多体素核磁共振谱量化估计新方法及其辅助肿瘤识别方法的研究

项目名称： 多体素核磁共振谱量化估计新方法及其辅助肿瘤识别方法的研究

项目编号： No.81201148

项目类型： 青年科学基金项目

立项/批准年度： 2013

项目学科： 影像医学与生物医学工程

项目作者： 郭玙

作者单位： 天津大学

项目金额： 23万元

中文摘要： 多体素核磁共振谱（MRS）能够提供人体被测组织内不同代谢物的空间分布，从而可以从分子水平区分肿瘤组织和周围正常组织，反映肿瘤内部的非均一特性，因此在癌症治疗领域有重要的应用价值。为解决目前已有MRS技术存在的对代谢物定量分析准确度较差的问题，本课题研究、探索多体素MRS量化估计及其临床应用新方法。（1）在课题组已有成果基础上进一步研究基于联合稀疏表示的多体素MRS量化估计新方法，通过更充分地利用不同体素核磁共振谱之间的相关性，提高算法在低噪声及严重基线干扰情况下量化估计的准确度, 进而为临床应用提供更准确和更丰富的代谢物空间分布图。（2）发展出基于代谢物空间分布图和核磁共振解剖图像模糊信息融合的肿瘤自动识别新方法和技术，为临床肿瘤治疗中肿瘤区域的定义和勾画提供更有效工具。

中文关键词： 多体素核磁共振谱；稀疏表示；活体核磁共振谱量化估计；肿瘤识别；模糊融合

英文摘要： Multi-voxel magnetic resonance spectroscopy (MRS) is a method for assessing tissue function by obtaining information about the composition and spatial distribution of cellular metabolites. It is able to distinguishing regions of tumor from normal tissues and visualizing the metabolic heterogeneity in tumor regions. It is considered as one of the most promising technologies for cancer therapy. In this project, two novel methods have been proposed in developing a new multi-voxel MRS technology for cancer therapy. (1) A MRS quantitation method based on joint sparse representations will be studied for quantitating multi-voxel MRS data. By taking the advantage of the correlation between spectra of different voxels, this method would be more robust under low signal-to-noise ratio condition and with serious baseline interference as compared with conventional methods. This research is for the purpose of improving the accuracy of metabolite quantification with multi-voxel MRS and providing the spatial distributions of additional metabolites that are difficult to be quantitated by conventional methods. (2) A more efficient automated tumor identification technique based on the fuzzy fusion of metabolite spatial distribution and conventional magnetic resonance images will also be studies. This research will provide a robust

英文关键词： multi-voxel magnetic resonance spectroscopy；sparse representation；magnetic resonance spectroscopy quantification；cancer identification；fuzzy fusion