This poster is published under an
open license. Please read the
disclaimer for further details.
Keywords:
Professional issues, Management, Genital / Reproductive system male, MR, Audit and standards, Diagnostic procedure, Health policy and practice, Quality assurance
Authors:
A. G. Wibmer, H. A. Vargas, T. Atkinson, D. M. Panicek, C. Moskowitz, J. Zheng, J. A. Eastham, H. Hricak; New York City, NY/US
DOI:
10.1594/ecr2014/C-0263
Results
Before the implementation of the certainty lexicon, the reporting radiologists used 49 different terms to express their levels of diagnostic certainty. In order to allow reasonable analysis,
these expressions were condensed to 20 groups (Figure 1).
Fig. 1: List of expressions that were used by the reporting radiologists before the implementation of a certainty lexicon.
After implementation of the lexicon, the readers adhered to the certainty lexicon in 85% of the cases.
In 15%,
however, the radiologists solely described the tumor’s relation to the prostatic capsule without explicetly using a lexicon term.
The level of diagnostoc certainty correlated to the prevalence of ECE on whole-mount histopathology (Figure 2).
Fig. 2: Expressions of diagnostic certainty for the presence of ECE on pre-prostatectomy prostate MRI before (pre-lexicon period) and after (lexicon period) the implementation of a lexicon of diagnostic certainty. For every term, the prevalence of ECE on whole-mount histopathology is given.
The diagnostic accuracy of the certainty lexicon was evaluated by receiver operater characetristic curve calculation. The area under the corresponding curve was 0.848 (95% confidence interval: 0.790 – 0.906) (Figure 3).
Fig. 3: This ROC-curve displays the diagnostic accuracy of the certainty lexicon. Its area ander the curve (AUC) is 0.848.