Method and system for automatic view planning for cardiac magnetic resonance imaging acquisition
Inventors
Lu, Xiaoguang • Guehring, Jens • Jolly, Marie-Pierre • Georgescu, Bogdan • Hayes, Carmel • Speier, Peter • Schmidt, Michaela • Bi, Xiaoming • Kroeker, Randall • Comaniciu, Dorin • Mueller, Edgar
Assignees
Siemens Canada Ltd • Siemens Healthineers AG • Siemens Corp • Siemens Medical Solutions USA Inc • National Institutes of Health NIH
Publication Number
US-8948484-B2
Publication Date
2015-02-03
Expiration Date
2031-11-10
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Abstract
A method and system for automated view planning for cardiac magnetic resonance imaging (MRI) acquisition is disclosed. The method and system automatically generate a full scan prescription using a single 3D MRI volume. The left ventricle (LV) is segmented in the 3D MRI volume. Cardiac landmarks are detected in the automatically prescribed slices. A full scan prescription, including a short axis stack and 2-chamber, 3-chamber, and 4-chamber views, is automatically generated based on cardiac anchors provided by the segmented left ventricle and the detected cardiac landmarks in the 3D MRI volume.
Core Innovation
The invention provides a method and system for automated view planning for cardiac magnetic resonance imaging (MRI) acquisition. It automatically generates a full scan prescription including a short axis stack and standard long axis views such as 2-chamber, 3-chamber, and 4-chamber views based on a single 3D MRI volume. The process involves segmenting the left ventricle (LV) in the 3D volume and detecting cardiac landmarks to anchor the heart chambers for view prescription.
The problem addressed is the conventional multi-step, operator-dependent, and time-consuming approach in cardiac MRI acquisition planning. Typically, view planning requires multiple localizer slices and manual adjustments by experts while the patient remains in the scanner. This invention addresses the clinical challenge of providing automatic and fast view planning for cardiac MRI acquisition, improving efficiency and reducing reliance on operator expertise.
The invention leverages machine learning techniques to localize and delineate cardiac anatomies in a 3D volume and detect cardiac landmarks to define scanning planes automatically from a single 3D MRI volume. This includes segmenting the LV by fitting a mesh model, detecting LV pose using marginal space learning classifiers, prescribing a short axis stack based on LV base and apex points, determining 3-chamber view plane using LV anatomical points, and detecting landmarks in a mid-ventricular short axis slice to prescribe 2-chamber and 4-chamber view planes.
Claims Coverage
The patent claims include three independent claims covering a method, an apparatus, and a computer-readable medium for automated cardiac MRI view planning from a single 3D MRI volume. These independent claims focus on segmenting the left ventricle and automatically generating scan prescriptions based on cardiac anchor points and landmarks.
Automatic segmentation of left ventricle in 3D MRI volume
Segmenting a left ventricle (LV) in the 3D MRI volume using machine learning methods and fitting an LV mesh model to the volume to delineate anatomical structures including the LV endocardium, epicardium, and outflow tract.
Automatic scan prescription generation based on cardiac anchor points
Automatically generating a cardiac MRI scan prescription including prescribing a short axis stack, determining a 3-chamber view scanning plane, detecting landmarks in a reconstructed mid-ventricular short axis slice, and determining 2-chamber and 4-chamber view scanning planes based on the detected landmarks.
LV pose estimation using marginal space learning
Estimating the LV position, orientation, and scale in the 3D MRI volume through a series of learned detectors using marginal space learning to fit the LV mesh model accurately.
Landmark detection using trained classifiers in mid-ventricular slice
Detecting anterior and posterior right ventricle (RV) insertion points and an RV lateral point in a reconstructed mid-ventricular short axis slice using trained detectors that utilize bounding box parameterization and context modeling.
Scan plane determination from LV and detected landmarks
Determining the 3-chamber view plane from LV apex, base, and outflow tract points; determining the 2-chamber view plane parallel to line connecting RV insertion points crossing LV apex and blood pool center; and determining the 4-chamber view plane crossing LV blood pool center, apex, and RV lateral point.
Automated acquisition of 3D MRI volume
Acquiring the 3D MRI volume as a single breath-hold acquisition using parallel MR imaging techniques to facilitate fast and high-quality image capture.
Performing diagnostic cardiac MRI scan using generated prescription
Utilizing the automatically generated scan prescription to perform a cardiac MRI diagnostic scan capturing the prescribed views with reduced operator intervention.
The claims collectively cover a fully automated system and method for cardiac MRI view planning from a single 3D MRI volume by segmenting the LV, estimating cardiac anchor points, detecting landmarks, prescribing standard cardiac views, and performing diagnostic scans based on these automated prescriptions.
Stated Advantages
Reduces operator dependence and manual time consumption in planning cardiac MRI views by automating the view planning process.
Allows for fast and fully automated generation of scan prescriptions from a single 3D MRI volume, improving clinical workflow efficiency.
Leverages machine learning to robustly segment and localize cardiac anatomy amidst complex structures and population variability.
Enables acquisition of diagnostic quality cardiac MRI views including short axis stack and standard long axis views without multiple localizer acquisitions.
Documented Applications
Automatic view planning for cardiac magnetic resonance imaging acquisition, facilitating clinical diagnosis, prognosis, and therapeutic decision-making.
Use in cardiac MRI systems to acquire preset Views such as short-axis stack, 2-chamber, 3-chamber, and 4-chamber views automatically based on a single 3D MRI volume.
Implementation in computer systems to segment the left ventricle and generate scanning prescriptions to improve MRI scan planning and operation.
Interested in licensing this patent?