Author: Jeremy Christophel, Zhihua Qi 👨🔬
Affiliation: Henry Ford Health 🌍
Purpose: To demonstrate a method to compare DICOM metadata from clinical scanners with institutional protocols as validation that clinical use matches the master protocol.
Methods: DICOM metadata is extracted from the RDSR and scan series archived in a Philips PACS server using DICOMweb Python scripts. The metadata is sorted into scan and reconstruction settings and then compared to the master protocol located in a SQL Server database. Any deviation between the scan and reconstruction settings with those in the master protocol are noted for further follow up.
Results: Nineteen pediatric routine head scans from a single GE LighSpeed VCT scanner located in the Emergency Room were compared to the master protocol. Out of eight separate scan settings in the master protocol, four had 100% agreement (kV, Detector Coverage, Rotation Time and Min mA) while four had disagreements (CTDI range, Max mA, Scan Mode and Noise Index). Differences between clinical images and the master protocol were mainly due to the use of the helical scan mode for some cases rather than the prescribed axial scan mode.
Conclusion: This is a simple method using open-source software to verify that clinical protocols match the prescribed settings of a master protocol. Some manual verification of data is necessary due to different labeling of scan series for various scanners and manufacturers. The future development of a large language model could facilitate an automated series name matching process for this approach.