Automatic Volumetric Segmentation of Encephalon by Combination of Axial, Coronal, and Sagittal Planes
Rodrigo Siega, Edson Jose Rodrigues Justino, Jacques Facon, Flávio Bortolozzi, Luiz Roberto Aguiar · Research in Computing Science · 2018
This paper describes a method of automatic volumetric segmentation of the human encephalon by Magnetic Resonance Imaging (M RI) using three anatomical planes of visualization (axial, coronal and sagittal).For mapping the volumetric topography of the encephalon we developed a set of algorithms for managing the different planes.It is intended for the segmentation of magnetic resonance images with T 1 weighting, Inversion Recovery (IR) and Gradient Echoes GRE (T 1 IR GRE).By combining filtering techniques and techniques of adaptive multiscale representation, directional transformation, and morphological filters, the method generates separated masks of the encephalon in the axial, coronal, and sagittal planes.Based on the masks of the three planes, reconstruction and rendering of the encephalon surface, which reveal the cortical mantle, are carried out.Tests performed using a database containing DICOM images of 30 volunteers show that the proposed method of automatic volumetric segmentation is promising for the study described in this paper.