Better representations of scanner utilization from DICOM metadata
- Mentors
- Pradeeban, PUNEET SHARMA, Nishchal Singi
- Organization
- Department of Biomedical Informatics, Emory University
- Technologies
- python, dicom, Front-end
- Topics
- machine learning
Niffler enables computing scanner utilization using real-time DICOM metadata extraction. Niffler acquires images from the PACS in real-time (meta-extraction) and on-demand (cold-extraction), then extracts DICOM metadata into a Mongo database or a CSV file, and performs computations on the metadata to compute utilization metrics for the scanners. This project performs the computations for the MR scanners, although it can be used for any modality. However, these computations were largely limited to the study level - how frequently a scanner idled between studies and how long it took for a scanner to perform a given study. Computing those metrics in a finger granularity, at the series level, is more challenging since the start time and end time of a series is harder to find with just public DICOM headers.
Furthermore, while this project has a scanner utilization computed in the backend, there is no integrated front-end to present the results elegantly. The created results are currently stored in CSV files and displayed through an Eaglescope dashboard. The front-end can be improved with better integration.