Abstract: A method is disclosed for determining a boundary surface network of the tubular object. In an embodiment, a representation of the tubular object is initially provided on the basis of image data and local dimension information is provided for points of the representation. A subdivided division structure presentation of the tubular object with division cells is then created, which based on the local dimension information, including a different spatial extent. Finally a boundary surface network is derived on the basis of the division structure presentation. Also described are a boundary surface network determination system and a division structure determination system for performing such a method.

Abstract: Preoperative resection planning is assisted by a computer. Rather than rely on interpolation of the user input, a graph of interconnections is used. The user inputs one or more polylines on one or more two-dimensional views. The polylines are used to assign resection and remnant seeds with a band of unassigned locations. The 2D seeds are used with the graph of interconnections to assign different voxels in the volume, including the unassigned locations, as being part of the resection volume or part of the remnant volume.

Abstract: A method and system for automatically detecting liver lesions in medical image data, such as 3D CT images, is disclosed. A liver region is segmented in a 3D image. Liver lesion center candidates are detected in the segmented liver region. Lesion candidates are segmented corresponding to the liver lesion center candidates, and lesions are detected from the segmented lesion candidates using learning based verification.

Abstract: In a computerized method and system to generate visualization command data for two-dimensional visualization of a vascular tree of a vascular system from tomography data acquired via an imaging system, tomography data of the vascular tree are registered in the form of prepared tomography data that include acquired tomography data of the vascular tree in which said vascular tree is segmented at least in a region of surrounding structures, and in which are present a number of curve lines of vessels of the vascular tree that branch with one another. Inlets are shaped between curve lines that are adjacent and/or connected with one another via node points. Representation data are generated in which the inlets are arranged flat, adjoining one another in a two-dimensional presentation plane. The visualization command data are derived from the representation data and presented as an output.

Abstract: A method and system for up-vector detection for ribs in a 3D medical image volume, such as a computed tomography (CT) volume is disclosed. A rib centerline of at least one rib is extracted in a 3D medical image volume. An up-vector is automatically detected at each of a plurality of centerline points of the rib centerline of the at least one rib. The up-vector at each centerline point can be detected using a trained regression function. Alternatively, the up-vector at each centerline point can be detected by detecting an ellipse shape in a cross-sectional rib image generated at each centerline point.

Abstract: For therapy response assessment, texture features are input for machine learning a classifier and for using a machine learnt classifier. Rather than or in addition to using formula-based texture features, data driven texture features are derived from training images. Such data driven texture features are independent analysis features, such as features from independent subspace analysis. The texture features may be used to predict the outcome of therapy based on a few number of or even one scan of the patient.

Abstract: A method and system for extracting rib centerlines in a 3D volume, such as a 3D computed tomography (CT) volume, is disclosed. Rib centerline voxels are detected in the 3D volume using a learning based detector. Rib centerlines or the whole rib cage are then extracted by matching a template of rib centerlines for the whole rib cage to the 3D volume based on the detected rib centerline voxels. Each of the extracted rib centerlines are then individually refined using an active contour model.

Abstract: A method is disclosed for displaying a digital medical image dataset. In an embodiment of the method, when a new set of diagnostic findings is generated, an associated new information field is also generated. A number of possible positions are determined for the new information field. Each of the positions is assigned an evaluation metric in accordance with a number of predefined criteria. The new information field is in this case arranged at that position having the best evaluation metric. Each existing information field that overlaps with the new information field is shifted until the existing information field is arranged free of overlap with the new information field. Other existing information fields that overlap with previously shifted information fields are shifted until all of the information fields are arranged free of overlap with one another.

Abstract: An apparatus and method for training a landmark detector receives training data which includes a plurality of positive training bags, each including a plurality of positively annotated instances, and a plurality of negative training bags, each including at least one negatively annotated instance. Classification function is initialized by training a first weak classifier based on the positive training bags and the negative training bags. All training instances are evaluated using the classification function. For each of a plurality of remaining classifiers, a cost value gradient is calculated based on spatial context information of each instance in each positive bag evaluated by the classification function. A gradient value associated with each of the remaining weak classifiers is calculated based on the cost value gradients, and a weak classifier is selected which has a lowest associated gradient value and given a weighting parameter and added to the classification function.

Abstract: In a computerized method and system to generate visualization command data for two-dimensional visualization of a vascular tree of a vascular system from tomography data acquired via an imaging system, tomography data of the vascular tree are registered in the form of prepared tomography data that include acquired tomography data of the vascular tree in which said vascular tree is segmented at least in a region of surrounding structures, and in which are present a number of curve lines of vessels of the vascular tree that branch with one another. Inlets are shaped between curve lines that are adjacent and/or connected with one another via node points. Representation data are generated in which the inlets are arranged flat, adjoining one another in a two-dimensional presentation plane. The visualization command data are derived from the representation data and presented as an output.

Abstract: A method and system for extracting rib centerlines in a 3D volume, such as a 3D computed tomography (CT) volume, is disclosed. Rib centerline voxels are detected in the 3D volume using a learning based detector. Rib centerlines or the whole rib cage are then extracted by matching a template of rib centerlines for the whole rib cage to the 3D volume based on the detected rib centerline voxels. Each of the extracted rib centerlines are then individually refined using an active contour model.

Abstract: A method and system for up-vector detection for ribs in a 3D medical image volume, such as a computed tomography (CT) volume is disclosed. A rib centerline of at least one rib is extracted in a 3D medical image volume. An up-vector is automatically detected at each of a plurality of centerline points of the rib centerline of the at least one rib. The up-vector at each centerline point can be detected using a trained regression function. Alternatively, the up-vector at each centerline point can be detected by detecting an ellipse shape in a cross-sectional rib image generated at each centerline point.

Abstract: A method and system for automatically detecting liver lesions in medical image data, such as 3D CT images, is disclosed. A liver region is segmented in a 3D image. Liver lesion center candidates are detected in the segmented liver region. Lesion candidates are segmented corresponding to the liver lesion center candidates, and lesions are detected from the segmented lesion candidates using learning based verification.