Abstract:

Diagnosis based medical imaging is part of neuroscience profesionals´ training, and Magnetic Resonance (MR) is one of the most used techniques in neuroanatomy, given its high spatial resolution, which facilitates the identification and localization of brain structures involved in pathological processes. Visual inspection of individual 2D images is limited by the increasing number of current explorations and generated data. 3D visualization of the specific elements contained in conventional 2D images offers a more efficient alternative and minimizes the neccesity for developing subjective mental reconstructions, dependient on students´ spatial aptitud. Recent research has pointed out the involvement of deep brain structures such as corpus striatum, amygdala and hippocampus not only in, neurological diseases, but also in psyquiatric disorders. These structures are unapproachable by non invasive means and hard to identify even through MR imaging given its 2D feature, insufficient to reflect its whole complexitiy, and high homogeneity regarding its histological composition, which makes it difficult to discriminate which different structures are contained in the 2D image. For these reasons, automatic processing of 2D images has been applied in vascular, muscular and bone tone 3D reconstruction, but not in deep brain structures. Volumetric generation from original 2D MR images would offer a more realistic and complete vision, congruent with the innate human ability of 3D visualization, minimizing the likelihood to make perceptual errors, which constitutes one of the error sources in detecting 2D image abnormalities by visual inspection.

Aim: This study aims to achieve the volumetric generation of bilateral telencephalic and diencephalic structures, relevant from the clinical point of view, but hard to discriminate with conventional 2D images, offering a digital tool for optimizing available didactic resources in neuroanatomy and neuroimaging.

Method: Method required the aquisition of MR images in a healthy subject, delimitation and interactive or manual segmentation of Regions of Interest (ROIs) for the target structures: hippocampus, amygdala, corpus striatum, thalami and subthalamic nuclei; mash surface reconstruction by the Amira® software and final renderization.

Results: Bilateral telencephalic and diencephalic structures were volumetrically generated. They offered a more comprehensive knowledge than traditional neuroanatomy manuals actually avalaible, constituting an useful tool for its teaching.

Discussion: This study continues the line of work started previously, whose reconstruction of brain structures anatomical images was made through anatomical imaging from the Visible Human Project (VHP). Future research should develop algorithms for automatic segmentation of these deep structures and their register as functional images.

Conclusion: Volumetric visualization provided a complete representation of the morphological and spatial characteristics of deep brain structures, optimizing the analysis of information within conventional morphological 2D images by visual inspection, usually implemented in the training of neuroscience students.

Keywords:

Deep Brain Structures, 2D and 3D Visualization, Magnetic Resonance, Teaching of Neuroimage, Teaching of Neuroantomy, Interactive Segmentation.