59794 Baseline 3D Fractal Dimension of Peritumoral Edema Predicts Survival and Early MRI Response in Glioma Patients
Summary
No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.
Summarise with VIRAT Free to try with your Science Catalogs account.Abstract
INTRODUCTION: Fractal geometry captures multiscale complexity of tumor and peritumoral regions on MRI. Comparing both 2D and 3D fractal dimensions (FD) may reveal complementary insights into invasive growth and treatment response not evident from conventional size metrics. METHODS: We analyzed the LUMIÈRE dataset (Figshare collection 5904905) of WHO grade II–IV glioma patients, including pre-treatment FLAIR and T1 post-contrast MRI, automated 3D segmentations, demographics, pathology, and expert RANO ratings. We computed compartmental 2D and 3D FDs using the Minkowski–Bouligand box-counting algorithm. Pearson correlation tested FD vs. OS, and Kruskal–Wallis compared baseline FD between non-responders (SD/PD) and responders (PR/CR). RESULTS: Baseline edema FD inversely correlated with OS for both 2D (r = –0.23, p = 0.048) and 3D (r = –0.25, p = 0.042), whereas necrotic and enhancing FDs showed no significant associations (all p > 0.8). First-time point RANO stratification revealed significantly higher baseline 2D and 3D FDs in SD/PD vs. PR/CR across allcompartments (p < 0.01). CONCLUSIONS: Higher baseline fractal complexity of peritumoral edema, measured in both 2D and 3D, is linked to poorer survival and early treatment resistance in glioma patients. This study uniquely compares 2D vs. 3D FD performance. Future work should validate these findings on larger, more balanced cohorts to confirm the relative prognostic power of 2D and 3D fractal metrics.
Fields of Study & Subjects
Radiomics and Machine Learning in Medical Imaging
Glioma Diagnosis and Treatment
Brain Tumor Detection and Classification
Publication details
- Journal / source
- Neurosurgery
- Volume
- 72
- Issue
- Supplement_1
- Pages
- 312–312
- Publication date
- 26 March 2026
- Language
- EN