Spatial Brain Tumor Concentration Estimation for Individualized Radiotherapy Planning

2024-12-18

arXiv.org

AI Summary

Conventional MRI fails to detect microscopic tumor infiltration in gliomas, and current radiotherapy planning—relying on a fixed 15-mm margin—often results in undertreatment or overtreatment. Method: We propose the first end-to-end, soft physics-constrained optimization framework that directly estimates a 3D tumor cell concentration field from preoperative multimodal MRI. Our approach integrates biophysical priors (e.g., diffusion-proliferation dynamics) with data-driven learning, employing differentiable MRI forward modeling and a parameterized, physics-guided concentration field optimized under a 3D biologically informed loss. Results: Evaluated on a two-center cohort of 192 patients, our method achieves significantly higher recurrence prediction AUC than state-of-the-art methods. It enables sub-minute inference per case (30× speedup), facilitating real-time clinical deployment. Moreover, it demonstrates robustness to imperfect inputs and seamless compatibility with diverse MRI contrasts.

Technology Category

• Machine Learning: Optimization
• Search and Optimization: Learning to Search
• Computer Vision: Medical and Biological Imaging

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Biophysical modeling of brain tumors has emerged as a promising strategy for personalizing radiotherapy planning by estimating the otherwise hidden distribution of tumor cells within the brain. However, many existing state-of-the-art methods are computationally intensive, limiting their widespread translation into clinical practice. In this work, we propose an efficient and direct method that utilizes soft physical constraints to estimate the tumor cell concentration from preoperative MRI of brain tumor patients. Our approach optimizes a 3D tumor concentration field by simultaneously minimizing the difference between the observed MRI and a physically informed loss function. Compared to existing state-of-the-art techniques, our method significantly improves predicting tumor recurrence on two public datasets with a total of 192 patients while maintaining a clinically viable runtime of under one minute - a substantial reduction from the 30 minutes required by the current best approach. Furthermore, we showcase the generalizability of our framework by incorporating additional imaging information and physical constraints, highlighting its potential to translate to various medical diffusion phenomena with imperfect data.

Problem

• Estimates 3D brain tumor infiltration from MRI data
• Overcomes computational limits of current tumor growth models
• Improves radiotherapy planning with patient-specific tumor spread

Innovation

• Lightweight optimization for 3D tumor estimation
• Fits MRI segmentations with smooth concentration
• Rapid runtime under one minute