It is widely accepted that MRI is one of the most versatile radiological techniques to study the human brain non-invasively. A reason for its versatility stems from its capability to create many different patterns of contrast in the brain, which depend on the data acquisition techniques employed. Each contrast mechanism is based on different physical and chemical properties of water molecules and, thus, each pattern often reflects different physiological and/or anatomical properties of the brain. In late '80s to early '90s, a new MRI contrast scheme called, "diffusion imaging” was introduced. This contrast is very unique because it is sensitive to structural orientation of axons, which can't be studied by any other radiological techniques. In this technique, we use water diffusion process as a probe to investigate brain axonal organizations based on a fact that water tends to diffuse along axons. From the beginning of the introduction, it was clear that the technique would provide exciting new information about brain white matter anatomy. However, it is far from simple to mathematically describe overwhelmingly complex neuroanatomy based on water diffusion process. Currently one of the most widely accepted models is based on a tensor model (thus called diffusion tensor imaging). In the first part of this lecture, I will go over the basics of the diffusion measurement and data processing based on the tensor model. Special emphasis will be placed on the understanding of data processing flow from raw MR images to calculated fiber orientation maps. In the second part, I will introduce several alternative approaches to describe the diffusion process and deduce the underlying anatomy. In the third part, I will discuss about strategies to apply computational neuroanatomy of white matter anatomy using the diffusion tensor imaging and its biological applications.