Research Of Algorithm In Body Composition Analysis Based On MRI

MRI(Magnetic Resonance Imaging)is recognized as one of the most advanced large-scale imaging equipment.A magnetic resonance imaging with a high-resolution has the imaging capabilities which can show in different angles and in different directions.It greatly enhances the physician’s diagnostic accuracy for lesions.The significance of body composition analysis is to detect chemical composition that affects people’s health as to find out whether there is missing or excess in human body.With the development of MRI technology,the application which combines body composition analysis and MRI technology has been paid more and more attention.The purpose of this research is to combine the Dixon water,fat separation method based on MRI or fat segmentation with body composition analysis,and apply it to the study of the brain,shoulder,knee and other parts.In addition a quantitative computation about body composition analysis is developed.The thesis takes a lot of papers as the theoretical basis.Principle and clinical applications of MRI imaging have been studied.Several methods of body composition analysis and technology on image segmentation with their research situation and applications have been expounded.Intensively research has been done on the Dixon water,fat separation method and fat segmentation.Test the feasibility of applying the Dixon water,fat separation method in the body composition analysis.Three-point Dixon water,fat separation method has been used to process the image of brain and compared with two-point Dixon water,fat separation method.Explore to apply fat separation method based on T1-weighted image to doing qualitative judgment in body composition analysis.In fat separation method,manual and auto methods have been used to process images about shoulder and knee.With two sets of contrast tests,the suitable application of the two methods in body composition analysis has been confirmed.A 3D fat distribution model of brain has been portrayed.Researched and developed a statistical approach of body composition analysis.Quantitative calculation methods of fat depends on the process strategy.For binary image,the amount of pixels is counted and the volume is calculated.For grayscale image,weighted statistics based on the gray value of image can be applied.The main results are as follows.With the experiments,Dixon water,fat separation method and fat segmentation method based on Ti-weighted image are effectively in water and fat separation in body composition analysis.And the three-point Dixon water,fat separation method has been testified more useful in diminishing the phase error than two-point Dixon water,fat separation method.The result of quantitative calculation in body composition analysis is as followed:the fat(or lipid)volume of brain takes 14.93%of the whole volume of the brain.The fat(or lipid)volume of shoulder takes 55.88%of the whole volume of shoulder image.The fat(or lipid)volume of knee takes 60.64%of the whole volume of knee image.It can be concluded that Dixon water,fat separation method and fat segmentation method based on T1-weighted image can fulfill the body composition analysis perfectly.The quantitative calculation method developed can calculate the amount the volume of fat(or lipid).This set of method can be applied in body composition analysis.