, a further contour of the lesion is predicted outside of the time series.

Topological evolution of tumor imagery

An embodiment of the invention may include a method, computer program product and computer system for image artifact removal. The method, computer program product and computer system may include computing device which may receive a primary image and analyze the primary image for global artifacts and local artifacts. The computing device may, in response to identifying a global artifact in the primary image, generate a secondary image with the global artifact removed utilizing a first generative adversarial network. The computing device may, in response to identifying a local artifact in the primary image, generate a patch with the local artifact removed utilizing a second generative adversarial network. The computing device may generate a hybrid image containing a reduction of global artifacts and a reduction of local artifacts by combining the secondary image and the patch utilizing a hybrid generative adversarial network.

Artifact removal in medical imaging

An embodiment of the invention may include a method, computer program product and computer system for three-dimensional medical image generation. The method, computer program product and computer system may include computing device which may receive a first three-dimensional medical image of a first patient from a first period of time, a two-dimensional medical image of the first patient from a second period of time and a plurality of three-dimensional medical images for a plurality of second patients. The computing device may input the three-dimensional medical image of the first patient, the two-dimensional medical image of the first patient and the plurality of three-dimensional medical images for a plurality of second patients into a generative adversarial network (GAN). The computing device may generate a synthetic three-dimensional medical image for the first patient based on the two-dimensional medical image from the second period of time utilizing the GAN.

Three-dimensional medical image generation

A system for generating simulated body parts for images may include a body part recognition convolutional neural network (CNN) to recognize a body part in an input image. The body part recognition CNN may be trained using first training data including training images including body parts contained in the input image being identified. The system may also include a body part generative adversarial network (GAN) to complete an image of the body part in the input image based on a body part identification output by the body part recognition CNN. The body part GAN may be trained using second training data including at least partial training images.

Generating simulated body parts for images

Machine logic (for example, software) for registering multiple medical images, each showing a common lesion, with each other. In performing this registration, registration points are chosen to be both: (i) outside of image portion that is potentially compromised by the lesion (in any of the multiple images); and (ii) as close to the lesion as possible. However, in at least one of the images the extent of the lesion is not known&#x2014;so, in order to accommodate this uncertainty about the lesion boundaries, lesion predicting machine logic rules are used to predict the size, shape and/or location of the lesion. Machine learning is used to intermittently adjust and improve the lesion predicting machine logic rules.

Medical image registration guided by target lesion

Provided are techniques for automated software testing using simulated user personas. A request to test software is received. Job roles, user software activities for the software to be tested, and objectives are automatically identified using a first machine learning model. A test operation sequence using the job roles, the user software activities, and the objectives to test the software is generated using a second machine learning model. The test operation sequence is executed to simulate different users having different job roles using the software with the user software activities to achieve the objectives.

Automated software testing using simulated user personas

A method, system and computer program product for determining changes in breast density. A generative adversarial network is trained to predict an appearance of a mammogram image for a patient's current examination based on mammogram images assigned labels of a first type of density classification. An appearance of a mammogram image for a patient's current examination is predicted using the generative adversarial network based on a mammogram image(s) obtained from the patient's prior examination assigned labels of the first type of density classification. A comparison is made between the predicted and actual mammogram images for the patient's current examination to determine if there is a difference between scores assigned to the predicted and actual mammogram images, and if so, whether such difference can be attributed to the subjective assessment by different physicians or changes in the standards of density classification or is due to a real change in breast density.

Growing community of inventors

San Diego, CA, United States of America

Sun Young Park