pymia

pymia provides preprocessing, integration, and evaluation of 2-D, 2.5-D, and 3-D medical imaging data within deep learning workflows to enable combined analysis of imaging and non-image clinical data across multiple computational frameworks.

Key Features:

• Flexible Data Handling: Processes 2-D, 2.5-D, and 3-D medical images in full- and patch-wise modes and integrates non-image data such as demographics and clinical reports into deep learning pipelines.
• Framework Independence: Operates independently of specific deep learning libraries and integrates with TensorFlow and PyTorch.
• Advanced Evaluation Capabilities: Computes stand-alone performance metrics and monitors training using domain-specific metrics for segmentation, reconstruction, and regression.

Scientific Applications:

• Medical Image Analysis: Supports segmentation, reconstruction, and regression tasks, including evaluation of complex 3-D medical imaging data.

Methodology:

Implements modular data processing and evaluation components to standardize medical image analysis workflows, reduce custom implementation requirements, and enable consistent performance assessment across heterogeneous datasets.