Bone is a common site of metastases for many primarily malignant tumors, being the third location after liver and lung [5]. Bone metastases are a major clinical concern as they can deteriorate the bone structure causing pathological fractures, vertebral/spinal instability, severe pain, spinal cord compression, functional impairment and increased mortality.
The vertebral column being the most frequent site of bone metastasis and one of the major organs in human body strongly correlated to neural and the skeletal system it plays an important role in various oncological, neurological and orthopaedic studies.
1.2.1. Vertebra Localization in CT Images
Not only studies or medical applications involving the investigation of vertebral body pathologies, …show more content…
Schmidt et al. [8] proposed a method to localize and label spine column from MR images using a multi-class classifier in combination with a probabilistic graphical model. Another similar work for detecting the lumbar vertebrae using multi-class classifier and graphical model was reported by Oktay et al. [9]. This work was further extended by Lootus et al. [10] by detecting vertebral regions in all 2D slices of a 3D MR volume using deformable part models. While Schmidt et al. [8] approach only considered the position, Kelm et al. [11] determined also the orientation and the scales by combining efficient local object detection based on Marginal Space Learning (MSL) with a global probabilistic model that incorporates pose priors on a nine-dimensional parameter …show more content…
[6, 12] proposed a local articulated model-based method, where the detection of vertebrae is done using a robust hierarchical algorithm. They aim to obtain robust performances, especially for cases with severe diseases and imaging artifacts, by first using a hierarchical strategy to learn anchor (distinctive) vertebrae, bundle (non-distinctive) vertebra, and intervertebral discs and by employing then a local articulated model to model the spatial relations across vertebrae and disc. This approach permits the detection in nearly arbitrary field of view scans, but it requires that at least one of the four ‘anchor’ vertebra (C2, T1, L1, S) be presented in order to graphically connect the other bundle vertebrae to
The vertebral column being the most frequent site of bone metastasis and one of the major organs in human body strongly correlated to neural and the skeletal system it plays an important role in various oncological, neurological and orthopaedic studies.
1.2.1. Vertebra Localization in CT Images
Not only studies or medical applications involving the investigation of vertebral body pathologies, …show more content…
Schmidt et al. [8] proposed a method to localize and label spine column from MR images using a multi-class classifier in combination with a probabilistic graphical model. Another similar work for detecting the lumbar vertebrae using multi-class classifier and graphical model was reported by Oktay et al. [9]. This work was further extended by Lootus et al. [10] by detecting vertebral regions in all 2D slices of a 3D MR volume using deformable part models. While Schmidt et al. [8] approach only considered the position, Kelm et al. [11] determined also the orientation and the scales by combining efficient local object detection based on Marginal Space Learning (MSL) with a global probabilistic model that incorporates pose priors on a nine-dimensional parameter …show more content…
[6, 12] proposed a local articulated model-based method, where the detection of vertebrae is done using a robust hierarchical algorithm. They aim to obtain robust performances, especially for cases with severe diseases and imaging artifacts, by first using a hierarchical strategy to learn anchor (distinctive) vertebrae, bundle (non-distinctive) vertebra, and intervertebral discs and by employing then a local articulated model to model the spatial relations across vertebrae and disc. This approach permits the detection in nearly arbitrary field of view scans, but it requires that at least one of the four ‘anchor’ vertebra (C2, T1, L1, S) be presented in order to graphically connect the other bundle vertebrae to