Classification and Prognostication in Pulmonary Thromboembolism Using Computed Tomography Image Analytics

项目摘要 肺血栓栓塞症仍然是西方世界发病率和死亡率的重要原因。而 急性肺栓塞（PE）的许多初始症状在适当的治疗下消退， 提高对疾病慢性影响的认识，包括慢性血栓栓塞性疾病的发展， 肺动脉高压（CTEPH）持续呼吸困难和运动障碍。许多患者最初 被诊断患有PE的患者可能已经患有慢性疾病，并且在这些患者中对急性疾病进行了不适当的治疗。 病例可能有害，并延迟转诊到具有治疗慢性病经验的专业中心。对 另一方面，尽管目前治疗，许多急性PE患者继续发展为慢性疾病 确保解决问题的选择和后续行动仍然是一个挑战，特别是在没有能力预测谁将解决问题的情况下。 患上慢性病。此外，治疗的确定和选择可能是困难的，特别是 在亚大块急性PE和CTEPH中，特别是在新出现的治疗选择中。定量 需要一些方法来帮助在疾病出现的早期确定疾病轨迹，帮助指导诊断， 治疗和提高我们对这种情况的病理生理学的理解。 计算机断层扫描（CT）成像是评估肺血栓栓塞症的基石。在 对于急性PE，它通常是可用于评估治疗方案的第一种成像方式。随着患者 恢复后，它被用来检测慢性或复发性血栓，指导慢性疾病的干预措施。CT进展 成像质量，图像处理（包括深度学习的应用），加上越来越多的 计算能力使得从CT成像中提取大量新特征成为可能。在这 我们寻求联合收割机我们团队在CT图像量化方面的经验与多中心纵向 数据开发CT成像功能，可以识别和预测疾病的慢性化，其对 肺循环及其对治疗的反应。在目标1中，我们利用三个学术机构的纵向数据， 医院（布里格姆妇女医院、马萨诸塞州总医院、西北大学）， 评估临床表现的CT特征，以预测慢性疾病的存在或发展。在目标2中， 研究的表现和后续行动的形象，以建立定量模型的影响，急性和慢性 疾病对肺循环的影响，以帮助诊断和改善非侵入性方法 预测持续性疾病与患者临床状态的相关性。在aim 3中，我们使用一个组合 接受手术的CTEPH患者和肺动脉高压患者的纵向成像 高血压，以确定患者将有最佳的手术结果。我们认为 在这些互补目标中开发的特征和模型的组合将提高我们使用 临床可用的CT成像，以改善表型、诊断和治疗决策， 我们对肺血栓栓塞性疾病纵向进展的理解。

Project Summary Pulmonary thromboembolism remains a significant cause of morbidity and mortality in the western world. While many of the initial symptoms in acute pulmonary embolism (PE) resolves with appropriate treatment, there is increasing awareness of chronic impact of the disease ranging from development of chronic thromboembolic pulmonary hypertension (CTEPH) to persisting dyspnea and exercise impairment. Many patients initially diagnosed with PE may already have chronic disease and inappropriate treatment for acute disease in these cases may be harmful and delay referral to specialized centers with experience in treating chronic disease. On the other hand many patients with acute PE go on to develop chronic disease despite current treatment options and follow-up to insure resolution remains a challenge particularly without the ability to predict who will develop chronic disease. Furthermore, prognostication and selection of treatments can be difficult, particularly in submassive acute PE and CTEPH, particularly with newly emerging treatment choices. Quantitative methods are needed to help define disease trajectories early in presentation, help guide prognostication and treatment and improve our understanding of the pathophysiology of this condition. Computed Tomography (CT) imaging is the cornerstone of evaluation of pulmonary thromboembolism. In acute PE, it is the often the first imaging modality available for assessing treatment options. As the patient recovers, it is used to detect chronic or reoccurring clot guide interventions in chronic disease. Advances in CT imaging quality, image processing (including application of deep learning), coupled with increasing computation power make possible the extraction of a large number of novel features from CT imaging. In this proposal we seek to combine our team’s experience in CT image quantification with multi-center longitudinal data to develop CT imaging features that can identify and predict disease chronicity, its impact on the pulmonary circulation and its response to treatment. In Aim 1 we utilize longitudinal data from three academic hospitals (Brigham and Women’s Hospital, Massachusetts General Hospital, Northwestern University) to assess CT features at presentation that predict the presence or development of chronic disease. In Aim 2, we study both the presentation and follow-up image to build quantitative models of the impact of acute and chronic disease on the pulmonary circulation in order to help with prognostication and improve non-invasive methods of predicting the relevance of persistent disease to the clinical state of patients. In aim 3 we use a combination of longitudinal imaging in CTEPH patients having undergone surgery and patients with pulmonary arterial hypertension to identify patients that would have the most optimal surgical outcomes. We believe that the combination of the features and models developed in these complementary aims will advance our ability to use clinically available CT imaging to improve phenotyping, prognostication and treatment decisions and improve our understanding of the longitudinal progression of pulmonary thromboembolic disease.