构建预测直肠癌新辅助治疗后病理学完全缓解（pCR）的多模态、多参数诊断模型

项目名称： 构建预测直肠癌新辅助治疗后病理学完全缓解（pCR）的多模态、多参数诊断模型

项目编号： No.81501621

项目类型： 青年科学基金项目

立项/批准年度： 2016

项目学科： 医药、卫生

项目作者： 张晓燕

作者单位： 北京大学

项目金额： 18万元

中文摘要： 局部进展期直肠癌（LARC）经新辅助治疗（CRT）后，约有15-40%的患者可达到病理完全缓解（pathological complete response，pCR），此类患者预后明显较好，总生存率可达90%，局部复发率接近于0，肿瘤相关死亡率仅6%。越来越多的研究表明此类患者有可能避免根治性切除，代之以非手术、观察或局部切除术。尽管有多种影像学和分子生物学手段预测直肠癌对放化疗的反应，但如何在术前准确的筛查pCR患者，目前仍缺乏有效、可靠且实用的方法。本研究拟建立一个可准确筛选LARC经CRT后能否达到pCR的模型。此模型包括多个影像学指标及参数以提高结果可靠性，后续研究将对筛选出的pCR直肠癌患者进行大样本的前瞻性随机对照研究以最终明确pCR与预后的关系。以期更好的指导治疗、改变治疗策略，最大限度改善直肠癌患者的预后并提高生活质量。

中文关键词： 直肠癌；特征提取；医学图像处理

英文摘要： Following neoadjuvant CRT, 15-40% of tumours have a pathological complete response (pCR), with nomalignant cells demonstrated at histological assessment ofthe surgical resection specimen. A complete pathologicalresponse after CRT has been shown to correlatepositively with outcome, and it has been suggested thatpatients with LARC who have a pCR to CRT could bemanaged non-operatively.However, there is currently no established method topredict patients who will go on to have a pCR at resection.Studies have variablyincorporated clinical examination, ultrasound, computed tomography (CT), MR and molecular biological means to evaluate tumor therapy reponse,while a pCR is currently not accuratelypredicted, nor does thisfindings reliably correlate with pCR.There is still lack of effective, reliable and practicalmethod to predict pCR..The purpose of this study was to establish an accurate diagnostic model to predict whether LARC after CRT can reach pCR. This model should includemulti-level, diversified, multi modal parameters to enhance the reliability of the results, such asclinical examination (including endoscopic), laboratory, imaging indexes, and et al. Then a large prospective randomized studies will be performed to determine the relationship between pCR and the prognosis, in order to guide treatment, make clinical treatment strategy better, improve the prognosis andthe quality life of patients with rectal cancer.

英文关键词： rectal cancer;feature extraction;Medical image processing