Estimation of liver elasticity using the finite element method and four‐dimensional computed tomography images as a biomarker of liver fibrosis
Estimation of liver elasticity using the finite element method and four‐dimensional computed tomography images as a biomarker of liver fibrosis
复制标题
使用有限元法和四维计算机断层扫描图像估计肝脏弹性作为肝纤维化的生物标志物
DOI：
10.1002/mp.14723
复制
发表时间：
2021
期刊：
影响因子：
3.8
通讯作者：
Tanaka Hidekazu
中科院分区：
文献类型：
--
作者：
Fujimoto Koya;Shiinoki Takehiro;Yuasa Yuki;Tanaka Hidekazu
目的当前的放射治疗计划程序通常仅根据解剖信息设计，并使用不包含器官功能信息的计算机断层扫描（CT）图像。在本研究中，我们开发了一种使用有限元法 (FEM) 和放射治疗计划期间获取的四维 CT (4DCT) 图像来估计肝脏弹性的方法，随后我们评估了其作为肝纤维化生物标志物的可行性。 材料和方法 纳入了 20 名接受 4DCT 和超声瞬态弹性成像 (UTE) 的患者。所有患者均患有慢性肝病或肝硬化。对 20 名患者的肝脏右叶进行了 UTE 的肝脏弹性测量。在 20 名患者中，有 18 名患者获得了天冬氨酸转氨酶 (AST) 与血小板比值指数 (APRI) 和纤维化 4 指数 (FIB-4) 的血清生物标志物，并在接受 4DCT 后 1 周内进行了测量。在呼气和吸气端点获得的 4DCT 图像之间的位移是使用实际（通过可变形图像配准）和模拟（通过 FEM）呼吸引起的位移来确定的。通过最小化实际和模拟呼吸引起的位移之间的误差来优化肝脏模型每个元件的弹性。然后，使用估计弹性图将每位患者的估计肝脏弹性定义为肝脏右叶和整个肝脏的平均杨氏模量。评估估计的肝脏弹性与通过 UTE 获得的弹性和两种血清生物标志物（APRI 和 FIB-4）的相关性。结果肝脏模型中实际呼吸诱导位移与模拟呼吸诱导位移之间误差的平均值±标准差 (SD) 为 0.54 ± 0.33 毫米。估计的肝脏右叶弹性与 UTE 具有统计学显着相关性（r = 0.87，P < 0.001）。此外，估计的整个肝脏弹性与 UTE（r = 0.84，P < 0.001）、APRI 评分（r = 0.62，P = 0.005）和 FIB-4 评分（r = 0.54，P = 0.021）具有统计显着相关性。结论在本研究中，通过基于 FEM 的模拟和从 4DCT 图像获得的实际呼吸引起的肝脏位移来估计肝脏弹性。此外，我们评估肝脏右叶的估计弹性与 UTE 密切相关。因此，估计的弹性有可能成为仅使用 4DCT 图像评估肝纤维化的可行成像生物标志物，而无需额外的检查或设备成本。由于我们的结果来自 20 名患者的有限样本，因此有必要评估较大组活检患者的每个肝段弹性估计的准确性，以便利用肝脏弹性信息制定放射治疗计划。
PurposeCurrent radiotherapy planning procedures are generally designed based on anatomical information only and use computed tomography (CT) images that do not incorporate organ‐functional information. In this study, we developed a method for estimating liver elasticity using the finite element method (FEM) and four‐dimensional CT (4DCT) images acquired during radiotherapy planning, and we subsequently evaluated its feasibility as a biomarker for liver fibrosis.Materials and methodsTwenty patients who underwent 4DCT and ultrasound‐based transient elastography (UTE) were enrolled. All patients had chronic liver disease or cirrhosis. Liver elasticity measurements of the UTE were performed on the right lobe of the patient’s liver in 20 patients. The serum biomarkers of the aspartate aminotransferase (AST)‐to‐platelet ratio index (APRI) and fibrosis‐4 index (FIB‐4) were available in 18 of the 20 total patients, which were measured within 1 week after undergoing 4DCT. The displacement between the 4DCT images obtained at the endpoints of exhalation and inspiration was determined using the actual (via deformable image registration) and simulated (via FEM) respiration‐induced displacement. The elasticity of each element of the liver model was optimized by minimizing the error between the actual and simulated respiration‐induced displacement. Then, each patient's estimated liver elasticity was defined as the mean Young’s modulus of the liver’s right lobe and that of the whole liver using the estimated elasticity map. The estimated liver elasticity was evaluated for correlations with the elasticity obtained via UTE and with two serum biomarkers (APRI and FIB‐4).ResultsThe mean ± standard deviation (SD) of the errors between the actual and simulated respiration‐induced displacement in the liver model was 0.54 ± 0.33 mm. The estimated liver’s right lobe elasticity was statistically significantly correlated with the UTE (r = 0.87,P< 0.001). Furthermore, the estimated whole liver elasticity was statistically significantly correlated with the UTE (r = 0.84,P< 0.001), APRI score (r = 0.62,P= 0.005), and FIB‐4 score (r = 0.54,P= 0.021).ConclusionIn this study, liver elasticity was estimated through FEM‐based simulation and actual respiratory‐induced liver displacement obtained from 4DCT images. Furthermore, we assessed that the estimated elasticity of the liver’s right lobe was strongly correlated with the UTE. Therefore, the estimated elasticity has the potential to be a feasible imaging biomarker for assessing liver fibrosis using only 4DCT images without additional inspection or equipment costs. Because our results were derived from a limited sample of 20 patients, it is necessary to evaluate the accuracy of elasticity estimation for each liver segment on larger groups of biopsied patients to utilize liver elasticity information for radiotherapy planning.