Automated analysis of lesions and atrophy in MS

项目描述（由申请人提供）：该项目的总体目标是开发一种能够定量分析多发性硬化症（MS）患者MRI数据的软件。该项目的成果将商业化为：用于临床试验的MS图像处理服务；我们的AutoAlign MR校准软件的ge专用扩展；最后作为提供MS特定功能（MS CAD，计算机辅助诊断）的医疗设备软件，供患者护理使用。该软件利用先进的配准、分割和量化技术，旨在解决该领域的长期问题。多发性硬化症是一种无法治愈的神经退行性疾病，大约每1000人中就有1人患病。它是中枢神经系统典型的炎症性自身免疫性疾病，可能是年轻人神经功能障碍的最常见原因。为了维持功能和减轻症状，患者必须接受终身治疗，通常包括昂贵的干扰素药物治疗。随着时间的推移，新症状的出现、缓解和恶化是MS临床病程的特征。由于这种可变性，当患者经历进展性MS发作时，临床医生寻求临床试验来指导治疗决策。国际多发性硬化症诊断小组建议在初次诊断时使用MRI，并在确定多发性硬化症阳性诊断后广泛用于监测疾病进展。在典型的临床环境中，MS患者的MRI评估仅限于放射科医生的视觉检查，并且没有对这些信息进行量化。此外，虽然多中心多发性硬化症临床试验通常使用连续MRI扫描，但由于扫描和扫描仪之间的基本获取水平可变性，这些信息会被MR注册问题所混淆。该项目中的技术解决了所有这些挑战，并将导致MS中MR评估的更大临床应用。

DESCRIPTION (provided by applicant): The overall goal of this project is to develop software which can quantitatively analyze MRI data from patients with multiple sclerosis (MS). The outcome of this project will be commercialized as: an MS image-processing service for clinical trials; a GE-specific extension of our AutoAlign MR alignment software; and finally as medical device software providing MS specific functionality (MS CAD, computer aided diagnosis) for patient care use. The proposed software, with its utilization of advanced registration, segmentation and quantification techniques, is intended to solve persistent problems in this field. MS is an incurable, neurodegenerative disease that affects approximately 1 in 1000 persons. It is the prototypical inflammatory autoimmune disorder of the central nervous system and may be the most common cause of neurological disability in young adults. To maintain functioning and alleviate symptoms, patients must undergo lifelong treatment, frequently including expensive interferon drug therapy. The appearance of new symptoms, remissions, and exacerbations over periods of years is characteristic of the clinical course of MS. Because of this variability, clinicians have sought paraclinical tests to guide treatment decisions when patients experience progressing MS attacks. MRI is recommended during the initial diagnosis by the International Panel on MS Diagnosis and is extensively utilized to monitor disease progression after a positive MS diagnosis has been determined. In the typical clinical setting, MRI evaluations of MS patients are limited to visual inspections by radiologists, and no quantification of this information is performed. Furthermore, while multicenter MS clinical trials typically utilize serial MRI scans, this information is confounded by MR registration problems because of fundamental, acquisition-level variability between scans and scanners. The technology in this project addresses all of these challenges, and should lead to greater clinical utility of MR evaluations in MS. Based on the 4 specific aims detailed below, during phase I of this project a software prototype which will be created to demonstrate the feasibility of this quantified MR approach to facilitate more accurate MS disease monitoring. Our first specific aim is to implement methods to provide a mask of the cerebral white matter, to aid in the detection, quantification and differentiation of MS lesions in various MR modalities. Our second specific aim is to develop algorithms to correct for patient motion between consecutive multimodal MRI scans, which is a prerequisite for multispectral MS lesion analysis. Our third specific aim is to develop prototype techniques for automatically identifying, quantifying and differentiating white matter abnormalities (i.e., putative MS lesions) on standard MR modalities. Our fourth specific aim is to validate this prototype against "gold standard" methods.