摘要
利用jMRUI时域分析软件模拟先验知识基础集定量检测1H MRS大脑代谢物.通过模拟短TE=28ms的1H MRS的多种代谢物基础集,经HLSVD奇异值分解滤除残余水峰,将先验知识应用在QUEST算法上对1H MRS大脑代谢物信号进行定量估计.结果显示,基于QUEST能有效获得1H MRS大脑代谢物的估计浓度.表明引用先验知识基础集进行1H MRS定量分析可克服短TE时谱线复杂和波峰重叠峰等问题,极大地提取代谢物的信息.
The time domain analysis method provided by the jMRUI software was applied to quantify in vivo ^1H magnetic resonance spectra obtained from the brain with simulated priori knowledge base set. The residual water signals were filtered through Hankel-Lanczos singular value decomposition (HLSVD) algorithm. The QUEST arithmetic was used for metabolite quantification, with simulated short echo spectra (TE= 28 ms) of the metabolites as prior knowledge. The results showed that that the concentrations of brain metabolites could be estimated effectively from the ^1H spectra using the QUEST method. It was demonstrated that, by introducing priori knowledge base sets, the difficulties in metabolite quantification concerning the complexity of the spectra and peak overlapping could be overcame.
出处
《波谱学杂志》
CAS
CSCD
北大核心
2008年第3期315-321,共7页
Chinese Journal of Magnetic Resonance
基金
国家自然科学基金资助项目(30570480
30470900)
广东省自然科学基金(7008110)
汕头大学研究与发展基金资助(L00015)