Keywords:
Abdomen, Liver, Oncology, MR, CT, Observer performance, Staging, Cancer, Cirrhosis, Workforce
Authors:
M. Sini, M. Erta, G. Corrias, L. Saba; Cagliari/IT
DOI:
10.26044/ecr2019/C-0875
Results
Cohen's Kappa statistics showed a small intra reader agreement before and after that residents read and applied the LIRADS manual (Fig.
3 Table 1: k ranges from 0.289 to 0.5 for CT and from 0.136 to 0.65 for MR).
We found an increased inter-reader agreement in every pair of readers after that they applied the LIRADS manual,
for CT,
and every pair of readers but one,
for MR.
For residents with less than 3 years of experience,
analyzing CTs,
the k ranged from 0.049 to 0.263 before LIRADS application,
to 0.205 to 0.761 after LIRADS application (Fig.
4; Table 2; green rows).
For residents with more than 3 years of experience,
analyzing CTs,
the k ranged from 0.361 to 0.508 before LIRADS application,
to 0.652 to 0.936 after LIRADS application (Fig.
4; Table 2; yellow rows).
The only case of agreement going down from k 0.382 to k 0282,
was the inter-reader agreement of MRs of residents with less the 3 years of experience (Fig.
5 Table 3,
green rows).
For residents with more than 3 years of experience,
analyzing MRs,
the k ranged from 0.192 to 0.533 before LIRADS application,
to 0.521 to 1.000 after LIRADS application (Fig.
5; Table 2; yellow rows).
In figure 6 show one case of a biopsy proven HCC that was given as such by 3 out of 6 readers,
before LIRADS application and by 100% of readers after LIRADS application.