Integrating DCE-MRI-Based Dural Drainage Function Indicators into Machine Learning Models for Improved Intracranial Tumor Prognosis.
Authors
Affiliations (9)
Affiliations (9)
- Department of Neurology Tongji Hospital Tongji Medical College Huazhong University of Science and Technology Wuhan China.
- College of Life Science and Technology Huazhong University of Science and Technology Wuhan China.
- Department of Neurology The Second Affiliated Hospital of Nanchang University, Nanchang Jiangxi China.
- Department of Oncology Tongji Hospital Tongji Medical College Huazhong University of Science and Technology Wuhan China.
- Central Research Institute United Imaging Healthcare Group Shanghai China.
- Paul C. Lauterbur Research Centre for Biomedical Imaging Shenzhen Institute of Advanced Technology Chinese Academy of Science Shenzhen Guangdong China.
- Shenzhen College of Advanced Technology University of Chinese Academy of Sciences Shenzhen Guangdong China.
- Wuhan Zhongke Industrial Research Institute of Medical Science Wuhan China.
- National Engineering Research Center for Big Data Technology and System Services Computing Technology and System Lab Cluster and Grid Computing Lab School of Computer Science and Technology Huazhong University of Science and Technology Wuhan China.
Abstract
Meningeal lymphatic vessels (mLVs) are crucial in intracranial tumor progression. This study investigated whether incorporating mLVs functional indicators into machine learning models enhances prognostic prediction for intracranial malignant tumors. We prospectively enrolled 246 patients, assessing baseline mLVs function via dynamic contrast-enhanced MRI. After 2.5 years' follow-up, 100 patients (51 survivors, 49 deceased) were finally included. The mean area under the receiver operating characteristic curve (AUROC) of the XGBoost model excluding DCE-MRI-based dural drainage function indicators (DDFIs) was 0.746 (95% CI: 0.621-0.871), which increased to 0.808 (95% CI: 0.733-0.883) with the inclusion of DDFIs. Kaplan-Meier survival curves demonstrated significantly better discrimination when DDFIs were included (<i>p</i> = 5.66 × 10<sup>-8</sup> vs. p = 1.22 × 10<sup>-4</sup>). The c-index of the Cox regression model excluding DDFIs was 0.919 (95% CI: 0.916-0.940), rising to 0.948 (95% CI: 0.946-0.955) with their inclusion. In the glioma subgroup (<i>n</i> = 43), AUROC rose from 0.804 (95% CI: 0.629-0.980) to 0.904 (95% CI: 0.717-1.000). These findings indicate that integrating mLVs function significantly refines long-term prognostic accuracy in intracranial malignant tumors, supporting its potential clinical utility.