Radiogenomics Framework for Non-Invasive Personalized Medicine

描述（由申请人提供）：肿瘤的分子谱现在用于确定预后和指导治疗。例如，EGFR 基因中存在突变很可能导致抗 EGFR 治疗。最近发现了一种图像表型，可以作为 EGFR 突变的生物标志物。这是放射基因组学这一新兴领域的可能性的先驱，放射基因组学的定义是将成像特征与潜在的分子特性直接联系起来。放射基因组学研究作为一个强大的新领域正在迅速获得认可，具有多种有前途的应用，例如非侵入性分子病变评估。当可以识别出反映相关分子畸变（例如 EGFR 基因突变）的图像替代物时，它们可以很容易地转化为临床护理。放射基因组学的附加价值可以很容易地转化，因为医学成像是肿瘤学常规管理的一部分。然而，这些早期应用并没有充分利用这些机会。首先，它们将相关性限制为少量手动注释的图像特征和一组预先选择的分子参数。其次，最初的应用仅限于单一组学，关注基因表达，没有考虑DNA突变、DNA拷贝数变化或DNA甲基化变化。我们将开发一个放射基因组学框架，以识别反映影响人脑肿瘤治疗和临床结果的相关分子肿瘤特性的非侵入性生物标志物。我们的目标不是通过计算手段模仿放射科医生的专业知识，而是为放射科医生和临床医生提供新的生物标志物。我们将提供创新的新算法来表示医学图像。一旦计算出这样的表示（例如，以大数据矩阵的形式），我们将识别预测临床结果的单变量和多变量图像特征。接下来，我们将使用复杂的方法与分子数据集成，以询问与临床相关结果相关的数据的不同观点。最终结果是一个放射基因组图，其中可以假设和验证分子特性和肿瘤异质性的图像特征。我们将拥有具有预后意义的图像特征和反映肿瘤的可操作分子特性（例如药物靶点活性或药物特征）的图像特征。

DESCRIPTION (provided by applicant): Molecular profiles of tumors are nowadays used to determine prognosis and to guide therapy. For example the presence of a mutation in the EGFR gene will most likely lead to anti-EGFR therapy. Recently an image phenotype was discovered that acts as a biomarker of EGFR mutation. This is a precursor of the possibilities of a new emerging field called radiogenomics defined as directly linking imaging features to underlying molecular properties. Research in radiogenomics is rapidly gaining recognition as a powerful new field that has several promising applications, such as non-invasive molecular lesion assessment. When image surrogates can be identified that mirror relevant molecular aberrations (e.g. a mutation in the EGFR gene) they can be readily translated in clinical care. The value added by radiogenomics can be readily translated, as medical imaging is part of routine management in oncology. However, these early applications have not taken full advantage of the opportunities. First, they limit the correlation to either a handful of manually annotated image features and a pre-selected set of molecular parameters. Secondly, the initial applications are limited to a single omics by focusing on gene expression, without taking into account DNA mutations, DNA copy number changes or DNA methylation changes. We will develop a radiogenomics framework to identify non-invasive biomarkers mirroring relevant molecular tumor properties that impact treatment and clinical outcome of human brain tumors. Our objective is not to mimic a radiologist's expertise through computational means, but to empower radiologists and clinicians with new biomarkers. We will offer innovative new algorithms to represent medical images. Once such a representation is computed (e.g., in the form of a large data matrix), we will identify univariate and multivariate image signatures predictive of clinical outcome. Next, we will use sophisticated methods for integration with molecular data to interrogate different views of the data with respect to a clinically relevant outcome. The end result is a radiogenomics map where image signatures of molecular properties and tumor heterogeneity can be hypothesized and validated. We will have image signatures that are prognostic and image signatures reflecting actionable molecular properties of a tumor such as drug target activity or drug signatures.