FluenceFormer: Transformer-Driven Multi-Beam Fluence Map Regression for Radiotherapy Planning

TL;DR

FluenceFormer is a transformer-based framework for accurate, geometry-aware fluence map prediction in radiotherapy planning, addressing limitations of prior convolutional methods by capturing long-range dependencies and ensuring physically consistent plans.

Contribution

The paper introduces FluenceFormer, a novel transformer-based, two-stage model with a physics-informed loss for improved fluence map regression in radiotherapy planning.

Findings

Swin UNETR backbone achieves best performance among tested models.

Reduces energy error to 4.5% compared to benchmarks.

Statistically significant improvements in structural fidelity.

Abstract

Fluence map prediction is central to automated radiotherapy planning but remains an ill-posed inverse problem due to the complex relationship between volumetric anatomy and beam-intensity modulation. Convolutional methods in prior work often struggle to capture long-range dependencies, which can lead to structurally inconsistent or physically unrealizable plans. We introduce \textbf{FluenceFormer}, a backbone-agnostic transformer framework for direct, geometry-aware fluence regression. The model uses a unified two-stage design: Stage~1 predicts a global dose prior from anatomical inputs, and Stage~2 conditions this prior on explicit beam geometry to regress physically calibrated fluence maps. Central to the approach is the \textbf{Fluence-Aware Regression (FAR)} loss, a physics-informed objective that integrates voxel-level fidelity, gradient smoothness, structural consistency, and beam-wise energy conservation. We evaluate the generality of the framework across multiple transformer backbones, including Swin UNETR, UNETR, nnFormer, and MedFormer, using a prostate IMRT dataset. FluenceFormer with Swin UNETR achieves the strongest performance among the evaluated models and improves over existing benchmark CNN and single-stage methods, reducing Energy Error to $\mathbf{4.5\%}$ and yielding statistically significant gains in structural fidelity ($p < 0.05$).