Multi-channel Enhancement and Validation of Subcortical Morphometry in HIV

描述(申请人提供)：尽管在高效抗逆转录病毒治疗的时代取得了重大进展，但人类免疫缺陷病毒(HIV)对中枢神经系统(CNS)的影响仍然普遍存在。在这种情况下，我们继续寻找敏感的、非侵入性的生物标志物来指导治疗，监测治疗干预，并识别有中枢神经系统衰退风险的个体。以前的工作已经支持了神经成像在识别艾滋病毒生物标记物方面的价值，特别是在白质和尾状核中。最近在多地点CNS HIV抗逆转录病毒治疗效果研究(CHECH)计划内的临床神经成像工作采用了多种方式，包括结构和扩散张量磁共振成像来评估大脑对艾滋病毒感染的反应。Charge使用了多通道MRI(T1、T2、PD)来进行组织分割，并手动划定全球区域，从而成功识别了与HIV相关的神经成像，如白质异常(AbWM)。然而，手工描绘单个结构(例如尾状核)在大规模研究中是不可行的。最近的方法学进步为自动化区域划分提供了希望，然而，这些方法尚未在艾滋病毒中得到验证。这些方法仅依赖于T1序列可能会导致将AbWM错误分类为灰质，从而导致结构体积估计的灵敏度降低和错误。由于AbWM与疾病严重程度的相关性，这种分类错误可能导致系统性偏差。利用来自CHEAT的数据，我们建议严格评估两种相对自动化的皮质下标记方法在HIV中的性能，并检查可能提高这些方法性能的统计前处理程序。特别是，我们将检查与HIV相关的AbWM相对于形态计量“黄金标准”对方法性能的影响程度。黄金标准感兴趣区域(例如，尾状区域)将根据现有的包裹体形态测量进行定义，从而提高了手动圈定的可靠性。我们还将利用来自PD、T2和扩散张量成像(DTI)体积的信息，通过使用交替条件预期(ACE)来研究统计增强输入T1体积的价值，ACE将使用补充体积来增强T1体积中的组织对比度。进一步的验证工作将检查结果结构体积与神经行为和医学生物标记物之间的关系。对结构神经成像方法学的系统评估和增强将使我们能够更好地表征HIV的潜在神经发病机制，潜在地验证这些工具在HIV中的使用，为最终用户提供指导，并为开发人员提供对未来工作的洞察。重要的是，这项工作将为《宪章》提供地区性形态测量数据，作为一项资源倡议，用于探索额外的、更具体的假设检验，例如尾状核体积作为艾滋病毒相关认知功能下降的生物标志物的价值、下降的风险或治疗干预的有效性。 公共卫生相关性：尽管在高效抗逆转录病毒治疗时代取得了重大进展，但人类免疫缺陷病毒(HIV)的中枢神经系统(CNS)效应仍然普遍存在，神经损伤和神经认知功能障碍对个人日常生活、工作表现和护理需求的影响可能是巨大的。正在进行的多站点CNS HIV抗逆转录病毒治疗效果研究(CHECURE)计划为描述此类效果提供了一个独特的多学科环境，在其中，我们建议对以前未在HIV中验证的结构性神经成像方法进行系统评估和增强。这项工作将为《宪章》提供地区性形态测量数据，作为一项资源倡议，用于探索更多、更具体的艾滋病毒神经成像相关因素，以指导治疗、监测治疗干预并识别有中枢神经系统衰退风险的个人。

DESCRIPTION (provided by applicant): Despite significant advances in the era of highly active anti-retroviral therapy, central nervous system (CNS) effects of human immunodeficiency virus (HIV) have remained prevalent. In this light, we continue to search for sensitive, non-invasive biomarkers to guide treatment, monitor therapeutic intervention, and identify individuals at risk for CNS decline. Previous work has supported the value of neuroimaging for identifying biomarkers in HIV, particularly in the white matter and caudate nucleus. Recent clinical neuroimaging efforts within the multi-site CNS HIV Anti-Retroviral Therapy Effects Research (CHARTER) program have employed multiple modalities including structural and diffusion tensor MR imaging to assess the brain's response to HIV infection. CHARTER has employed multi-channel MRI (T1, T2, PD) to produce tissue segmentations with manual delineation of global regions, resulting in successful identification of neuroimaging correlates of HIV, such as abnormalities in the white matter (AbWM). Manual delineation of individual structures (e.g., caudate), however, is prohibitive on large-scale studies. Recent methodological advancements offer promise for automating regional delineation, however, these methods have not yet been validated in HIV. The reliance of these methods solely on T1 sequences can result in the misclassification of AbWM as gray matter, leading to decreased sensitivity and errors in structural volume estimates. Such classification errors may result in systematic bias, due to the correlation of AbWM with disease severity. Utilizing data from CHARTER, we propose to critically assess the performance of two relatively automated subcortical labeling methods in HIV and to examine a statistical pre-processing procedure that may enhance performance of these methods. In particular, we will examine the degree to which HIV-related AbWM affects method performance relative to a morphometric "gold standard." The gold standard regions of interest (e.g., caudate) will be defined upon existing CHARTER morphometry, thus increasing the reliability of the manual delineations. We also will investigate the value of statistically enhancing the input T1 volumes, with information from PD, T2, and diffusion tensor imaging (DTI) volumes, through the use of alternating conditional expectations (ACE), which will use the supplemental volumes to enhance tissue contrast in the T1 volumes. Further validation efforts will examine the relationship between resultant structural volumes and neurobehavioral and medical biomarkers. The systematic assessment and enhancement of structural neuroimaging methodology will allow us to better characterize the underlying neuropathogenesis of HIV, to potentially validate these tools for use in HIV, to provide guidance to end-users, and to offer insight to developers for future work. Importantly, this work will provide regional morphometric data to the CHARTER as a Resource initiative for the exploration of additional, more specific hypothesis testing, such as the value of caudate volume as a biomarker of HIV-related cognitive decline, risk for decline, or effectiveness of therapeutic intervention. PUBLIC HEALTH RELEVANCE: Despite significant advances in the era of highly active anti-retroviral therapy, central nervous system (CNS) effects of human immunodeficiency virus (HIV) have remained prevalent, and the impact of neurological impairment and neurocognitive dysfunction on individuals daily lives, work performance, and care needs can be significant. The on-going multi-site CNS HIV Anti-Retroviral Therapy Effects Research (CHARTER) program provides a unique multidisciplinary setting for the characterization of such effects, within which we propose a systematic assessment and enhancement of structural neuroimaging methods, not previously validated in HIV. This work will provide regional morphometric data to the CHARTER as a Resource initiative for the exploration of additional, more specific neuroimaging correlates of HIV to guide treatment, monitor therapeutic intervention, and identify individuals at risk for CNS decline.