基于有限振幅超声回波信号的非线性参数B/A诊断巴马香猪肝纤维化的实验研究

Early detection and staging of liver fibrosis is important for diagnosis, treatment decision and evaluation of chronic liver disease. But it is difficult through clinical data, laboratory tests and routine image examinations. Ultrasound shear wave elastography is useful but it can be easily affected by respiration, heartbeat, patient condition and equipment diversity. Biopsy is accurate but invasive. Studies showed ultrasound nonlinear parameter B/A is potentially valuable in evaluating porcine liver fibrosis in vitro. However, to our knowledge, there is no B/A measurement method on current Color Doppler ultrasound equipment(CDUE)so far. Therefore, based on previous work, this study firstly aims to develop a new method of measuring B/A using CDUE for Bama miniature porcine liver in vivo. This new method is to investigate the nonlinear partial differential equation of finite amplitude based on media density using echo mode configuration, then to acquire Bama liver B/A in vivo by derivating the relationship of probe input voltage vs pulse echo signal and the correspondence of fundamental wave vs harmonic wave. Secondly, analyze the influence of probe frequency and measuring depth to normal Bama liver B/A measurement. Thirdly, explore the value of B/A in diagnosing and staging porcine liver fibrosis. In detail, establish Bama porcine liver fibrosis of different degree models, measure liver B/A in vivo and conduct liver biopsy for all pigs. Then obtain excised liver tissue blocks by partial hepatectomy and measure the blocks’ B/A in vitro for part of the pigs. Finally, analyze the difference of liver B/A measured in vivo vs in vitro and investigate the accuracy of B/A for diagnosing and staging porcine liver fibrosis taking pathology as gold standard. All the above work is to lay foundation and accumulate experience for realizing vivo B/A detection on CDUE for clinical application.

肝纤维化早期发现和分级对慢性肝病诊断、治疗选择和评估都很重要，临床、实验室及常规影像均不可靠；剪切超声弹性有一定准确性，但受呼吸、心跳、患者条件及不同设备等影响；病理诊断准确但有创。研究显示非线性超声参数B/A 能区分离体猪肝纤维化程度，但至今尚未见到应用现有彩超对活体组织检测B/A。本课题拟在前期工作基础上，用现有彩超的脉冲回波处理模型研究有限振幅超声波基于介质密度的非线性偏微分方程，推导脉冲回波信号与探头输入电压、回波信号中基波与谐波的对应关系来获取B/A值，并率先用彩超检测正常巴马香猪肝脏B/A；然后探讨探头频率及检测深度对活体猪肝B/A的影响；最后建立巴马香猪不同程度肝纤维化模型并检测其肝脏活体B/A值且进行肝穿，部分猪以手术获取肝组织块并测量其离体B/A值，分析猪肝活体与离体B/A差异，以肝穿病理为金标准探索活体B/A对猪肝纤维化诊断和分级的准确性。为B/A检测临床应用奠定基础。

肝纤维化为我国带来了巨大的医疗负担，而肝纤维化的分级诊断对于治疗有重要意义，但现有方法或多或少存在一定的不足，因此寻求无创、准确、快速的肝纤维化诊断和分级方法，仍然具有非常现实和重要的临床价值。.已有研究表明超声非线性参数B/A能够区分不同程度肝纤维化，但如何在活体中获取B/A仍然值得探索。本研究拟提出一种能够基于使用现有超声仪而计算活体B/A的方法，并验证方法的准确性。.本课题的主要研究内容主要包括以下：.肝纤维化模型制作：预实验使用巴马香猪，但是巴马香猪的皮下脂肪厚度过高，与人类相差较为明显，且对超声回波信号影响较大，因此改用新西兰白兔作为实验动物进行实验，并成功建立新西兰白兔肝纤维化—肝硬化模型，最终共39只白兔纳入研究。.计算B/A的算法：根据高、低电压两组回波信号的基波与谐波幅值，可粗略计算介质的B/A值，并且与金标准测得的B/A值有一定相似度。本实验方法测得的丙三醇、正常兔肝及肝纤维化兔肝的B/A值在7.13-7.31、7.77-8.13及7.72-8.27的范围内。.B/A诊断肝纤维化的准确性：本研究测得的B/A值诊断兔显著肝纤维化、进展期肝纤维化及肝硬化的AUROC分别在0.55-0.65、0.51-0.74及0.67-0.73的范围内。.深度学习的应用：直接计算获取的非线性参数B/A不同等级肝纤维化肝脏之间的差异较小，推测有未知参数影响了计算结果。为了解决可能存在的未知参数对实验的影响，进一步使用深度学习的方法诊断肝纤维化。使用卷积神经网络（CNN）学习超声非线性图像。最终发现深度学习的方法诊断显著肝纤维化、进展期肝纤维化及肝硬化的AUROC分别在0.6-0.83、0.61-0.90及0.63-0.93的范围内。.本研究提出了一种基于有限振幅回波信号的非线性参数B/A计算方法，使用现有超声仪，从高、低电压回波信号中基波与二次谐波的振幅成功计算得到了无气水、丙三醇的非线性参数B/A值， 并且与金标准测得的B/A值有一定相似度，但是准确性不足，而基于超声非线性图像的深度学习方法能够有效提高准确性，但是仍待进一步验证。