Automated Measurement of Bowel Damage Using Enterography Imaging to Predict Clinical Outcomes in Crohn’s Disease.

目前克罗恩病（CD）的管理依赖于监测粘膜免疫反应的客观终点。 炎症虽然结构性肠损伤导致一半以上的CD患者接受手术， 结构性肠损伤的评估具有挑战性，难以量化并纳入治疗决策- 制作。横断面成像可以调查深部肠损伤和纤维狭窄的变化，但时间和 所需的专业知识，以及定性特征对观察者间差异的敏感性，对研究提出了挑战。 更广泛地使用成像数据来个性化护理。这项研究的长期目标是开发方法， 客观地测量结构性肠损伤，并对CD的临床结局进行个体化预测。的 本申请的总体目标是测试（i）计算图像分析方法收集 使用普通肠造影成像研究的肠损伤的传统和新特征，以及（ii） 评价这些指标改善CD预后预测的能力。核心假设是， 通过计算机图像分析方法收集的肠损伤特征将提高诊断的准确性。 预测CD治疗和临床结局的模型。这一中心假设将通过三个测试 具体目标：（1）确定肠造影研究的计算分析性能， 在常规护理过程中用于预测CD临床结局的肠损伤测量，（2） 比较肠造影图像分析预测治疗反应的性能， 现有的实验室和内窥镜措施，以及（3）评估图像分析能力，以确定 使用常规成像在CD中显示潜在组织组织学。在第一个目标中， 国家前瞻性CD自然历史数据集将进行图像分析，以提取用于 模型手术、住院和类固醇使用结果。这方面的进一步工作将检验该协议 放射科专家和计算机得出的肠道测量结果之间的差异。在第二个目标中， 新的生物疗法将接受定期的肠道造影，以比较 计算得出的肠道特征与炎症生物标志物和内窥镜检查的关系， 反应最后，在第三个目标中，接受选择性肠切除术治疗CD的患者将具有 术前肠造影高维图像特征将用于模拟组织学分级， 炎症和纤维化。拟议的研究在接近结构性肠损伤方面具有创新性， 相关的，但独立的和同样重要的，伴随评估炎症的预后， 治疗CD。此外，计算图像分析不仅在客观性方面， 再现性，以及如何测量CD负担的概念。这项研究意义重大，因为 它将证明结构性肠损伤特征对于最准确的 在临床护理和治疗试验中预测CD病程和治疗反应性。

Current management of Crohn’s disease (CD) relies on monitoring objective endpoints of mucosal inflammation. While structural bowel damage drives surgery in more than half of patients with CD, assessments of structural bowel damage are challenging to quantify and incorporate into treatment decision- making. Cross-sectional imaging can survey deep bowel damage and fibrostenotic changes, but the time and expertise needed, and the susceptibility of qualitative features to interobserver variation, pose challenges in the broader use of imaging data to personalize care. The long-term goal of this research is to develop methods to objectively measure structural bowel damage and individualize predictions of clinical outcomes in CD. The overall objectives in this application are to test (i) the ability of computational image analysis methods to collect traditional and novel characterizations of bowel damage using common enterography imaging studies and (ii) to evaluate these measures’ ability to improve predictions of CD outcomes. The central hypothesis is that bowel damage features collected by computational image analysis methods will improve the accuracy of models predicting therapeutic and clinical outcomes in CD. This central hypothesis will be tested through three specific aims: (1) Determine the performance of computational analysis of enterography studies capturing bowel damage measurements for predicting CD clinical outcomes in the regular course of care, (2) Prospectively compare the performance of enterography image analysis for predicting therapeutic response to existing laboratory and endoscopic measures, and (3) Evaluate image analysis capacity to determine underlying tissue histology in CD using conventional imaging. In the first aim, enterography studies in a national prospective CD natural history dataset will undergo image analysis to extract measurements used to model surgical, hospitalization, and steroid use outcomes. Further work in this aim will test the agreement between expert radiologists and computer-derived bowel measurements. In the second aim, subjects starting new biologic therapies will undergo scheduled enterography to compare the prognostic capabilities of computationally derived bowel features to inflammatory biomarkers and endoscopy for predicting therapeutic response. Finally, in the third aim, patients undergoing elective surgical resection of intestine for CD will have pre-operative enterography. High dimensional image features will be used to model histologic grading of inflammation and fibrosis. The proposed research is innovative in approaching structural bowel damage as a related, but independent and equally important, companion assessment to inflammation in the prognosis and treatment of CD. Further, computational image analysis opens new horizons not only in objectivity and reproducibility, but also concepts of how to measure CD burden. The proposed research is significant because it will demonstrate the indispensable importance of structural intestinal damage features for the most accurate predictions of CD course and therapeutic responsiveness in both clinical care and therapeutic trials.