Three Dimensional Volumetric Modeling of the Internal Brain Structure Using Magnetic Resonance Imaging Slices.

摘要

The conventional cross sections of the brain, provided by magnetic resonance imaging (MRI) scanners, comprise a sparse dataset of 2-D gray-level images, that is neither capable of representing the 3-D nature of the brain, nor differentiating its various component parts in a convenient way. The target of the developed work is to fuse more information from the original MRI cross sections, which leads to building a 3-D computerized color-coded model of the normal human brain. The proposed model is beneficial in many areas like medical training, radiation treatment, 3-D model matching, or volume mensuration of brain component parts. This paper presents a revision of the different methods for building 3-D brain models, along with their advantages and disadvantages. A proposed method for building a 3-D brain model is then introduced. The method consists of three stages: interpolation of the original MRI slices, segmentation of the different brain tissues, and 3-D volumetric reconstruction. The resultant model can be geometrically transformed and arbitrarily dissected. The results are shown throughout the paper. Finally, the conclusions drawn from this work, as well as possible future extensions of the work, are listed.