J Neurosurg 142:1298–1306, 2025
This study compares manual and automatic methods for assessing tumor resection extent and residual volume in glioblastoma patients. It finds that both methods have comparable prognostic value, suggesting that automatic segmentation with Raidionics is a viable alternative for future studies.
• Objective: The study compares the prognostic value of manual versus automatic methods for assessing the extent of resection (EOR) and residual tumor (RT) volume in glioblastoma patients.
• Methods: Patients from 12 hospitals in Europe and North America underwent glioblastoma resection and were included in the study. Data were collected from local tumor registries and patient medical records.
• Results: Both manual and automatic RT volumes were negative prognostic factors for overall survival. Automatic segmentation with Raidionics showed comparable prognostic properties to manual measurements.
• Automatic Segmentation: Raidionics, an open-access software, performed automatic segmentation using pretrained deep learning models, which showed high quality and robustness.
• Survival Analysis: Cox regression models indicated that patients with gross-total resection had significantly longer overall survival compared to those with subtotal resection.
• Advantages of Automatic Methods: Automatic segmentation offers fast, quantitative image assessments and reduces interobserver variability, making it suitable for clinical trials.
• Limitations: Some cases showed a mismatch between manual and automatic segmentation, often due to poor-quality MR images or heterogeneous tumors.
• Conclusion: Automatic segmentation is a viable alternative to manual methods for evaluating tumor remnants, with similar prognostic value for survival in glioblastoma patients.


You must be logged in to post a comment.