Abstract
dc:descriptionThis doctoral thesis, entitled “Histogram-Based Analysis of Intracranial Tumors,” focuses on the use of advanced magnetic resonance imaging (MRI) and radiomics analysis for the non-invasive evaluation and prognostication of intracranial neoplasms, with particular emphasis on glioblastoma (GBM). The research comprises two complementary studies that investigate (a) radiomic predictors of survival, and (b) microstructural and topological reorganizations induced by the tumor in the brain.Study 1 – Survival Outcome Prediction in Glioblastoma: Insights fromMRI Radiomics (Curr Oncol, 2024) The first study assessed the prognostic value of radiomic features extracted from multiparametric diffusion and perfusion MRI for predicting survival in patients with IDH wild-type glioblastoma. The cohort included 49 patients who underwent pre-treatment MRI. Diffusion and perfusion maps were generated, and the tumor was segmented into four subregions: edema, necrotic, enhancing, and non-enhancing solid areas. A total of 1,862 radiomic features (morphological, first-order, and texture features) were extracted from each subregion and imaging map. After removing highly correlated features (r > 0.7) and performing univariate Cox regression, 25 statistically significant features were identified and incorporated into a multivariate Cox model with LASSO regularization. The final model identified smaller necrotic subregion volume, heterogeneous vascularization in the non-enhancing solid area, and tissue disruption within enhancing and edematous regions as significant predictors of poor prognosis. Age emerged as an independent prognostic factor. The model demonstrated satisfactory performance (C-index = 0.66, 95% CI 0.54–0.80, AUC > 0.65 across all time points) following internal and external validation (bootstrapping, hdnom package). This study supports that diffusion- and perfusion-derived radiomic features serve as non-invasive prognostic biomarkers of survival, reflecting the microstructural and vascular heterogeneity of GBM.Study 2 – The impact of glioblastoma on the contralesional hemisphere: Insightsfrom structural connectivity and cortical surface analysis (Neuroradiology, 2025) The second study explored neuroplastic and microstructural changes in the contralesional (opposite) hemisphere of patients with glioblastoma, aiming to elucidate mechanisms of neural reorganization and compensatory connectivity induced by tumor presence. The dataset included MRI scans from 31 newly diagnosed IDH wild-type glioblastoma patients who underwent pre-treatment multiparametric MRI, including diffusion tensor imaging (DTI). White matter tractography and graph-theoretical network models were generated to quantify changes in network topology and microstructure within the contralesional hemisphere. Results revealed that glioblastoma triggers neuroplastic hypertrophy in homologous cortical regions of the contralesional hemisphere, likely representing compensatory functional mechanisms, whereas high-metabolic-demand hub regions exhibited cortical atrophy, possibly due to energy depletion. Tumor-related white matter infiltration appeared to activate alternative contralesional pathways that may support functional recovery but also facilitate tumor cell migration. Correlations between structural alterations, Ki-67 proliferation index, and patient survival suggest that these microstructural changes reflect both tumor aggressiveness and the dynamics of neuroplastic remodeling. Overall ContributionThis dissertation combines two complementary approaches—the radiomic quantification of tumor heterogeneity and the mapping of brain reorganization mechanisms—to provide a comprehensive understanding of glioblastoma. Through the combined use of multiparametric MRI, and advanced statistical and computational modeling, it highlights the complex interplay between tumor biology and neural architecture. The findings demonstrate that radiomic analysis can serve as a reliable non-invasive prognostic tool, while the observed topological alterations represent potential biomarkers of neuroplasticity and tumor behavior. Overall, this work contributes to a deeper understanding of glioblastoma pathophysiology and paves the way for personalized prognostic assessment and therapeutic targeting in neuro-oncology.
Degree
thesis:*- Grantor dc:publisher
- University of Ioannina
- Year dc:date
- 2026
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Styliara, Effrosyni
- Στυλιαρά, Ευφροσύνη
Subjects
dc:subject × 17Rights
- Language dc:language
- gre
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/61780