Value of internal stratification analysis of abdominal wall muscles in predicting complications after orthotopic liver transplantation
- VernacularTitle:基于CT图像的腹部肌肉内部分层分析对原位肝移植术后并发症的预测价值
- Author:
Xin SHI
1
;
Chongxiao LIANG
2
;
Bei ZHANG
1
;
Jiping WANG
1
Author Information
- Publication Type:Journal Article
- Keywords: Myosteatosis; Liver Transplantation; Postoperative Complications
- From: Journal of Clinical Hepatology 2025;41(2):314-321
- CountryChina
- Language:Chinese
- Abstract: ObjectiveTo divide the muscle into different subzones according to different density ranges using the stratified analysis on the basis of myosteatosis， and to investigate the effect of muscle density changes on complications （Clavien-Dindo grade ≥Ⅲ） after orthotopic liver transplantation （OLT）. MethodsA retrospective analysis was performed for the medical records of 145 patients who underwent OLT in The First Hospital of Jilin University from May 2013 to September 2020， and with the plain CT scan images of the largest level of lumbar 3 vertebrae of each patient as the original data， Neusoft Fatanalysis software was used to measure related muscle parameters. The independent-samples t test was used for comparison of normally distributed continuous data between two groups， and the Mann-Whitney U test was used for comparison of non-normally distributed continuous data between two groups. The chi-square test or Fisher test was for comparison of categorical data between two groups. RIAS software was used to extract clinical features and perform analysis and modeling， and three machine learning models of logistic regression （LR）， support vector machine （SVM）， and random forest （RFC） were constructed. The receiver operating characteristic （ROC） curve， the calibration curve， and the decision curve were plotted for each model to calculate the area under the ROC curve （AUC）， sensitivity， specificity， precision， F1 score， and accuracy. ResultsThe three machine learning models of LR-C， SVM-C， and RFC-C were established based on the 7 clinical features before muscle stratification analysis， among which the RFC-C model had an AUC of 0.803， a sensitivity of 0.588， and a specificity of 0.778 in the test set. Among the models of LR-CS， SVM-CS， and RFC-CS established based on the 16 clinical features after muscle stratification analysis， the LR-CS and SVM-CS models had an AUC of 0.852 in the test set， with a sensitivity of 0.765 and 0.706， respectively， and a specificity of 0.889 and 0.926， respectively. Comparison of the AUC， sensitivity， specificity， precision， F1 score， and accuracy of each model in the test set before and after muscle stratification analysis showed that there were improvements in the parameters of the predictive model after muscle stratification analysis. Comparison of the decision curves and calibration curves of each predictive model showed that the LR-CS and SVM-CS models had good efficacy in predicting postoperative complications （Clavien-Dindo grade≥Ⅲ） in OLT patients. ConclusionOn the basis of myosteatosis， the division of the muscle into different subzones according to different densities using the stratified analysis has a certain value in predicting postoperative complications in patients with OLT.
