109. Adoption of RadLex playbook to manage CT dosimetric big data starting from a multicenter Italian project

Abstract

Purpose To manage and analyse dosimetric data provided by 12 different CT scanners, installed in four Italian hospitals. Methods The interest in CT dose management has been growing during last years because of the increasing awareness of risks related to high radiation dose exams. A task of PREP, a Research Project of Regione Lombardia, is to create a database of CT dosimetric data using the NEXO[DOSE]® (Bracco, Italy) system, to analyse exposure values of the most common CT exams. Until today, about 450.000 exams from four different hospitals were collected in a cloud-based database. Results For each exam, NEXO[DOSE]® reports patients’ demographic information and scan protocol parameters, as total DLP and single series exposure values. When exams from different devices/PACS systems have to be compared, it is necessary to cluster them in homogenous sets according to scan region and acquisition task. This standardization is not always straightforward since each facility uses many ways to name the same procedure. For each single device we selected the DICOM tag, which was more appropriate to describe the procedure, and associated this description to a label from RadLex playbook, a standard system for naming radiology procedures written by RSNA. We created almost 90 sets of exams, including more than 400 procedure names. Conclusions 94% of all the CT exams are included in 9 labels and homogenous data may be used for risk assessment. Preliminary results show that single-phase head CT is the most common exam (40%) followed by multiphase chest-abdomen-pelvis (17.5%), multiphase abdomen-pelvis (9.3%) and single-phase chest (9.6%). This homogeneity will be a preliminary step to approach multi center optimization process.