基于多模态影像肝细胞肝癌β-catenin突变与预后预测的深度放射组学研究

肝癌发病率高，早期症状隐匿疾病进展迅速，发现时往往已是晚期。目前肝癌的治疗方式如手术切除、肝移植、经肝动脉化疗栓塞术和以索拉非尼为代表的靶向治疗均存在一定局限性，总体治疗效果仍不理想。Wnt/β-catenin通路突变是肝癌发生发展的重要分子基础，术前确定患者β-catenin突变状态对于肝癌生物学恶性行为的预测、治疗方案选择等有重要价值。目前术前穿刺活检存在出血，肿瘤种植转移的风险，且无法全面反映肝癌内部异质性。影像学检查如CT和MRI具有非侵入性、高分辨率、时空连续性等优势。深度学习和放射组学研究证实影像学可以表征复杂疾病表型及预测患者预后。本项目拟基于无创的术前多模态影像，采用深度放射组学方法实现术前预测肝癌β-catenin突变状态和预后情况，并在临床前瞻性研究中验证模型的效能。本项目将对肝癌的精准诊疗提供一种有效的工具，同时也为其它实质性肿瘤的的诊断和评估带来革命性改变。

The incidence and mortality of Hepatocellular carcinoma (HCC) has been increasing recently. It is often discovered in the late stage because of early hidden symptoms and rapid progression. At present, the treatment of HCC such as surgical resection, liver transplantation, transcatheter arterial chemoembolization (TACE) and targeted therapy represented by sorafenib have certain limitations so that the overall treatment effect is still not ideal. The mutation of Wnt / β-catenin pathway is an important molecular basis for the occurrence and development of HCC. Thus, detecting β-catenin mutation status before treatment is important to predict HCC potential biological behavior and choose effective treatment strategy. But now it could only be acquired via biopsy, which is invasive, difficult to perform longitudinally, and limited to a single region of a tumor. Imaging examination such as CT and MRI are non-invasive and with high resolution. Deep learning and radiomics studies have confirmed that imaging can characterize tumor phenotypes, track spatial and temporal heterogeneity and predict prognosis of patients. Our project aims to build a deep learning-based radiomics model to predict β-catenin mutation before treatment and prognosis of HCC based on non-invasive preoperative multi-modality imaging, and to verify the effectiveness of the model in prospectively. If succeed, it will provide an effective tool for accurate diagnosis and treatment of HCC, and thereby bring revolutionary influences on the precision treatment for other solid tumors.