Explore chapters and articles related to this topic
Deformable Models and Image Segmentation
Published in Ayman El-Baz, Jasjit S. Suri, Level Set Method in Medical Imaging Segmentation, 2019
Ahmed ElTanboly, Ali Mahmoud, Ahmed Shalaby, Magdi El-Azab, Mohammed Ghazal, Robert Keynton, Ayman El-Baz, Jasjit S. Suri
As to brain segmentation, numerous researchers made use of deformable models. Ho et al., for example, segmented the brain tumors using level-set evolution with region competition [73]. A signed local statistical force that replaced the constant propagation term was proposed to avoid boundary leakage problems. The probabilities for background and tumor regions were computed from a pre- and post-contrast image and mixture-modeling of the histogram. Goldenberg et al. [74] used a shape-based level set approach that exploited the coupled surfaces model to segment the brain cortex. A neonatal image segmentation method was proposed [94], where local intensity information, atlas spatial prior, and cortical thickness constraint were combined in a single level-set framework. A longitudinally guided level-sets method for neonatal image segmentation was done by the same group [95], combining local intensity information, atlas spatial prior, cortical thickness constraint, and longitudinal information into a variational framework. A novel patch-driven level set method for the segmentation of neonatal brain MR images using sparse representation techniques was later proposed by the same group [96]. After building an atlas, the probability maps were integrated into a coupled level set framework. In [97–99], the brain was extracted using a hybrid approach that integrates geometrical and statistical models. The brain is parceled to a set of nested iso-surfaces using the fast marching level-set method in order to accurately discriminate between brain and non-brain tissues.
Comparison of 3D visualization results of segmented white and gray matter from T1 and T2 – weighted MRI data
Published in Waldemar Wójcik, Sergii Pavlov, Maksat Kalimoldayev, Information Technology in Medical Diagnostics II, 2019
Due to the isosurface function, surfaces created with dots of the same intensity value are created. You can have more control over the colors used to render the surface using the isocolors command. If the desired effect is only to show a certain part of the volume, you can extract them using the subvolume command. It is also used for flat sections (slices) of volume. The presentation of MR images as cross-sections of the brain in a three-dimensional coordinate system is done using the conturplot function. Using and space allows you to view the overall structure of the volume. In combination with the isocaps function, this technique can show information about the data in the interior of the created surface. The data can be smoothed with the smooth3 function. Then, using the appropriate path leads to displaying this data in a form that uses the original grayscale map. The isonormals function for surface rendering uses normal vertices derived from smoothed data. The resulting surfaces use one color to represent the volume.
Vascular Tree Segmentation from Different Image Modalities
Published in Ayman El-Baz, Jasjit S. Suri, Cardiovascular Imaging and Image Analysis, 2018
Ali Mahmoud, Ahmed Shalaby, Fatma Taher, Maryam El-Baz, Jasjit S. Suri, Ayman El-Baz
Yim et al. [21] proposed a methodology for deforming the isosurface to conform to the boundaries of objects in the image with minimal a priori assumptions of object shape. As in conventional methods, external forces attract the surface toward edges in the image. However, smoothing is produced by a moment that aligns the normals of adjacent surface triangles. Notably, the moment produces no translational motion of surface triangles.
Integration of cortical thickness data in a statistical shape model of the scapula
Published in Computer Methods in Biomechanics and Biomedical Engineering, 2020
Jonathan Pitocchi, Roel Wirix-Speetjens, G. Harry van Lenthe, María Ángeles Pérez
An algorithm was developed in Python 3.5 to automatically estimate sampled cortical thickness (Figure 1) starting from the initial three-dimensional model. For each point in the model surface, HU values were sampled along the line passing through that point and perpendicular to the surface, with a sampled distance of 0.1 mm. As a result, a HU profile was obtained for each point in the surface. First, a threshold of 226 HU was applied to detect the cortical bone in the profile and separate cortical and trabecular values. The full-width at half maximum method (FWHM) was used to estimate the cortical thickness by setting the 10th percentile of the trabecular intensities as the value for the trabecular bone. Using a Variable Wrapped Offset algorithm, it was possible to build the inner surface (representing the trabecular bone) by setting the measured cortical thickness as local offset for each point of the outer surface. The algorithm makes use of a vtkContourFilter to generate the Isosurface mesh from the scalar values (VTK: vtkContourFilter Class Reference).
Secular trends in cranial size and shape among black South Africans over the late 19th and 20th centuries
Published in Annals of Human Biology, 2020
Frederick E. Grine, Christine Lee, Carrie S. Mongle, Brendon K. Billings, Ian J. Wallace, Victor Mngomezulu
The stacked CT scan images of each cranium were visualised as an isosurface rendering in Aviso 9 LTE. The rendering was manipulated to produce a direct lateral view (perpendicular to the sagittal plane) and a direct superior view (perpendicular to the horizontal [Frankfurt] plane) of the cranium (Figure 1). Maximum cranial length between glabella and opisthocranion was measured from the lateral view, and maximum cranial breadth between the euyria was measured from the superior view of the cranium in Aviso 9 LTE (Figure 1). In the majority of specimens, the calotte had been removed from the base using a narrow blade electric bone saw during dissection. However, in no instance did this interfere with the determination of either maximum cranial length or breadth. The two cranial measurements were recorded together with the sex, ontogenetic age and year of birth for each individual.
The influence of the α/β ratio on treatment time iso-effect relationships in the central nervous system
Published in International Journal of Radiation Biology, 2020
Bleddyn Jones, Thomas Klinge, John W. Hopewell
In a very specific comparison of the impact of this variation in BED values, as a consequence of using different α/β ratios, the estimated equivalent acute or instantaneous dose, with no repair, is shown in Table 1. The treatment plans associated with the shortest and longest overall treatment time (20.8 and 108.6 min) were selected for this comparison. For the mean BED to the prescription iso-dose of 12 Gy the equivalent mean instantaneous dose only varied minimally, by 0.03 and 0.09 Gy for the short and long treatment times, respectively. However, since the BED for the prescription dose varies significantly, due to the detailed dose prescription to each voxel on the iso-surface, there was a greater variation in the equivalent instantaneous doses around the original prescription iso-surface of ∼ 0.6 Gy and ∼ 1.7 Gy for the short and long treatment time cases, respectively. This is supportive of the trend shown in Figure 6, suggesting an increasing trend in the range of BED values with increasing treatment time. This indicates that any variation resulting from a change in the α/β ratio used are very small compared with the normal variation in the dose scheduling to individual voxels on the prescription iso-surface.