DeepImageJ

DeepImageJ integrates pre-trained deep learning (DL) models into ImageJ to enable biomedical image analysis tasks such as pixel and object classification, instance segmentation, denoising, and virtual staining.

Key Features:

• Access to Pre-trained Models: Provides access to the BioImage Model Zoo, a repository of pre-trained DL models curated for biomedical image analysis.
• Versatile Applications: Supports pixel classification, object classification, instance segmentation, denoising, and virtual staining for image-processing workflows.
• Compatibility and Extensibility: Includes utilities to incorporate new models and maintains compatibility with state-of-the-art solutions and model formats.
• Deployment via deepImageJ format: Enables deployment of models from several training frameworks into ImageJ using the deepImageJ format.

Scientific Applications:

• Cell biology: Enables segmentation, classification, and denoising of cellular images for quantitative analysis.
• Pathology: Facilitates instance segmentation and virtual staining of histological images for tissue analysis.
• Neuroscience: Supports segmentation and denoising workflows for neuronal imaging datasets.

Methodology:

Integrates and applies pre-trained DL models within ImageJ, accesses models from the BioImage Model Zoo and uses the deepImageJ format for model deployment; execution is supported on standard CPU-based computers without requiring GPUs.