Adaptive Partial Scanning Transmission Electron Microscopy with Reinforcement Learning
This work addresses the need for more efficient and specimen-adaptive microscopy techniques, representing an incremental improvement over traditional static compressed sensing methods.
The paper tackles the problem of reducing electron dose and scan time in scanning transmission electron microscopy by introducing an adaptive partial scanning system that dynamically adjusts scan paths based on the specimen, using a recurrent neural network trained with reinforcement learning to cooperate with a convolutional neural network for scan completion.
Compressed sensing can decrease scanning transmission electron microscopy electron dose and scan time with minimal information loss. Traditionally, sparse scans used in compressed sensing sample a static set of probing locations. However, dynamic scans that adapt to specimens are expected to be able to match or surpass the performance of static scans as static scans are a subset of possible dynamic scans. Thus, we present a prototype for a contiguous sparse scan system that piecewise adapts scan paths to specimens as they are scanned. Sampling directions for scan segments are chosen by a recurrent neural network based on previously observed scan segments. The recurrent neural network is trained by reinforcement learning to cooperate with a feedforward convolutional neural network that completes the sparse scans. This paper presents our learning policy, experiments, and example partial scans, and discusses future research directions. Source code, pretrained models, and training data is openly accessible at https://github.com/Jeffrey-Ede/adaptive-scans