Related Experiment Video
Updated: May 31, 2025
09:07
Sonodynamic Therapy for the Treatment of Glioblastoma Multiforme in a Mouse Model Using a Portable Benchtop Focused Ultrasound System
Published on: February 10, 2023
2.3K
Prognostic Modeling of Overall Survival in Glioblastoma Using Radiomic Features Derived from Intraoperative
Santiago Cepeda1, Olga Esteban-Sinovas1, Vikas Singh2
1Department of Neurosurgery, Río Hortega University Hospital, 47014 Valladolid, Spain.
Cancers
|January 25, 2025
Summary
Intraoperative ultrasound radiomics show promise for predicting glioblastoma survival. Combining imaging features with clinical data significantly improved prognostic model accuracy in this multi-institutional study.
Area of Science:
- Neurosurgery
- Radiology
- Oncology
- Medical Imaging Analysis
Background:
- Accurate glioblastoma prognostic models are crucial for treatment optimization.
- Intraoperative ultrasound (iUS) is an underutilized imaging modality for glioblastoma survival prediction.
- This study investigates the prognostic value of iUS radiomics in a multi-institutional glioblastoma cohort.
Purpose of the Study:
- To evaluate the prognostic potential of intraoperative ultrasound (iUS) radiomics in glioblastoma patients.
- To compare the performance of iUS radiomics, clinical data, and their combination in predicting overall survival (OS).
Main Methods:
- Retrospective analysis of glioblastoma patients from the multicenter BraTioUS database.
- Extraction of radiomic features from a single 2D iUS slice per patient.
- Development of Cox proportional hazards models using radiomic features, clinical data, and combined data with fivefold cross-validation.
Main Results:
- A cohort of 114 glioblastoma patients was analyzed.
- The combined model (iUS radiomics + clinical data) achieved a superior concordance index (C-index) of 0.87.
- The combined model outperformed models using only radiomic features (C-index: 0.72) or clinical data (C-index: 0.73).
Conclusions:
- Intraoperative ultrasound radiomics capture unique glioblastoma features for tissue characterization.
- Radiomic features from iUS show significant potential for developing accurate glioblastoma survival prediction models.
- Combining iUS radiomics with clinical data enhances prognostic model performance.
