TRACE: A Concept Bottleneck Model for Longitudinal 3D Glioblastoma Response Assessment

2026-06-29

AI Summary

Current deep learning approaches for predicting glioblastoma treatment response lack clinical interpretability and intervenability. This work proposes TRACE, a novel model that formalizes the RANO 2.0 guidelines into a computable concept graph for the first time. TRACE employs a shared 3D encoder to extract tumor measurements from longitudinal multimodal MRI as foundational concepts and applies deterministic rules to derive downstream RANO assessments. The architecture incorporates concept bottlenecks and intervention-consistent training to ensure clinical alignment and model correctability. Evaluated on the LUMIERE dataset, TRACE achieves a macro F1 score of 0.4769 for four-class response classification and 0.7085 for binary progression/non-progression prediction—performance comparable to black-box models—while demonstrating that targeted concept interventions can further improve predictive accuracy.

Abstract

Longitudinal glioblastoma response assessment requires comparing subtle tumor changes across MRI time points using structured clinical criteria such as RANO. However, most deep learning methods predict response labels directly from imaging features, which limits clinical inspection, verification, and correction. We introduce TRACE, a RANO 2.0-aligned concept bottleneck model for interpretable 4-class glioblastoma response classification on longitudinal 3D MRI. TRACE processes paired baseline and follow-up multimodal MRI scans with a shared 3D vision encoder, predicts clinically meaningful tumor measurements as root concepts, computes downstream RANO-derived concepts through deterministic rules, and incorporates scan interval and new-lesion information as passthrough concepts. This design frames response assessment as structured concept reasoning rather than direct image-to-label prediction. Using 5-fold patient-wise cross-validation on the LUMIERE dataset, TRACE achieves a 4-class macro F1 of 0.4769 and a binary progression-versus-non-progression macro F1 of 0.7085. It improves over a concept bottleneck baseline and remains within the range of published non-interpretable deep learning approaches. Ablation studies show that the expert RANO graph and intervention-consistency training are important for performance, while intervention experiments demonstrate that correcting concepts can improve downstream predictions. These results suggest that structured concept bottlenecks offer a transparent and clinically aligned direction for longitudinal glioblastoma response assessment, while highlighting the need for larger protocol-aligned datasets and external validation.

Problem

• glioblastoma
• longitudinal response assessment
• RANO criteria
• interpretability
• concept bottleneck

Innovation

• Concept Bottleneck Model
• Longitudinal 3D MRI
• RANO 2.0
• Interpretable AI
• Glioblastoma Response Assessment