EXACT

EXACT provides collaborative, algorithm-aided annotation, plugin-based integration of deep learning, and dataset version control for reproducible analysis of image and related signal data.

Key Features:

• Collaborative Annotation: Supports multi-user annotation of image datasets, enabling contributions from distributed researchers to shared annotated image data.
• Algorithm-aided Annotation: Integrates algorithmic assistance for annotation and validation of image data, including workflows for identifying false annotations.
• Flexible Plugin System: Employs a plugin architecture enabling tasks such as counting mitotic figures using a screening mode, a validation view for detecting false annotations, and incorporation of advanced deep learning technologies for image analysis.
• Version Control: Maintains a version control system that tracks changes in datasets and links experimental results, including deep learning experiments, to specific dataset versions to support reproducibility.

Scientific Applications:

• Medical Imaging: Handles multi-gigapixel medical whole slide images for detailed analysis and annotation.
• Deep Learning Integration: Supports deep learning-based image analysis for tasks such as cytology scoring.
• Interdisciplinary Research: Facilitates multi-centre whole slide image tumor annotations and cross-disciplinary projects requiring shared annotated datasets.
• Specialized Applications: Has been applied to non-image signal analysis such as whale sound spectroscopy clustering.

Methodology:

Integrates algorithm-aided annotation and validation, a plugin architecture for task-specific algorithms, deep learning-based image analysis, a version control system that tracks dataset changes and links experiments to dataset versions, and synchronization with an EXACT Server via a REST API.