CAJ | 학술논문

目的探讨超声定量诊断大鼠肝纤维化的价值.方法收集正常或肝纤维化大鼠共90只的标准化声像图,观测肝包膜厚度并提取声像图纹理的13个灰度共生矩阵参数(熵、对比、方差、均值和、方差和、和熵、方差差、差熵及相关信息2递增、相关信息1、角二阶矩、相关及逆差矩)与肝纤维化病理学诊断结果比较,进行逐步判别分析建立判别大鼠肝纤维化程度的超声定量模型,用交互检验评价模型的效率.结果肝包膜厚度及其他13个指标与肝纤维化病理学分期均具有相关性(r值依次为0.817、0.894、0.808、0.844、0.828、0.795、0.864、0.725、0.821、0.848、-0.743、-0.909、-0.438、-0.855,P值均＜0.05),14个指标在病理学分期间的差异均具有统计学意义(F值分别为43.12、60.55、50.70、43.65、45.68、23.63、56.60、21.48、46.19、24.66、39.52、75.74、15.37、63.98,P值均＜0.05).交互检验结果显示建立于超声定量指标基础上的判别模型对大鼠肝纤维化S0、S1、S2、S3和S4分期的准确率分别为83.3％、84.2％、70.0％、50.0％和88.2％,73.3％的大鼠能够被准确分期；对无纤维化组(S0),轻度纤维化组(S1),中重度纤维化组(S2及S3)和早期肝硬化组(S4)分组的准确率分别为91.7%、84.2％、69.0％和88.2％,78.9％的大鼠能够被准确分组.结论超声检查结合声像图的纹理分析对定量诊断大鼠肝纤维化具有较高的准确性.
목적 탐토초성정량진단대서간섬유화적개치.방법 수집정상혹간섬유화대서공90지적표준화성상도,관측간포막후도병제취성상도문리적13개회도공생구진삼수(적、대비、방차、균치화、방차화、화적、방차차、차적급상관신식2체증、상관신식1、각이계구、상관급역차구)여간섬유화병이학진단결과비교,진행축보판별분석건립판별대서간섬유화정도적초성정량모형,용교호검험평개모형적효솔.결과 간포막후도급기타13개지표여간섬유화병이학분기균구유상관성(r치의차위0.817、0.894、0.808、0.844、0.828、0.795、0.864、0.725、0.821、0.848、-0.743、-0.909、-0.438、-0.855,P치균＜0.05),14개지표재병이학분기간적차이균구유통계학의의(F치분별위43.12、60.55、50.70、43.65、45.68、23.63、56.60、21.48、46.19、24.66、39.52、75.74、15.37、63.98,P치균＜0.05).교호검험결과현시건립우초성정량지표기출상적판별모형대대서간섬유화S0、S1、S2、S3화S4분기적준학솔분별위83.3％、84.2％、70.0％、50.0％화88.2％,73.3％적대서능구피준학분기；대무섬유화조(S0),경도섬유화조(S1),중중도섬유화조(S2급S3)화조기간경화조(S4)분조적준학솔분별위91.7%、84.2％、69.0％화88.2％,78.9％적대서능구피준학분조.결론 초성검사결합성상도적문리분석대정량진단대서간섬유화구유교고적준학성.
Objective To evaluate the efficacy of an ultrasound-based quantitative method to diagnose liver fibrosis using a rat model.Methods Ultrasonography was performed on the livers of 90Sprague-Dawley rats with or without thioacetamide-induced fibrosis.The liver capsule thickness and 13texture parameters of gray level co-occurrence matrix were extracted from the standard sonograms.After sacrifice,severity of liver fibrosis (S0-S4 classification) was diagnosed by histopathology.Analysis of variance and correlation statistical tests were used to analyze the differences between groups and determine the relationships between each of the 14 quantitative ultrasound index points and the histological results,respectively.Discriminant analysis models were developed for quantitative diagnosis of liver fibrosis,and the leave-one-case-out method was used to verify the efficiency of models.Results All 14 indices were significantly correlated with the histological stages of fibrosis (P ＜ 0.05).The accuracy of the discriminant model for S0,S1,S2,S3 and S4 was 83.3％,84.2％,70.0％,50.0％ and 88.2％,respectively.In addition,73.3％ of cross-validated rats were accurately classified.Grouping S0 as no fibrosis,S1 as mild fibrosis,S2with S3 as moderate to severe fibrosis and S4 as early cirrhosis increased the accuracy of the discriminant model for these four groups (respectively,91.7％,84.2％,69.0％ and 88.2％) and allowed for 78.9％ of crossvalidated rats to be correctly identified.Conclusion Ultrasonography combined with texture analysis was a novel and accurate method to diagnose liver fibrosis in a rat model; further studies may provide insights into its applicability for quantitating liver fibrosis in other animal models or in clinic.