Predicting Brain Tumor Response to Therapy using a Hybrid Deep Learning and Radiomics Approach

2025-09-08

AI Summary

Accurate assessment of glioblastoma (GBM) treatment response is critical for clinical decision-making, yet current RANO criteria rely on subjective visual interpretation and suffer from inter-observer variability. To address this, we propose an automated, interpretable four-class longitudinal prediction framework integrating deep learning and radiomics. Specifically, fine-tuned ResNet-18 extracts 2D region-of-interest features from multimodal MRI; these are fused with high-dimensional 3D radiomic features (>4,800), dynamic tumor volume changes, and centroid displacement—clinically meaningful quantitative biomarkers. CatBoost is employed for classification. To our knowledge, this is the first GBM response prediction method in the BraTS challenge to jointly model multimodal imaging, volumetric dynamics, spatial shifts, and handcrafted radiomic features. In the four-class response classification task, it achieves a mean ROC AUC of 0.81 and Macro F1-score of 0.50, significantly improving both predictive accuracy and reproducibility over conventional approaches.

Technology Category

• Machine Learning: Multimodal Learning
• Computer Vision: Multi-modal Vision
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization

Abstract

Accurate evaluation of the response of glioblastoma to therapy is crucial for clinical decision-making and patient management. The Response Assessment in Neuro-Oncology (RANO) criteria provide a standardized framework to assess patients' clinical response, but their application can be complex and subject to observer variability. This paper presents an automated method for classifying the intervention response from longitudinal MRI scans, developed to predict tumor response during therapy as part of the BraTS 2025 challenge. We propose a novel hybrid framework that combines deep learning derived feature extraction and an extensive set of radiomics and clinically chosen features. Our approach utilizes a fine-tuned ResNet-18 model to extract features from 2D regions of interest across four MRI modalities. These deep features are then fused with a rich set of more than 4800 radiomic and clinically driven features, including 3D radiomics of tumor growth and shrinkage masks, volumetric changes relative to the nadir, and tumor centroid shift. Using the fused feature set, a CatBoost classifier achieves a mean ROC AUC of 0.81 and a Macro F1 score of 0.50 in the 4-class response prediction task (Complete Response, Partial Response, Stable Disease, Progressive Disease). Our results highlight that synergizing learned image representations with domain-targeted radiomic features provides a robust and effective solution for automated treatment response assessment in neuro-oncology.

Problem

• Automating brain tumor response classification from MRI scans
• Overcoming observer variability in glioblastoma therapy assessment
• Integrating deep learning with radiomics for response prediction

Innovation

• Hybrid deep learning and radiomics for tumor response
• Fine-tuned ResNet-18 extracts 2D MRI features
• CatBoost classifier fuses 4800+ radiomic clinical features