Publications
Text queries can be conducted by Author, Title, or Keyword.
Extracting and Representing the Cortical Sulci
Source: IEEE Computer Graphics and Applications 1999 May;19(3):49-55.
Author: Zhou Y, Thompson PM, Toga AW.
Abstract:
Cortical sulci are deep valley-like regions between two juxtaposed cortical folds. Accurate shape representation is required for the quantitative morphological analysis of brain structure. Current sulcal modeling techniques, more or less, depend on manual interpretation. In this paper we propose a robust voxel-coding methodology for automatic extraction and parameterization of the sulcal surfaces from volume data sets. The cortical sulci are interpreted as a collection of centered 2D contours (within the slices), which are generated in two steps: initial contour extraction and final contour generation. A voxel-coding technique based on the exterior boundary of the cerebral cortex is used for the extraction of the initial contours, providing the fundamentals for unambiguous 3D reconstruction of the sulcal surface. Another similar voxel-coding technique based on the interior boundary of sulcal regions and a 2D deformable curve model are employed for the generation of the final contours, providing a parametric representation of them. On the basis of the characteristics of generated contours, the integration and parametric representations of the sulcal surface can be obtained easily by testing adjacency relationships of contour endpoints within adjacent slices without any ambiguity in contour connections. Generated results on MRI and post mortem data sets are shown and comparisons with manually-derived surface models are given, demonstrating the potential of our methodology. The resulting contours can be applied to the development of maps, atlases and the creation of models for research and education. BACKGROUND: A number of methods model cortical features. Most focus on the outer cortical surface, composed of a pattern of fissures and folds named sulci and gyri, respectively. The complexity and variability of sulci have made them difficult to model. The 3D skeleton extraction method offers a powerful voxel-based technique for medial surface generation, successfully applied in computer vision for image compression and reconstruction, and in visualization for endoscopic navigation. OBJECTIVE: Our goal was to create a simple, efficient method to extract sulcal surfaces and output them as parametric representations without manual interaction. In this article we introduce a pixel-coding method, called voxel-coding, that employs both skeleton extraction and deformable model mechanisms. CONTOUR GENERATION: Contour generation employs a series of 2D slices. The algorithm has 3 steps: data processing, extraction of initial contours, and final contour generation. CONCLUSIONS: Our efficient voxel-coding methodology for sulcal surface extraction doesn't require manual interaction in generating parametric representations. Sulcal characteristics motivated the methodology-the structures are convoluted ribbons starting at the external limit of the cerebral cortex and extending deeply into the brain. Voxel coding successfully stimulates these features by combining a variety of techniques. 2D generation of the contours avoids a complicated 3D search and computation.
Source: IEEE Computer Graphics and Applications 1999 May;19(3):49-55.
Author: Zhou Y, Thompson PM, Toga AW.
Abstract:
Cortical sulci are deep valley-like regions between two juxtaposed cortical folds. Accurate shape representation is required for the quantitative morphological analysis of brain structure. Current sulcal modeling techniques, more or less, depend on manual interpretation. In this paper we propose a robust voxel-coding methodology for automatic extraction and parameterization of the sulcal surfaces from volume data sets. The cortical sulci are interpreted as a collection of centered 2D contours (within the slices), which are generated in two steps: initial contour extraction and final contour generation. A voxel-coding technique based on the exterior boundary of the cerebral cortex is used for the extraction of the initial contours, providing the fundamentals for unambiguous 3D reconstruction of the sulcal surface. Another similar voxel-coding technique based on the interior boundary of sulcal regions and a 2D deformable curve model are employed for the generation of the final contours, providing a parametric representation of them. On the basis of the characteristics of generated contours, the integration and parametric representations of the sulcal surface can be obtained easily by testing adjacency relationships of contour endpoints within adjacent slices without any ambiguity in contour connections. Generated results on MRI and post mortem data sets are shown and comparisons with manually-derived surface models are given, demonstrating the potential of our methodology. The resulting contours can be applied to the development of maps, atlases and the creation of models for research and education. BACKGROUND: A number of methods model cortical features. Most focus on the outer cortical surface, composed of a pattern of fissures and folds named sulci and gyri, respectively. The complexity and variability of sulci have made them difficult to model. The 3D skeleton extraction method offers a powerful voxel-based technique for medial surface generation, successfully applied in computer vision for image compression and reconstruction, and in visualization for endoscopic navigation. OBJECTIVE: Our goal was to create a simple, efficient method to extract sulcal surfaces and output them as parametric representations without manual interaction. In this article we introduce a pixel-coding method, called voxel-coding, that employs both skeleton extraction and deformable model mechanisms. CONTOUR GENERATION: Contour generation employs a series of 2D slices. The algorithm has 3 steps: data processing, extraction of initial contours, and final contour generation. CONCLUSIONS: Our efficient voxel-coding methodology for sulcal surface extraction doesn't require manual interaction in generating parametric representations. Sulcal characteristics motivated the methodology-the structures are convoluted ribbons starting at the external limit of the cerebral cortex and extending deeply into the brain. Voxel coding successfully stimulates these features by combining a variety of techniques. 2D generation of the contours avoids a complicated 3D search and computation.