CASIA OpenIR  > 毕业生  > 博士学位论文
基于扩散张量成像的模式分类、纤维束分析方法及其应用
其他题名Pattern Classification, Fiber Tracts Analysis Methods Based on Diffusion Tensor Imaging and Their Applications
林富春
学位类型工学博士
导师蒋田仔
2006-05-28
学位授予单位中国科学院研究生院
学位授予地点中国科学院自动化研究所
学位专业模式识别与智能系统
关键词扩散张量成像 特征提取 模式分类 长度归一化 组图映射 Dti Feature Extraction Pattern Classification Length-normalized Parameterization Tractography Group Based Mapping
摘要扩散张量成像(DTI)是一种非侵入性的能提供活体内水分子扩散运动的成像技术,它能检测传统MRI所不能观察到的组织的微观变化,是MR成像技术的重大突破。本文以重大神经疾病的临床应用为目标,主要研究工作包括基于扩散张量成像的模式分类、长度规一化的白质纤维束定量分析方法、组图映射的白质纤维束定量分析方法及它们在多发性硬化研究中的应用。 基于脑影像信息的模式分类是目前脑影像研究中的热点课题。由于脑影像数据具有数据量大、维数高的特点,这对特征提取的要求很高。本文提出了一种基于扩散张量信息的二维直方图的特征提取方法,它有效的综合了DTI所提供的表观扩散系数(ADC)与分数各向异性(FA)的信息,并将此方法用于复发型视神经脊髓炎(RNMO)与复发好转型多发性硬化(RRMS)的模式分类。这两种疾病不管在临床表现上还是在脑影像上都有很多相似的特征,因此在临床上这两个疾病比较难以鉴别。我们的基于扩散张量信息的二维直方图的模式分类方法能较好的区分这两种疾病,其“留一法”正确分类率达到85.7%,这对临床诊断是具有重要意义的。而且,此方法不受病程的影响,因此,我们的方法可用于RNMO疾病的早期诊断。此外,我们还将该方法用于脑MRI阴性扫描的RNMO病人的模式判别,其“留一法”正确分类率达到87.5%,而且,还得到了RNMO疾病的判别区域。其结果表明,RNMO病人的表现正常的脑灰质存在隐匿性损伤。 由于个体间的白质纤维束的长度受其脑大小的影响,因此,其长度在不同个体之间是不一样的,这对组水平的纤维束定量分析带来了困难。为了克服这个问题,我们提出了一种长度规一化的白质纤维束定量分析方法,其优点在于建立了不同个体间纤维束的可比性,而且可以沿着纤维束做定量分析,为分析疾病的病理机制提供重要的参考。我们把该方法用于RNMO病人的锥体束研究,发现RNMO病人的锥体束的下方存在着异常,这种异常认为是继发于脊髓的病灶引起的。 对于器质性的疾病,其FA值比正常组织低,特别是在病灶区,其FA值更低,这导致了纤维跟踪在这些区域是不可靠的。为此,我们改进了一种组图映射的白质纤维束定量分析方法。我们把这种改进的方法用于RRMS病人的锥体束研究,发现RRMS病人的表现正常的锥体束有异常,这种异常认为是由Wallerian变性引起的。
其他摘要DTI is a non-invasive imaging technique capable of characterizing the diffusion properties of water molecules in vivo and detecting microstructural structural tissue changes not visible on conventional MRI. With aims at clinical applications in some serious neurologic diseases, this dissertation focuses on three topics of DTI, including pattern classification with two-dimensional histogram from DTI, fiber tracts quantitative analysis methods based on length-normalized parameterization and tractography based group mapping and their applications in MS. Feature extraction is very important for brain classification because of the characteristic of brain imaging data, such as mass data, and high dimension. In this paper, we proposed a novel method for feature extraction on the basis of two-dimensioanl histogram from DTI, which makes full use of information from ADC and FA of the brain. Then, we discriminated patients with RNMO from RRMS with the proposed method. The correct recognition rate of our method reached 85.7%, validated by the leave-one-out method. It is very helpful in differentiating RNMO from RRMS patients. Moreover, our proposed method is little influenced by the disease duration, which may be useful in early diagnosing RNMO from RRMS patients, particularly before progression and fulfillment of all clinical diagnostic criteria. On the other hand, we used the proposed method in discriminative analysis of RNMO patients without visible brain lesions. The correct recognition rate of our method reached 87.5%, validated by the leave-one-out method. More interesting, eigenimages, which reflect the discriminative pattern of RNMO patients from healthy subjects, were also generated. Diffuse damages in brain tissue of RNMO were detected from these eigenimages. We introduced a length-normalized parameterization method to establish anatomical correspondence of white matter fiber tracts across subjects. The major merit of our method is to provide comparability across subjects along the fiber tract on the basis of diffusion tensor tractography and to make it possible for the quantitative analysis along it. Then, we applied this method to investigate the presence of abnormal diffusion along the PYT of RNMO patients without visible brain lesions. The statistical results indicate that the diffusion indices were significantly different from those of healthy controls, especially in the lower part of the PYT. The differences may be caused by secondary degeneration to lesions in the spinal cord. For some diseases, such as MS, we proposed an improved method for quantitatively analyze white matter fiber tracts based on tractography group based mapping. Then, we used this method to investigate the presence of abnormal diffusion on the PYT of RRMS patients. Our findings confirm the presence of abnormal diffusion in the normal-appearing PYT of RRMS patients and suggest that Wallerian degeneration might be its mechanism.
馆藏号XWLW992
其他标识符200318014603016
语种中文
文献类型学位论文
条目标识符http://ir.ia.ac.cn/handle/173211/5914
专题毕业生_博士学位论文
推荐引用方式
GB/T 7714
林富春. 基于扩散张量成像的模式分类、纤维束分析方法及其应用[D]. 中国科学院自动化研究所. 中国科学院研究生院,2006.
条目包含的文件
文件名称/大小 文献类型 版本类型 开放类型 使用许可
CASIA_20031801460301(1750KB) 暂不开放CC BY-NC-SA请求全文
个性服务
推荐该条目
保存到收藏夹
查看访问统计
导出为Endnote文件
谷歌学术
谷歌学术中相似的文章
[林富春]的文章
百度学术
百度学术中相似的文章
[林富春]的文章
必应学术
必应学术中相似的文章
[林富春]的文章
相关权益政策
暂无数据
收藏/分享
所有评论 (0)
暂无评论
 

除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。