Tissue segmentation in MRI

多发性硬化症(MS)是年轻人最常见的神经系统疾病，在加拿大约有5万人受到影响。这种疾病的特征是对中枢神经系统的短暂攻击，导致肢体麻木或无力，协调、眩晕或视觉功能障碍。目前还没有治愈的方法。在过去的十年里，一些新的治疗方法已经被批准用于临床，主要是基于它们有能力减少新的皮损数量。这些药物可以减少30%-50%的发作次数。对更新、更有效的治疗方法的评估需要进行临床试验，在临床试验中，通过识别磁共振成像(MRI)中看到的新病变来衡量“疾病负担”。在制药行业中，病变识别的最新技术包括主观人工或半人工程序。这一过程受制于观察者之间和观察者内部的可变性，这降低了检测药物效应的统计能力。为了解决这个问题，必须增加研究对象的数量，这是一个非常昂贵的选择。

Multiple sclerosis (MS) is the most common neurological disease of young adults, affecting approximately 50,000 people in Canada. The disease is characterized by transient attacks on the central nervous system and results in numbness or weakness of a limb, in coordination, vertigo or visual dysfunction. At present there is no cure. Over the last decade, a number of new therapies have been approved for clinical use based primarily on demonstration of their ability to reduce the number of new lesions. These drugs reduce the number of attacks by 30-50%. Evaluation of newer, more efficient treatments requires clinical trails where the "burden of disease" is measured by identification of new lesions seen in magnetic resonance imaging (MRI). The state of the art for lesion identification in the pharmaceutical industry consists of subjective manual or semi-manual procedures. This process is subject to inter- and intra-observer variability that diminishes the statistical power to detect a drug effect. To address this problem, one must increase the number of subjects in the study a very expensive option.