Search papers, labs, and topics across Lattice.
This paper introduces a clinical DVH metric loss (CDM loss) for deep learning-based 3D dose prediction in head and neck radiotherapy, incorporating differentiable D-metrics and surrogate V-metrics. The CDM loss is combined with a lossless bit-mask ROI encoding to improve training efficiency. Experiments on 174 H&N patients demonstrate that CDM loss significantly improves target coverage and satisfies clinical constraints compared to MAE- and DVH-curve based losses, while the bit-mask encoding reduces training time by 83%.
By directly optimizing clinical dose-volume histogram (DVH) metrics, this method produces 3D dose predictions that more closely align with clinical treatment planning criteria than traditional voxel-wise approaches.
Purpose: Deep-learning-based three-dimensional (3D) dose prediction is widely used in automated radiotherapy workflows. However, most existing models are trained with voxel-wise regression losses, which are poorly aligned with clinical plan evaluation criteria based on dose-volume histogram (DVH) metrics. This study aims to develop a clinically guided loss formulation that directly optimizes clinically used DVH metrics while remaining computationally efficient for head and neck (H\&N) dose prediction. Methods: We propose a clinical DVH metric loss (CDM loss) that incorporates differentiable \textit{D-metrics} and surrogate \textit{V-metrics}, together with a lossless bit-mask region-of-interest (ROI) encoding to improve training efficiency. The method was evaluated on 174 H\&N patients using a temporal split (137 training, 37 testing). Results: Compared with MAE- and DVH-curve based losses, CDM loss substantially improved target coverage and satisfied all clinical constraints. Using a standard 3D U-Net, the PTV Score was reduced from 1.544 (MAE) to 0.491 (MAE + CDM), while OAR sparing remained comparable. Bit-mask encoding reduced training time by 83\% and lowered GPU memory usage. Conclusion: Directly optimizing clinically used DVH metrics enables 3D dose predictions that are better aligned with clinical treatment planning criteria than conventional voxel-wise or DVH-curve-based supervision. The proposed CDM loss, combined with efficient ROI bit-mask encoding, provides a practical and scalable framework for H\&N dose prediction.
