Macro-vasculature: A Novel Image Biomarker of Lung Cance

摘要 肺癌仍然是美国和世界范围内与癌症相关的死亡的主要原因。这个 与肺癌相关的高死亡率在一定程度上是由于未充分利用和有限的肺癌机会。 阻碍早期诊断的筛查。一些临床医生和政策制定者对肺的担忧 低剂量计算机断层扫描(LDCT)检查的癌症筛查是基于对提前时间偏差的担忧 假阳性率高。一种可减少筛查的肺癌生物标志物的开发 假阳性和对不确定结节的改进分类将缓解一些相关的担忧 肺癌筛查。尽管调查表明，使用LDCT扫描进行筛查可能会减少肺癌 死亡率比胸部X光低20%，据报道，~96%的可疑发现(大多是不确定的 结节)证实为非癌性(假阳性)。筛查不确定症的临床处理 结节通常导致不必要的、昂贵的和潜在有害的后续程序(例如，后续CT 扫描、正电子发射断层扫描(PET)/CT检查、侵入性活检)。我们开发了一种令人兴奋的 新的基于图像的大血管特征用于区分良恶性结节。我们建议 进一步开发这一功能，并通过一系列CT协议和其他机构的扫描进行验证。我们 还将把大血管特征与临床信息(例如，年龄、性别、吸烟史、 肺功能)，并评估该模型区分筛查检测到的良恶性的能力 不确定的结节。我们将调查放射科医生对不确定结节的分类是否随着 综合模型的输出与没有综合模型的分类相比较。这件事的成功 该项目可能导致一种新的、强大的肺癌生物标志物，以准确评估筛查检测到的 不确定的结节，可以显著减少肺内不必要的后续程序的数量 癌症筛查。

ABSTRACT Lung cancer remains the leading cause of cancer related deaths in the United States and worldwide. The high mortality associated with lung cancer is in part due to underutilization of and limited access to lung cancer screening that impedes early diagnosis. The apprehension of some clinicians and policymakers towards lung cancer screening with low-dose computed tomography (LDCT) exams is based on concerns of lead-time bias and high false-positive rate. The development of a robust lung cancer biomarker that reduces screen-detected false positives and improves classification of indeterminate nodules would relieve some of the concerns related to lung cancer screening. Although investigations show that screening with LDCT scans may reduce lung cancer mortality by 20% compared to chest x-ray, it is reported that ~96% of suspicious findings (mostly indeterminate nodules) turn out to be non-cancerous (false positives). Clinical management of screen-detected indeterminate nodules often leads to unnecessary, costly, and potentially harmful follow-up procedures (e.g., follow-up CT scan, positron emission tomography (PET)/CT exam, invasive biopsies). We have developed an exciting and novel image-based macro-vasculature feature to discriminate benign from malignant nodules. We propose to further develop the feature and validate it across a range of CT protocols and scans from other institutions. We will also integrate the macro-vasculature features with clinical information (e.g., age, gender, smoking history, lung function) and evaluate the model's ability to discriminate benign from malignant screen-detected indeterminate nodules. We will investigate if a radiologist's classification of indeterminate nodules improves with the output of the integrative model compared to classification without the integrative model. The success of this project may lead to a novel and robust lung cancer biomarker to accurately assess screen-detected indeterminate nodules that can significantly reduce the number of unnecessary follow-up procedures during lung cancer screening.