SliDL

SliDL provides Python utilities to apply deep learning to whole-slide images (WSIs) for analysis supporting disease detection, diagnosis, and monitoring.

Key Features:

• Pre- and Post-Processing Tools: Provides tools for handling WSI-specific preprocessing and postprocessing tasks including artifact handling and preparation of data for model training.
• Annotation Management: Supports management of slide annotations used for training and evaluating deep learning models on WSIs.
• Tile Extraction: Extracts smaller image tiles from large WSIs to enable patch-based model training and inference.
• Tissue Detection: Automates identification and segmentation of tissue regions within whole-slide images.
• Model Evaluation: Supplies utilities to assess performance of models trained on WSI data.
• PyTorch Integration: Integrates with PyTorch for model training and inference workflows.

Scientific Applications:

• Disease detection and diagnosis: Applies deep learning to WSIs to support early detection, diagnosis, and monitoring of diseases from histopathology slides.
• Computational pathology research: Enables development, training, and evaluation of deep learning models for WSI analysis and benchmarking.
• Workflow automation for pathology: Facilitates automation of tasks in pathology that rely on tile extraction, tissue detection, and model-based inference.

Methodology:

Includes tile extraction, tissue detection and segmentation, annotation management, artifact handling, and post-training model evaluation, with integration for PyTorch-based model training and inference.