基于人工智能的超声影像组学精确评估肝细胞癌侵袭范围的研究

Accurate assessment of invasion margin for hepatocellular carcinoma (HCC) preoperatively is extremely important during ablation of liver cancer. The assessment of 3-dimension invasion margin of HCC is an urgent clinical problem to be solved. Contrast-enhanced ultrasound and elastography are the up-to-date and important ultrasound techniques, which can reflect image characteristics of the elasticity and perfusion changes showing at the edge between HCC tumor and liver parenchyma; artificial intelligence based Radiomics can provide comprehensive quantifications of large amounts of image features extracted from medical images, and has the potential to uncover disease characteristics on images that fail to be appreciated by human eyes. There has been no study regarding whether or not the combination of the cutting-edge ultrasound techniques and the artificial intelligence based Radiomics can accurately assess the HCC invasion margin. We have already: 1. Established the technique to analyze dynamic clip and static images on ultrasonography using artificial intelligence based Radiomics; 2. Established an accurate scheme for matching ultrasound images and pathology slices, which is the key step to the two-dimensional assessment of tumor invasion margin; 3. Established a three-dimension imaging technique in ultrasound based on the multiple orthogonal planar image data. Thus, we propose a hypothesis that the combination of artificial intelligence, up-to-date ultrasonography, and three-dimensional imaging technique is able to assess the HCC invasion margin accurately. This project aims to establish a three-dimensional imaging technique for accurate assessment of HCC invasion margin preoperatively, using the combination of contrast-enhanced ultrasound, elastography, and artificial intelligence technology, in order to enable precise assessment of tumor invasion margin of HCC, to provide a technical basis and intervention strategies to reduce tumor recurrence after ablation.

术前精确判断肝癌侵袭范围有助制定消融范围，提高疗效。实现肝癌侵袭范围的立体定量评估是亟待解决的临床难题。弹性成像、超声造影是重要的超声技术，可反映肝癌侵袭边缘弹性和灌注异常等病变特征；人工智能影像组学可挖掘海量肉眼无法识别的图像特征，精准解读临床影像；联合超声新技术和影像组学能否判断肝癌侵袭范围，目前未见报道。申请人前期：1.建立人工智能分析动静态超声影像技术；2.建立超声图像及手术病理配准方案，完成侵袭范围二维评估关键步骤；3.具备基于多个正交平面的超声三维成像技术。基于前期研究和技术基础，我们提出设想：影像组学、超声新技术和三维成像可实现肝癌侵袭范围的立体评估。本项目拟联合弹性成像、超声造影和影像组学，将肝癌侵袭范围由人眼分辨转化为人工智能的精准计算，进而建立肝癌侵袭范围的三维成像技术，以实现肝癌侵袭范围的立体定量评估，为制定个体化消融边距，减少消融术后肿瘤复发提供技术基础及干预策略。

肿瘤消融是HCC的三大根治性治疗方式之一，肝细胞癌具有向周边组织浸润侵袭的生物学特征，术前精确判断肝癌侵袭范围有助制定消融范围，提高疗效。因此，如何术前精确判断肿瘤侵袭范围成为研究热点。肝细胞癌与正常肝组织交界区会出现弹性及血流灌注的变化，弹性成像及超声造影是重要的超声技术，在常规超声成像的基础上可反映肝癌侵袭边缘弹性和灌注异常等病变特征，更准确描绘肝癌实际大小和形状，影像组学可挖掘海量肉眼无法识别的图像特征，精准解读临床影像。申请人前期：1.建立分析动静态超声影像的组学技术；2.建立超声图像及手术病理配准方案，完成侵袭范围二维评估关键步骤；3.具备基于多个正交平面的超声三维成像技术。基于前期研究和技术基础，本项目拟联合弹性成像、超声造影和影像组学，将肝癌侵袭范围由人眼分辨转化为人工智能的精准计算，进而建立肝癌侵袭范围的三维成像技术，以实现肝癌侵袭范围的立体定量评估，为制定个体化消融边距，减少消融术后肿瘤复发提供技术基础及干预策略。.项目分析了肿瘤内部、肿瘤边界、瘤周肝组织超声特征对判断肝细胞癌侵袭的价值，以此制定病例影像数据的收集标准；分析不同模态超声成像技术对肝细胞癌侵袭的判断价值，综合不同成像技术及肿瘤不同部分影像数据建立判断肝细胞癌侵袭的超声影像组学模型；构建肝细胞癌微浸润的三维成像技术并申请专利；对肝细胞癌微浸润征象及预测模型在肝细胞癌预后预测中的作用进行分析和验证。项目最终通过影像这种无创、便捷的检查手段对早期肝细胞癌进一步分层，优化早期肝癌手术和消融的个性化治疗决策。