| Fat is an essential part of the body widely distributing in human body.In the Magnetic Resonance imaging,T1 of fat is shorter and T2 of fat is longer compared with muscle,so fat appears as high signal intensity in both T1 weighted and T2 weighted images,which tends to cover the lesions.What’s more,there is a chemical shift between water and fat.In the conventional sequences,such as spin echo and gradient echo sequence,Chemical shift artifact appears along frequency encoding direction.Artifact appears on the border of water and fat,so the image quality is seriously corroded.Recently,lots of Dixon-based water-fat separation methods were proposed one by one to improve image quality and provide new contrasts,including fat and water images.However,the success rate is still a challenge.The first part proposed a two-point water-fat separation new method based on double masks and 3D regional iterative phasor extraction(DM-3DRIPE)to ensure reliable separation.Researches proved that healthy condition is strongly associated with fat content in human body.Take fatty liver for example,when the cell number of hepatic steatosis is greater than 5%,it is identified as fatty liver.Greater cell numbers of hepatic steatosis,more fat content,severer fatty liver.Water-fat separation is extension of fat quantification,which is used to detect fat content.Nowadays the accuracy and repeatability of fat quantification has been proved by several researches,but all of these are based on commercial magnetic resonance(MR)of Philips、GE、Siemens and so on and these have not been reported on domestic MR.So the second part focused on investigating accuracy and repeatability of fat quantification on domestic 1.5T MR.Firstly,the paper proposed a two-point Dixon water-fat separation method based on DM-3DRIPE,which is an improvement of regional iterative phasor extraction(RIPE)algorithm.Determination of field phasor is the key of two-point water-fat separation.The field phasor of low signal-noise ratio(SNR)is affected by noise easily.To minimize the effect of noise,double masks were proposed to control the quality of image.Field phasor of high quality is firstly determined,and then was assisted to calculate field phasor of low quality.To reduce the effect of inhomogeneous field map,3D information was combined with RIPE to reliably calculate phasor.Results of simulation data showed water and fat maps were successfully achieved in the image containing low SNR region(SNR=0.8)or 4 times inhomogeneous field map than usual.Compared with traditional RIPE method,the DM-3DRIPE method decreased the error ratio from 28.33%to 0.73%for 413 in vivo images.DM-3DRIPE method was shown more robust when applied to low SNR region and strong inhomogeneous field map.The second part of the work was evaluating the accuracy and repeatability of multi-point Dixon-based fat quantitative technique on domestic 1.5 T superconducting MR.To evaluate the accuracy,different fat contents agarose phantom(agarose is similar with human tissue)made of oil and water was created.And then data pretreatment was carried out to eliminate the phase error caused by bipolar readout gradient.The fat fraction of phantom was calculated,and the influencing factors including T2*、peaks of fat、T1 and noise were eliminated or reduced.On the second day,phantom was rescanned to exam it’s repeatability.Fitting function between measured values and true values was:y= 1.02x-0.69,determination coefficient R2 was 0.9998,which showed that fat quantitative technique had high accuracy on domestic 1.5 T MR.Quantitative result of rescanned data was still accurate.In human trails,head、knee and pelvis data showed the effectiveness of quantitative technique. |
PDF Full Text Request