基于Savitzky-Golay滤波理论的超声造影像素级编码技术监测肝癌靶向治疗的数学模型建立

The molecularly targeted chemotherapy has become an important treatment strategy for patients with hepatocellular carcinoma (HCC) clinically. However, to date, there is no available noninvasive monitoring method which could assess treatment response early, accurately and be used repeatedly in clinical setting. Our previous research has reported that contrast-enhanced ultrasound (CEUS) with quantitative analysis provided functional information about the microcirculation of HCC and reflected their treatment response after anti-angiogenesis therapy. On this basis, we are supposed to develop a powerful software to rapidly process large volume of raw data from CEUS through the smoothing and reducing noise with automatic recognition of logic by using Savitzky-Golay digital differentiators and encoding technology in a pixel-level, quantitatively assessing tumor perfusion accurately and easily. The orthotopic nude mouse models with three metrocytes of human MHCC97H, HCCLM3 and HCCLM6 will be treated with sorafenib or pazopanib and monitored with CEUS perfusion imaging regularly. Multiple techniques including immunohistochemistry, western-blotting and polymerase chain reaction (PCR) will be used to determine the angiogenesis of tumors and change of vascular endothelial growth factor (VEGF)/VEGFR axises after targeted therapy will be compared. We will investigate the feasibility of CEUS perfusion imaging with quantitative analysis to seek and screen the valuable perfusion parameters in the evaluation of treatment response of HCC with targeted therapy. Finally, a diagnostic model based on CEUS perfusion imaging and Savitzky-Golay differentiators technology will be established and verified for the evaluation of treatment response of HCC after targeted therapy, laying the foundation for the consecutive acquisition of biological change of tumors, quantitative assessment of angiogenesis and treatment response of HCC after molecular-targeted chemotherapies with minimally invasive CEUS perfusion imaging.

肝癌的分子靶向治疗已成为许多患者的治疗策略，但临床上缺乏一种能早期、准确、可多次重复地评估肿瘤治疗效果的无创监测手段。我们的前期研究证实，作为功能影像学的超声造影定量分析可间接反映肝肿瘤微循环血流灌注、检测肝癌抗血管生成治疗后反应。本项目在此基础上，使用Savitzky-Golay滤波理论对超声造影原始数据进行平滑除噪和逻辑识别、像素级彩色编码，研制并完善一款快速处理大数据的造影参数灌注成像软件；通过以MHCC97H、HCCLM3和HCCLM6为母细胞的人肝癌裸鼠原位瘤模型构建的肝癌分子靶向治疗平台，与免疫组化、Western-blotting、PCR等技术观察的靶向治疗后VEGF/VEGFR轴变化相比较，创建超声造影像素级编码技术定量评估肝癌分子靶向治疗效果的数学模型，并临床验证，为临床通过无创或微创的影像学手段获得连续的肿瘤生物学变化、定量评价肿瘤血管生成和靶向治疗效果奠定基础。

随着新型肝癌靶向治疗药物的发明及其在临床的使用，临床上迫切需要一种能早期、准确且可多次重复使用的无创监测肿瘤治疗效果的影像学手段。超声造影是目前临床常用的可观察肝肿瘤微循环灌注的功能影像学技术，在临床诊断中以肉眼主观判断为主，不能敏感地发现肿瘤治疗后微循环的细微变化。本项目设计开发了一款使用Savitzky-Golay滤波理论的超声造影定量分析（Savitzky-Golay based contrast-enhanced-ultrasound quantification, SGCQ）软件，通过建立高转移性人肝癌细胞株裸鼠原位移植肝癌模型，对肿瘤微循环灌注进行量化评估，以及经Sorafenib和Lenvatinib抗肿瘤新生血管生成治疗的系列观察，结果显示SGCQ软件技术可直接分析完整的超声造影动态数据，自动识别和追踪图像的运动校正，可操作性和准确性高；在原位移植肝癌模型靶向治疗用药第4天即可发现肿瘤灌注参数（灌注时间、平均渡越时间、肿瘤相对于肝实质的灌注曲线下面积比）的显著变化，且与病理肿瘤微血管密度相关性好，具有早期无创量化评估肝细胞癌靶向治疗效果的作用。在临床病例研究中，SGCQ软件获得的肝细胞癌灌注参数中峰值减半斜率、曲线上升斜率、曲线下面积与转移性肝癌的均有统计学差异；简化的差值标化法的曲线下面积和平均渡越时间与肿瘤微血管密度CD31、VEGF相关性最高，创建了SGCQ监测肝癌靶向药物治疗效果评估的诊断模型。鉴于超声造影检查具有便捷、无肾损失和不增加晚期肝肿瘤患者痛苦的优势，该超声造影像素级编码技术监测肝癌靶向治疗的诊断模型有助于早期无创监测肝癌靶向药物治疗效果，在指导临床医师及时更换不敏感的靶向药物、减少患者不良反应发生率和降低医疗费用等方面具有重要的临床价值。