Abstract
In relation to multislice X-ray CT (Computed Tomography) images of a mouse body, an automated method that evaluates the spatial distribution of two types of body fat (subcutaneous and visceral) as a quantitative trait is presented. A single X-ray CT image shows distribution of degree of X-ray attenuation (CT value) on a transverse section that is perpendicular to the body axis. By extracting regions where CT values are specific to body fat on the sections along the body axis, the spatial distribution of body fat can be recognized. However, since the two types of body fat have the same spectrum of CT values, they cannot be distinguished from each other only by their CT values. Therefore, we must first reconstruct the shape of the body cavities (thoracic and abdominal ones) that spatially divide those two types of body fat, and then classify the fat pats according to whether they are inside the cavities or not. The possibility of the numerical method that recognizes structure of a mouse body by using the landmarks on organs such as bones, muscles, lungs, and skin is discussed with showing the results of classification and quantitation of body fat.
