Age-cognizant, fully-automated, high-dimensional modelling of brain tumour appearances on MR imaging

摘要：在美国儿科人群中，脑瘤仍然是癌症死亡的最常见原因，而胶质瘤是35岁以下成年人癌症死亡的第二大原因。胶质瘤患者的肿瘤特征具有丰富的多样性，即使在同一肿瘤级别内也是如此，这对疾病预后预测是一个巨大的挑战。年龄特定的遗传差异被认为是恶性胶质瘤临床观察到的各种差异的原因，最近世卫组织对脑瘤的重新分类首次纳入了除组织病理学之外的分子标记。这为识别复杂的“成像指纹”提供了机会，这种指纹不仅可用于非侵入性基因分型，还可用于建立特定年龄肿瘤生长速度的模型，预测肿瘤转化和脑侵袭。这项研究的目的是整合神经肿瘤学中的多模式数据，结合成像、临床、病理和遗传变量来开发年龄识别的、全自动的、高维的脑肿瘤模型，以达到个人水平的临床分诊、诊断、预后和治疗推断的目的。

Summary: Brain tumours remain the most common cause of cancer death in the US paediatric population and gliomas the second leading cause of cancer mortality in adults under 35. The rich diversity of tumour characteristics among glioma patients, even within the same tumour grade, is a big challenge for disease outcome prediction. Age-specific genetic differences are thought to account for the various clinical differences observed in malignant gliomas and the recent WHO reclassification of brain tumours incorporates, for the first time, molecular markers in addition to histopathology. This provides an opportunity to identify a complex "imaging fingerprint" that could not only be useful for non-invasive genotyping but also to model age-specific tumour growth rates, predict tumour transformation and brain invasion. The aim of this study is to integrate multimodal data in neuro-oncology, combining imaging, clinical, pathological and genetic variables to develop age-cognizant, fully-automated, high-dimensional modelling of brain tumours for the purpose of individual-level clinical triage, diagnosis, prognosis, and therapeutic inference.