Deep Reinforcement Learning for Data-Driven Adaptive Scanning in Ptychography
This work addresses dose reduction in ptychography, a domain-specific imaging technique, with incremental improvements over existing methods.
The paper tackles the problem of reducing radiation dose in ptychographic imaging by using deep reinforcement learning to adaptively scan specimens, focusing on critical regions, and shows that this method outperforms conventional approaches in reconstruction resolution.
We present a method that lowers the dose required for a ptychographic reconstruction by adaptively scanning the specimen, thereby providing the required spatial information redundancy in the regions of highest importance. The proposed method is built upon a deep learning model that is trained by reinforcement learning (RL), using prior knowledge of the specimen structure from training data sets. We show that equivalent low-dose experiments using adaptive scanning outperform conventional ptychography experiments in terms of reconstruction resolution.