An Open Library of CT Patient Projection Data

Area of Science:

• Medical Imaging
• Computer Science
• Algorithm Development

Background:

• Limited access to patient CT projection data hinders the development and validation of novel reconstruction algorithms.
• Existing data formats may lack vendor neutrality or comprehensive information required for advanced research.

Purpose of the Study:

• To establish an open-access library of computed tomography (CT) patient projection data.
• To provide a vendor-neutral dataset in an extended DICOM format (DICOM-CT-PD) for algorithm research.
• To facilitate the development and validation of CT reconstruction algorithms.

Main Methods:

• Creation of a DICOM-CT-PD format, extending DICOM to include sinogram data, acquisition geometry, patient information, and pathology.
• Collection of diverse CT scan types (head, chest, abdomen, ECG-gated, dual-energy) with associated projection data, reconstructed images, and text files.
• Development of instructional materials, including a format dictionary, reader, and user manual, to support data utilization.

Main Results:

• A substantial collection of projection data has been gathered, with 328 head, 228 chest, and 228 abdomen cases collected.
• The library will offer projection data at various dose levels, reconstructed images, and radiologist performance metrics.
• The final library will feature a curated selection of scans from multiple manufacturers and advanced acquisition types.

Conclusions:

• The DICOM-CT-PD library addresses a critical need for accessible CT projection data.
• This open-access resource is expected to significantly accelerate innovation in CT reconstruction algorithm development.
• The library will be freely available to academic researchers, promoting collaborative advancements in medical imaging.