Site-Agnostic 3D Dose Distribution Prediction with Deep Learning Neural Networks

TL;DR

This paper introduces a deep learning model capable of predicting 3D dose distributions across multiple treatment sites, requiring minimal fine-tuning for new sites, thereby enhancing flexibility and data efficiency in radiotherapy planning.

Contribution

The authors develop a site-agnostic deep learning model for 3D dose prediction that adapts to new treatment sites with minimal fine-tuning, unlike previous site-specific models.

Findings

Model can be fine-tuned for new sites with limited data

No changes needed to input channels or model parameters for new sites

Improves data utilization and adaptability in dose prediction

Abstract

Typically, the current dose prediction models are limited to small amounts of data and require re-training for a specific site, often leading to suboptimal performance. We propose a site-agnostic, 3D dose distribution prediction model using deep learning that can leverage data from any treatment site, thus increasing the total data available to train the model. Applying our proposed model to a new target treatment site requires only a brief fine-tuning of the model to the new data and involves no modifications to the model input channels or its parameters. Thus, it can be efficiently adapted to a different treatment site, even with a small training dataset.