Automatic Cerebellar MRI Labeling in Health and Disease

小脑功能障碍与多种疾病有关，这些疾病的发病率目前还很高。 治疗不充分。中风、肿瘤、感染、自身免疫性疾病、药物副作用和慢性 酒精中毒都会影响小脑，更具体的小脑疾病，如 橄榄桥小脑变性或小脑性共济失调。这些情况可能与一系列 致残的医疗条件，包括步态和平衡问题，说话含糊或缓慢，模糊或重音 视力、震颤和轻度痴呆。事实上，小脑功能障碍会导致大脑相当大的退化。 生活质量也可能会缩短寿命。越来越多的证据表明， 患者的小脑可被识别为特定的行为、运动和认知障碍。为了 研究这些关系，然后使临床方案可用，有必要开发和测试 从磁共振图像分析小脑解剖结构的自动程序。这个 拟议的研究计划满足了这一需求。具体地说，我们建议(1)优化成像和 小脑磁共振成像的组织分类方法；(2)发展和评价 小脑自动标记法；(3)小脑区形态差异的研究和建模； 以及(4)对正常人和小脑性共济失调患者的局部小脑体积进行初步研究。 病人。拟议的研究计划在几个方面都是新颖的。组织分类与分割 将使用多个数据集进行超分辨率，并将包含保持拓扑的可变形模型 和核密度先验形状模型。标注(地块)将包含拓扑保持， 使用多参数特征的多图谱可变形配准。因为我们的临床应用是 与小脑萎缩相关的疾病，我们的方法必须对正常和萎缩都有效 小脑。我们对脊髓小脑性共济失调(特别是SCA6)的初步研究提供了一个独特的机会 研究一种特定于小脑的疾病，了解区域萎缩及其与神经学的相关性 赤字。预计这里开发的自动化程序不仅将促进大规模的 神经科学研究小脑的结构/功能关系，但也将提供基础 作为一种临床工具，帮助治疗和管理所有类型的小脑疾病。

Dysfunction of the cerebellum is associated with a wide variety of diseases that have significant morbidity yet inadequate therapy. Strokes, tumors, infection, autoimmune disease, medication side effects, and chronic alcoholism can all affect the cerebellum, as can more specific diseases of the cerebellum such as olivopontine cerebellar degeneration or cerebellar ataxia. These conditions may be associated with a host of disabling medical conditions, including gait and balance problems, slurred or slow speech, blurred or double vision, tremor, and mild dementia. Indeed, cerebellar dysfunction can cause considerable degradation in the quality of life and may also be life-shortening. There is increasing evidence that regional changes in the cerebellum can be identified with specific behavioral, motor, and cognitive deficits in patients. In order to study these relationships and then to make clinical protocols available, it is necessary to develop, and test automatic procedures for analysis of the cerebellar anatomy from magnetic resonance images. The proposed research program addresses this need. Specifically, we propose to (1) optimize imaging and tissue classification methods for magnetic resonance imaging of the cerebellum; (2) develop and evaluate automatic cerebellar labeling methods; (3) investigate and model shape differences of cerebellar regions; and (4) conduct a pilot study of regional cerebellum volumes in normal subjects and cerebellar ataxia patients. The proposed research program is novel in several ways. Tissue classification and segmentation will use multiple data sets for superresolution and will incorporate topology-preserving deformable models and kernel density prior shape models. Labeling (parcellation) will incorporate topology-preserving, multiple-atlas deformable registration using multiparametric features. Since our clinical application is to diseases associated with cerebellar atrophy, our methods must work on both normal and atrophied cerebella. Our pilot study on spinocerebellar ataxia (specifically SCA6) provides a unique opportunity to study a cerebellar-specific disease and to learn about regional atrophy and its correlation with neurological deficits. It is expected that the automatic procedures developed herein will not only facilitate large-scale neuroscientific studies of the structure/function relationships in the cerebellum, but will also provide the basis for a clinical tool to assist in treatment and management of all types of cerebellar diseases.