DICOM re-encoding of volumetrically annotated Lung Imaging Database Consortium (LIDC) nodules

Area of Science:

• Medical Imaging
• Radiology
• Data Science

Background:

• The Lung Imaging Data Consortium and Image Database Resource Initiative (LIDC) dataset contains expert annotations for lung nodules from CT scans.
• These annotations were originally in a project-specific XML format, limiting direct use with standard tools.

Purpose of the Study:

• To convert LIDC-IDRI lung nodule annotations into standard DICOM objects.
• To enhance the findability, accessibility, interoperability, and reusability (FAIR) of the LIDC-IDRI data for researchers.

Main Methods:

• Utilized open-source tools to parse XML annotations and save them as DICOM Segmentation and Structured Reporting objects.
• Validated the resulting DICOM objects for standard compliance, data consistency, and interoperability with existing tools.

Main Results:

• Successfully created a standardized dataset of lung nodule annotations in DICOM format.
• The dataset includes annotations for 6859 nodules across 875 subjects, clustered into 2651 distinct nodules.
• Data is accessible via The Cancer Imaging Archive (TCIA).

Conclusions:

• The standardized DICOM dataset preserves the original LIDC-IDRI annotations.
• Facilitates the development of automated tools for lung lesion characterization and image phenotyping.
• Improves data integration with other standardized collections, promoting FAIR data principles.