摘要:
An image segmentation method segments a plurality of image features in an image. The plurality of image features are segmented non-simultaneously in succession. The segmenting of each image feature includes adapting an initial mesh to boundaries of the image feature. The segmenting of each image feature further includes preventing the adapted mesh from overlapping any previously adapted mesh.
摘要:
A method of data processing is provided for estimating a position of an object in an image from a position of a reference object in a reference image. The method includes learning the position of the reference object in the reference image and its relation to a set of reference landmarks in the reference image, accessing the image, accessing the relation between the position of the reference object and the set of the reference landmarks, identifying a set of landmarks in the image corresponding to the set of the reference landmarks, and applying the relation to the set of landmarks in the image for estimating the position of the object in the image.
摘要:
A reconstruction processor (34) reconstructs acquired projection data (S) into an uncorrected reconstructed image (T). A classifying algorithm (66) classifies pixels of the uncorrected reconstructed image (T) at least into metal, bone, tissue, and air pixel classes. A clustering algorithm (60) iteratively assigns pixels to best fit classes. A pixel replacement algorithm (70) replaces metal class pixels of the uncorrected reconstructed image (T) with pixel values of the bone density class to generate a metal free image. A morphological algorithm (80) applies prior knowledge of the subject's anatomy to the metal free image to correct the shapes of the class regions to generate a model tomogram image. A forward projector (88) forward projects the model tomogram image to generate model projection data (Smodel). A corrupted rays identifying algorithm (100) identifies the rays in the original projection data (S) which lie through the regions containing metal objects. A corrupted rays replacement algorithm (102) replaces the corrupted regions with corresponding regions of the model projection data to generate corrected projection data (S′). The reconstruction processor (34) reconstructs the corrected projection data (S) into a corrected reconstructed 3D image (T′).
摘要:
A reconstruction processor (34) reconstructs acquired projection data (S) into an uncorrected reconstructed image (T). A classifying algorithm (66) classifies pixels of the uncorrected reconstructed image (T) at least into metal, bone, tissue, and air pixel classes. A clustering algorithm (60) iteratively assigns pixels to best fit classes. A pixel replacement algorithm (70) replaces metal class pixels of the uncorrected reconstructed image (T) with pixel values of the bone density class to generate a metal free image. A morphological algorithm (80) applies prior knowledge of the subject's anatomy to the metal free image to correct the shapes of the class regions to generate a model tomogram image. A forward projector (88) forward projects the model tomogram image to generate model projection data (Smodel). A corrupted rays identifying algorithm (100) identifies the rays in the original projection data (S) which lie through the regions containing metal objects. A corrupted rays replacement algorithm (102) replaces the corrupted regions with corresponding regions of the model projection data to generate corrected projection data (S). The reconstruction processor (34) reconstructs the corrected projection data (S) into a corrected reconstructed 3D image (T′).
摘要:
The invention relates to a method of segmenting a three-dimensional structure from a three-dimensional, and in particular medical, data set while making allowance for user corrections. The method is performed with the help of a deformable three-dimensional model whose surface is formed by a network of nodes and mashes that connect these nodes. Once the model has been positioned at the point in the three-dimensional data set at which the structure to be segmented is situated and positions of nodes have, if necessary, been changed by known methods of segmentation, any desired nodes can be displaced manually. The nodes of the model are recalculated by making weighted allowance for the nodes that have been displaced manually.
摘要:
A method for selecting vertices for performing deformable registration of imaged objects is provided. The selected vertices form corresponding pairs, each pair including a vertex from a first imaged object and a vertex from a second imaged object. The corresponding vertex pairs are sorted in order of distance between the vertices making up the corresponding vertex pair. The corresponding vertex pair with the greatest distance is given top priority. Corresponding vertex pairs that lie within a selected distance from the selected corresponding vertex pair are discarded. In this manner, the number of vertex pairs used for deformable registration of the imaged objects is reduced and therefore allows for processing times that are clinically acceptable.
摘要:
The invention relates to a method for data processing. At stage 3 the position of the reference object in the reference image and its relation to a set of reference landmarks in the reference image is established at step 6. In order to enable this, the reference imaging of learning examples may be performed at step 2 and each reference image may be analyzed at step 4, the results may be stored in a suitably arranged database. In order to process the image under consideration, the image is accessed at step 11, the suitable landmark corresponding to the reference landmark in the reference image is identified at step 13 and the spatial relationship established at step 6 is applied to the landmark thereby providing the initial position of the object in the actual image. In case when for the object an imaging volume is selected, the method 1 according to the invention follows to step 7, whereby the scanning 17 is performed within the boundaries given by the thus established scanning volume. In case when for the object a model representative of the target is selected, the method 1 follows to the image segmentation step 19, whereby a suitable segmentation is performed. In case when for the model a deformable model is selected, the segmentation is performed by deforming the model thereby providing spatial boundaries of the target area. The invention further relates to an apparatus and a computer program for image processing.
摘要:
An imaging system (10) includes imaging modalities such as a PET imaging system (12) and a CT scanner (14). The CT scanner (14) is used to produce a first image (62) which is used for primary contouring. The PET system (12) is used to provide a second image (56), which provides complementary information about the same or overlapping anatomical region. After first and second images (62, 56) are registered with one another the first and second images (62, 56) are concurrently segmented to outline a keyhole (76). The keyhole portion of the second image (56) is inserted into the keyhole (76) of the first image (62). The user can observe the composite image and deform a boundary (78) of the keyhole (76) by a mouse (52) to better focus on the region of interest within previously defined keyhole.
摘要:
The invention relates to a method of segmenting a three-dimensional structure from a three-dimensional, and in particular medical, data set while making allowance for user corrections. The method is performed with the help of a deformable three-dimensional model whose surface is formed by a network of nodes and mashes that connect these nodes. Once the model has been positioned at the point in the three-dimensional data set at which the structure to be segmented is situated and positions of nodes have, if necessary, been changed by known methods of segmentation, any desired nodes can be displaced manually. The nodes of the model are re-calculated by making weighted allowance for the nodes that have been displaced manually.
摘要:
The invention relates to a method for segmentation of a three-dimensional structure in a three-dimensional data set, especially a medical data set. The method uses a three-dimensional deformable model, wherein the surface of the model consists of a net of polygonal meshes. The meshes are split into groups, and a feature term is assigned to each group. After the model has been placed over the structure of interest, the deformable model is recalculated in consideration of the feature terms of each group.