目的: 探讨基于影像组学的超声图像在肝纤维化诊断中的应用价值。方法: 将SD大鼠随机分为对照组和模型组,模型组通过腹腔注射40%四氯化碳-橄榄油混悬液建立肝纤维化模型,对照组给予等剂量生理盐水。成模后,选取大鼠肝脏无明显血管结构的切面进行超声成像。成像结束后处死大鼠,取肝组织行HE染色,依据《病毒性肝炎防治方案》中的肝纤维化分期标准进行病理学评估,以病理结果为金标准,验证超声影像组学模型对大鼠肝纤维化的诊断效能。采用最小绝对值收缩与选择算子(LASSO)方法筛选影像组学特征,结合逻辑回归算法构建基于超声影像的组学诊断模型,并通过受试者工作特征曲线(ROC)、校准曲线及决策曲线评估其诊断价值。结果: 共纳入34只大鼠,其中模型组28只,对照组6只,每只大鼠提取477个影像组学特征。经过LASSO分析,筛选出2个关键影像组学特征: X20(灰度共生矩阵对比度,GLCM)和 X187(小区域突出,SAE)。采用上述2个特征构建多因素logistics回归模型,其ROC曲线下面积为0.929(95%CI,0.843-1),敏感度为0.929(95%CI,0.833-1),特异度为0.800(95%CI,0.552-1),阳性预测值为0.929(95%CI,0.833-1),阴性预测值为0.800(95%CI,0.552-1),F1分数为0.929。结论: 超声影像组学模型对于肝纤维化的预测概率与实际概率具有较好的拟合度,具有临床应用的潜力。
Objective: To explore the application value of ultrasound images based on radiomics in the diagnosis of liver fibrosis. Methods: SD rats were randomly divided into control and model groups. Liver fibrosis was induced in the model group by intraperitoneal injection of 40% carbon tetrachloride-olive oil suspension, while controls received equal saline. Ultrasound imaging was performed on liver sections without obvious vascular structures. After imaging, rats were euthanized and liver tissues were collected for HE staining. Pathological evaluation was conducted according to the liver fibrosis staging criteria (Prevention and Treatment Strategies for Viral Hepatitis), which served as the gold standard to validate the diagnostic efficacy of the ultrasound-based radiomics model. Radiomics features were selected using the least absolute shrinkage and selection operator (LASSO) method, and a logistic regression model was constructed. Diagnostic performance was assessed using receiver operating characteristic (ROC) curve, calibration curve, and decision curve analysis. Results: A total of 34 rats were finally enrolled for radiomics analysis, including 28 in the model group and 6 in the control group. 477 radiomics features were extracted from each rat. After LASSO analysis, two key radiomics features were selected: X20:gray level co-occurrence matrix contrast (GLCM) and X187:small area prominent (SAE). The area under the ROC curve was 0.929(95%CI, 0.843-1), the sensitivity was 0.929(95%CI, 0.833-1), the specificity was 0.800(95%CI, 0.552-1), the positive predictive value was 0.929(95%CI, 0.833-1), the negative predictive value was 0.800(95%CI, 0.552-1), and the F1 score was 0.929. Conclusion: The ultrasound radiomics model has a good fit between the predicted probability and the actual probability of liver fibrosis, and has the potential for clinical application.
[1] Roehlen N, Crouchet E, Baumert TF. Liver Fibrosis: Mechanistic Concepts and Therapeutic Perspectives[J]. Cells, 2020, 9(4):2.
[2] Man S, Deng Y, Ma Y, et al. Prevalence of Liver Steatosis and Fibrosis in the General Population and Various High-Risk Populations: A Nationwide Study With 5.7 Million Adults in China[J]. Gastroenterology, 2023, 165(4): 1025-1040.
[3] Chen H, Cai J, Wang J, et al. Targeting Nestin+ hepatic stellate cells ameliorates liver fibrosis by facilitating TβRI degradation[J]. J Hepatol, 2021, 74(5): 1176-1187.
[4] Ginès P, Graupera I, Lammert F, et al. Screening for liver fibrosis in the general population: a call for action[J].Lancet Gastroenterol Hepatol, 2016, 1(3): 256-260.
[5] 张倩, 李志艳. 超声技术在肝纤维化诊断中的应用进展[J]. 传染病信息, 2021, 34(2): 177-180.
[6] 赫晓磊, 王俊, 王艳. CT影像组学可辅助诊断慢性乙型肝炎肝纤维化[J]. 分子影像学杂志, 2024, 47(7): 732-735.
[7] Li X, Li R, Wang M, et al. Emodin alleviates liver fibrosis by inhibiting ferroptosis via the PI3K/Akt signaling pathway[J]. Mol Med Rep, 2021, 24(5): 1-11.
[8] 中华医学会传染病与寄生虫病分会, 中华医学会肝病学分会.病毒性肝炎防治方案[J]. 中华传染病杂志, 2001, 19(1): 55-61.
[9] Khanam A, Saleeb PG, Kottilil S. Pathophysiology and Treatment Options for Hepatic Fibrosis:Can It Be Completely Cured[J]. Cells, 2021, 10(5):11-13.
[10] 詹晓丽, 徐莉萍, 金晨阳, 等. 超声影像组学对桥本甲状腺炎背景下甲状腺结节的诊断价值[J]. 中国超声医学杂志, 2023, 39(9): 961-965.
[11] Haralick RM, Shanmugam K, Dinstein I.Textural Features for Image Classification[J]. IEEE Trans Syst Man Cybern, 1973, 3(6): 610-621.
[12] Kumar V, Gujral DM, Bharath AA. Radiomics: the process and the challenges[J]. Magn Reson Imaging, 2012, 30(9): 1234-1248.
[13] Thibault G, Fertil B, Navarro C, et al. Shape and texture indexes application to cell nucleiclassification[J]. Int J Pattern Recognit Artif Intell, 2013, 27(1): 1357002.
[14] Zwanenburg A, Vallieres M, Abdalah MA, et al. The Image Biomarker Standardization Initiative[J]. Radiology, 2020, 295(2): 328-338.
[15] 付甜甜, 蒋卓韵, 丁红, 等. 计算机辅助超声影像组学预测肝纤维化进程的实验研究[J]. 中国超声医学杂志, 2019, 35(12): 1131-1133.
