Bridging the gap: Integrating cutting-edge techniques into biological imaging with deepImageJ

Area of Science:

• Life Sciences
• Bioimage Analysis
• Computational Biology

Background:

• Fiji/ImageJ is a widely used platform for life science image analysis.
• Deep learning (DL) offers powerful tools for complex image processing tasks.
• Integrating DL into existing bioimage analysis workflows can be challenging.

Purpose of the Study:

• To present advancements in deepImageJ, a Fiji/ImageJ plugin for bioimage analysis.
• To demonstrate the plugin's enhanced capabilities in deploying deep learning models.
• To showcase streamlined workflows for complex, 3D, and large-scale image analysis.

Main Methods:

• Integration of the Java Deep Learning Library for compatibility with TensorFlow, PyTorch, and ONNX.
• Demonstration of deepImageJ's ability to run multiple DL engines within Fiji/ImageJ.
• Application of deepImageJ in three case studies: image-to-image translation with nuclei segmentation, 3D nuclei segmentation, and large volume segmentation.

Main Results:

• DeepImageJ now supports diverse deep learning frameworks, enhancing its flexibility.
• The plugin successfully handles complex pipelines, 3D image analysis, and large image volumes.
• Case studies confirm the effectiveness of deepImageJ for integrated and advanced bioimage segmentation tasks.
• Compatibility with the BioImage Model Zoo is established.

Conclusions:

• Advancements in deepImageJ provide a more flexible and user-friendly framework for bioimage analysis.
• The plugin makes advanced deep learning more accessible and efficient for life science researchers.
• DeepImageJ is poised to enable next-generation image processing in the life sciences.