In this work we used data from the following information systems:
- RISCa. Cantabrian Radiology Information System (Bahía Software company).
- DISCa. Cantabrian Dosimetric Information System, which is composed of Bayer's dose management system called “Radimetrics” and QlickTech’s business intelligence tool called QlikView.
150000 DISCA descriptions were analyzed, corresponding to studies of the first half of 2019.
These descriptions were compared with their corresponding RISCA peers, and the percentage of wrong or incomplete descriptions and their cause were determined.
Approximately 40% of the descriptions were wrong or incomplete due to problems in the DICOM workflow (heterogeneity and incompleteness problems).
Fist, in order to solve these problems, it was proposed to use in “Radimetrics” the HL7 ORU^R01 messages that are sent from the RIS to the PACS to ask if the study has been stored. These messages contain the study's accession number and the code and description of the SERAM catalog.
This solution solves the heterogeneity problem: all descriptions would correspond to the SERAM catalog. However, they are not always complete because some descriptions correspond to a part of the study, not to the whole study (incompleteness problem).
Finally, this second problem is solved through an algorithm implemented in "QlikView" that merges the information from RISCa before sending it to DISCa using, again, HL7 ORU^R01 messages.
This algorithm reduces the percentage of wrong or incomplete descriptions from 40% to 10%.
This solution solves the problem of heterogeneity: all descriptions correspond to the SERAM catalog. However, they are not always complete because some descriptions are still named only after a part of the study (incompleteness problem).
Finally, this second problem is solved through an algorithm implemented in a Data Mining tool ("QlikView") that consolidates the information from the RISCa before sending it to DISCa using, again, HL7 ORU^R01 messages.
This algorithm reduces the percentage of wrong descriptions from 40% to 10%.