Keywords:
Outcomes, Haemorrhage, Outcomes analysis, Observer performance, CT, Neuroradiology brain, Interventional vascular, Emergency
Authors:
C. Pineda, F. X. ZARCO, A. Lopez Rueda, M. Benegas Urteaga, D. Campodonico, C. Ruiz, M. L. Olondo, S. Capurro, L. Oleaga Zufiría, S. Amaro; Barcelona/ES
DOI:
10.1594/ecr2013/C-1656
Results
The dichotomization of each observer is presented in the Table 3 .
The interobserver agreement obtained was excellent between experienced neuroradiologists in both conventional CT and dual-energy CT (k = 0,83 and k =1 respectively).
We obtain a poor correlation comparing the readings of conventional TC between the experienced neuroradiologists and the second-year resident; k = 0,41 (NR1 vs R2) and k = 0,31 (NR2 vs R2).
However,
we found a good correlation comparing dual-energy CT readings between the experienced neuroradiologists and the second-year resident; k = 0,68 (NR1 vs R2) and k = 0,68 (NR2 vs R2).
The comparison between the kappa index is shown in the Table 4 .
The improvement of the R2 k-index is secondary to decrease in detection of false bleeding; in this situation,
misclassified patients would not benefit from anticoagulant therapy.
The two remaining cases misclassified in dual-energy CT were false bleeding too,
so all the cases represents "false positive" of bleeding.
However,
if the fault was on the contrary,
the failure to detect a hemorrhage could expose a bleeding patient to anticoagulation therapy.
The mistake and the risk would not be acceptable.