基于太原市两所三级医院CT影像资料模拟建立成人马蹄肾患者肾脏深度估算公式  

A formula for estimating the depth of the kidney in adult horseshoe kidney patients based on CT imaging data from two tertiary hospitals in Taiyuan

作　　者：宋一东 姜子德 马晨 郝丽萍[1] Song Yidong;Jiang Zide;Ma Chen;Hao Liping(Department of Office,Taiyuan Hospital of Intergrative Medicine,Taiyuan 030003,China;Department of Operating Peking University First Hospital Taiyuan Hospital(Taiyuan Central Hospital),Taiyuan 030009,China)

机构地区：[1]太原中西医结合医院办公室,太原030003 [2]北京大学第一医院太原医院(太原市中心医院)手术室,太原030009

出　　处：《实用医技杂志》2025年第2期92-97,I0001,共7页Journal of Practical Medical Techniques

摘　　要：目的基于太原市两所三级医院CT影像资料评估现有肾脏深度估算公式在马蹄肾患者中的适用性,并建立适用于马蹄肾患者肾脏及峡部深度的估算公式。方法回顾2022年1月至2024年8月在太原中西医结合医院及太原市中心医院进行CT平扫确诊马蹄肾的84例成人患者,测量其肾脏深度、峡部深度、峡部椎体厚度及肾区身体总厚度(T);记录患者性别、年龄(Age)、身高及体质量(W);CT测量深度为金标准。将84例患者随机分为2组,第1组用于评估现有肾脏深度估算公式在马蹄肾患者中的适用性,第2组用于建立新的马蹄肾患者肾脏及峡部深度的估算公式。采用多元逐步线性回归分析获得马蹄肾患者肾脏及峡部深度的新公式,然后使用第1组数据验证新公式的准确性。随后利用机器学习进一步优化肾深度估算模型,计算决定系数(R^(2))、均方误差(MSE)和平均绝对误差(MAE)以评估模型性能。同时将新公式作为对照,比较其预测精度。结果①现有肾脏深度估算公式均明显低估马蹄肾患者肾脏深度,双肾深度平均误差最大约为-2.673 cm,最小约为-1.174 cm。②多元逐步线性回归分析表明,马蹄肾患者肾脏深度估算的重要变量为:W和T;峡部软组织深度和椎体厚度估算的重要变量分别为:W、T和Age、W。新公式为:右肾深度(cm)=0.273×T+0.043×W+1.086(r=0.821,P<0.05;标准回归系数:肾区身体总厚度=0.500,体质量=0.367),左肾深度(cm)=0.245×T+0.041×W+0.676(r=0.833,P<0.05;标准回归系数:肾区身体总厚度=0.520,体质量=0.353);峡部深度(cm)=软组织深度+椎体厚度,软组织深度(cm)=0.144×T+0.044×W+0.536(r=0.580,P<0.05;标准回归系数:身体总厚度=0.272,体质量=0.335),椎体厚度(cm)=0.012×Age+0.018×W+3.683(r=0.532,P<0.05;标准回归系数:年龄=0.326,体质量=0.438)。③新公式所估算肾区身体总厚度的肾脏深度准确性优于现有公式,且新公式适用于估算马蹄肾患者峡�Objective The objective of this study is as follows:firstly,to assess the applicability of existing renal depth estimation formulae in patients with horseshoe kidneys based on CT imaging data from two tertiary hospitals in Taiyuan City;secondly,to establish estimation formulae applicable to renal and isthmus depths in patients with horseshoe kidneys.Methods A total of 84 adult patients with a confirmed diagnosis of horseshoe kidneys by CT scanning at Taiyuan Hospital of Integrative Medicine and Taiyuan Central Hospital between January 2022 and August 2024 were included in the review.Their renal depths,isthmus depths,and isthmus vertebral body thicknesses were measured.The thicknesses of the thoracic vertebral bodies and the total renal area body thicknesses(T,cm)were measured.The patients′sexes,ages,heights(H,cm),and body weights(W,kg)were recorded.The depths of the CT measurements were used as the gold standard.The 84 patients were randomly divided into two groups.The first group was used to evaluate the applicability of the existing renal depth estimation formula in patients with horseshoe kidneys,while the second group was used to establish a new renal and isthmus depth estimation formula in patients with horseshoe kidneys.A series of stepwise linear regression analyses were employed to derive novel formulas for kidney and isthmus depths in patients with horseshoe kidneys.The accuracy of these new formulas was subsequently validated using data from the initial cohort.Subsequently,the kidney depth estimation model underwent further optimisation through the implementation of machine learning,whereby the coefficient of determination(R^(2)),mean square error(MSE)and mean absolute error(MAE)were calculated for the purpose of evaluating the model performance.Additionally,the new formula was utilised as a control for the purpose of comparing its prediction accuracy.Results①The extant formulas for estimating kidney depth all yielded significantly inaccurate results for patients with horseshoe kidneys.The mean abso

关 键 词：融合肾 肾小球滤过率 体层摄影术 X线计算机 

分 类 号：G63[文化科学—教育学]