肝癌超声智能诊断和分子影像监控热化疗的研究

Ultrasound (US) is the first choice for screening and interventional treatment of liver cancer. However, there are some limitations for US imaging such as low signal-to-noise ratio, difficulty in distinguishing lesion features, and difficulty in capturing visual information in molecular level of tumor. In this study, digital information technology and molecular imaging technology will be used to carry out intelligent analysis and thermo-chemotherapy monitoring of liver cancer by US. The following studies are planned; 1) Establishing the deep model for automatic recognition of US features of liver cancer and its correlation with pathological information through the independently developed network architecture Light3. And the intelligent diagnosis of liver cancer by US will be realized; 2) Designing a deep convolutional neural network with rotational invariance and an adversarial network model to establish an intelligent early warning system for ablation effect of liver cancer. The intelligent decision by US for the treatment plan of liver cancer and the evaluation of prognosis will be realized; 3) Multi-functional molecular probe (complex material of Fe-MOF) will be synthesized by using molecular imaging technology and nanomedicine technology, with the functions of sensitization of thermal ablation field, targeted therapy of drug and multi-modality image of US/MRI/fluorescence tracing of tumor and drug target. The probe is expected to improve the inactivation ability of tumor in situ by thermal chemotherapy and reduce the recurrence and metastasis of liver cancer. Based on these, a new integrated system for the diagnosis and treatment of liver cancer covering US intelligent diagnosis, personalized treatment decision-making and accurate monitoring of thermos-chemotherapy will be established to improve the precise diagnosis and treatment level of liver cancer and 5-year survival rate.

超声是肝癌筛查和介入诊疗首选的影像手段。但超声影像存在信噪比低、病变特征不易识别、肿瘤分子水平的超声可视化信息捕捉困难等问题，对此本研究利用数字化信息技术和分子影像技术进行肝癌超声智能分析和热化疗监控。拟开展如下研究：1）通过自主研发的网络架构Light3建立肝癌超声特征自动识别及与病理信息相关性的深度模型，实现肝癌超声智能诊断；2）设计具有旋转不变性的深度卷积神经网络和生成对抗网络结构模型，建立精准的肝癌消融疗效智能预警系统，实现超声智能决策肝癌治疗方案和评价预后；3）利用分子影像技术和纳米医学技术合成多功能分子探针(铁基MOF复合材料)，具有消融热场增敏、药物靶向治疗以及超声/MRI/荧光多模态影像示踪药物靶点活性和肿瘤功能，提高热化疗原位灭活肝癌能力并降低转移复发率。由此建立覆盖超声智能诊断、个性化治疗决策和精准监控热化疗的肝癌诊疗一体化新体系，提高肝癌精准诊疗水平和5年生存率。

超声是肝癌筛查和介入诊疗首选的影像手段。但超声影像存在信噪比低、病变特征不易识别、肿瘤分子水平的超声可视化信息捕捉困难等问题，对此本研究利用数字化信息技术和分子影像技术进行肝癌超声智能分析和热化疗监控。开展如下研究：1）通过自主研发的网络架构Light3建立肝癌超声特征自动识别及与病理信息相关性的深度模型，实现肝癌超声智能检出及诊断，肝占位检出率高达99%以上，肝癌诊断AUC达0.932，不同年资医师的诊断水平在AI辅助下均有显著提升(p＜0.05)；2）设计具有旋转不变性的深度卷积神经网络和生成对抗网络结构模型，建立精准的肝癌消融疗效智能预警系统，实现超声智能决策肝癌治疗方案和评价预后；3）利用分子影像技术和纳米医学技术合成多功能分子探针(铁基MOF复合材料)，具有消融热场增敏、药物靶向治疗以及超声/MRI/荧光多模态影像示踪药物靶点活性和肿瘤功能，提高热化疗原位灭活肝癌能力并降低转移复发率。由此建立覆盖超声智能诊断、个性化治疗决策和精准监控热化疗的肝癌诊疗一体化新体系，提高肝癌精准诊疗水平。