highdicom

highdicom provides a high-level Python API to create, parse, encode, and decode DICOM objects for integrating machine learning workflows with clinical imaging data (pathology and radiology), supporting NumPy array representations and adherence to the DICOM standard.

Key Features:

• High-Level API: Provides high-level abstractions for encoding and decoding image-derived information in DICOM objects.
• Interoperability: Simplifies creation, parsing, and handling of DICOM-compliant files to facilitate integration between ML systems and enterprise medical imaging systems.
• Efficient Data Representation: Uses NumPy arrays for efficient representation of image pixel data and derived information.
• DICOM Compliance: Produces DICOM-conformant objects and metadata to maintain standard-compliant medical imaging data.

Scientific Applications:

• Image-based diagnostics: Supports machine learning workflows applied to pathology and radiology image-based diagnostics.
• Model training and evaluation: Enables training and evaluation of machine learning models using real-world DICOM imaging datasets.
• Imaging modalities: Applicable to slide microscopy and computed tomography (CT) imaging modalities.

Methodology:

Implements a high-level Python interface that abstracts low-level DICOM details and provides functions to create, parse, encode, and decode DICOM objects while using NumPy arrays for pixel and derived data representation to produce DICOM-compliant files and metadata.