RLCorrector: Reinforced Proofreading for Cell-level Microscopy Image Segmentation
This addresses the need for efficient error correction in microscopy image segmentation for connectomics research, representing an incremental improvement over existing semi-automatic proofreading approaches.
The paper tackles the problem of error-prone segmentation in electron microscopy images for connectomics by proposing RLCorrector, a fully automatic proofreading method using reinforcement learning, which outperforms conventional methods in various tests.
Segmentation of nanoscale electron microscopy (EM) images is crucial but still challenging in connectomics research. One reason for this is that none of the existing segmentation methods are error-free, so they require proofreading, which is typically implemented as an interactive, semi-automatic process via manual intervention. Herein, we propose a fully automatic proofreading method based on reinforcement learning that mimics the human decision process of detection, classification, and correction of segmentation errors. We systematically design the proposed system by combining multiple reinforcement learning agents in a hierarchical manner, where each agent focuses only on a specific task while preserving dependency between agents. Furthermore, we demonstrate that the episodic task setting of reinforcement learning can efficiently manage a combination of merge and split errors concurrently presented in the input. We demonstrate the efficacy of the proposed system by comparing it with conventional proofreading methods over various testing cases.