CopilotJ brings conversational AI workflows to ImageJ bioimage analysis
Published in Artificial Intelligence, Journals, Scanners and Imaging, Tools.
Published October 31, 2026.
CopilotJ connects natural-language requests with ImageJ bioimage analysis, according to a correspondence published October 28 in Nature Methods.
The system uses several cooperating AI agents to interpret tasks, retrieve relevant knowledge, generate instructions and operate imaging software. A dedicated bridge exchanges information with ImageJ and executes analysis steps. After validation on a representative example, a pipeline can be saved and reused for batch processing.
The authors demonstrate tasks ranging from basic operations to multistage workflows. Tests on synthetic microscopy images assessed cell counting against known ground truth. Across the language models evaluated, repeated-task completion stability ranged from 84% to 93%, with each task repeated 15 times.
That stability measure should not be interpreted as universal analytical accuracy. Users still need to check segmentation, measurements and workflow suitability for their own images.
The software is released under Apache-2.0. Source code and an ImageJ plug-in are available.
Source: Lu, Z., Yuan, Z., Chen, Y. et al., Nature Methods, October 28, 2026. Read the original correspondence.