Machine learning model based on multi-parameter MRI radiomics for differentiating chordoma from chondrosarcoma of the skull base:a multi-center retrospective study
FAN Cungeng
YUE Yushu
ZHANG Ling
Abstract：Objective To explore the value of machine learning models based on multi-parameter MRI biomarkers in the preoperative differentiation between skull base chordoma and skull base chondrosarcoma.Methods A total of 180 patients with pathologically confirmed skull base tumors(136 cases of chordoma and 44 cases of chondrosarcoma)were retrospectively included.Their clinical characteristics were collected.The tumor regions were segmented using the nnUNet framework and MRI radiomics features were extracted.Three subsets consisting of clinical characteristics,radiomics features,and fusion features were constructed and preprocessed.Then,11 machine learning models were trained with the pathological results as labels,and the optimal model was selected based on indicators such as the area under the ROC curve(AUC)and accuracy.Results In the integrated feature model,the logistic regression model performed the best,with an AUC of 0.92(95%CI:0.86-0.98)and an accuracy rate of 87%.Baseline analysis revealed that tumor diameter,calcification,fibrous septum,and enhancement degree were statistically significant for discrimination(P<0.05).Decision curve analysis indicated that the integrated model had a higher net benefit rate within the clinical risk threshold range of 6%-91%.Conclusion The fusion model based on multi-parameter MRI imaging histology and clinical features can effectively distinguish between skull base chordoma and skull base chondrosarcoma,providing a reference for precise preoperative diagnosis and treatment.
Keywords：MRI radiomicsmachine learningskull base tumormulticenter study
Publication Date：2025-12-20
Online Publishing Date：2026-01-07(First online date of this platform, not the publication date of the document)
Pages：6( 1526-1531 )