CDeep3M

CDeep3M performs large-scale image segmentation using deep learning to analyze microscopy images from light, electron, and X-ray modalities.

Key Features:

• Image segmentation: Performs large-scale segmentation of microscopy images using deep learning techniques.
• Supported modalities: Processes images acquired by light microscopy, electron microscopy, and X-ray microscopy.
• Pre-trained models: Utilizes pre-trained neural network models available in the CIL-CDeep3M model zoo.
• Model evaluation (CDeep3M-Preview): Provides a preview mechanism to evaluate pre-trained models on user datasets or publicly hosted datasets.
• Deployments: Supports deployment on Google Colab, AWS, Docker, and Singularity.
• Model and data co-hosting: Co-hosts trained deep neural networks alongside microscopy images in a model zoo and a cell image library and accepts contributions of trained models and image datasets.

Scientific Applications:

• Biological image segmentation: Segments cells and other structures in light, electron, and X-ray microscopy images for quantitative analysis.
• Cross-dataset evaluation: Evaluates and benchmarks pre-trained neural network performance across user-provided and publicly hosted datasets.
• Reproducible model sharing: Enables sharing and reuse of trained neural networks and microscopy datasets to support reproducibility of image analysis.

Methodology:

Applies deep learning via pre-trained neural network models for image segmentation and supports evaluation of those models on user or public datasets through the CDeep3M-Preview mechanism.