42 Chapter 3 N= % Estimated number of rectal cancer cases/MRIs reviewed on yearly basis Median (range) 35 (7-300) < 50 24 57% 50-100 13 31% > 100 5 12% Years of experience in reading rectal MRI Median (range) 3 (0.5-20) < 1 year 3 7% 1-5 years 26 62% >5 years 13 31% Effect of virtual training on diagnostic performance and interobserver agreement Table 2 shows the average diagnostic accuracy for the 42 course participants, and their interobserver agreement. Though results were largely not statistically significant in our small number of test cases, the majority of variables showed a tendency towards increased diagnostic performance when comparing the pre-course and post-course results. The increase in accuracy ranged between +2% and +17%, with the most considerable improvement for assessment of EMVI where mean accuracy improved from 74% to 91%. When looking at the continuous variables (tumour length, tumour height, number of lymph nodes), the difference between the participants and expert reference became smaller for the post-course versus pre-course test, indicating an improved concordance. A similar trend was observed for IOA, which improved for all variables when comparing the pre- and post-course results, except for the assessment of tumour deposits. Improvement in IOA was most evident for morphology (composition), tumour length, distance from the anorectal junction, number of suspicious mesorectal nodes and EMVI. For these variables, there was no overlap in 95% confidence intervals between the IOA results pre- and post-test, indicating a significant positive effect. Impact of participants’ experience level When comparing the baseline (pre-course) performance of the participants with <5 versus ≥5 years of prior experience in reading rectal MRI, we observed an overall trend towards better concordance with the expert-reference for the more experienced participants (except for T-stage). Results were largely non-significant, except for assessment of tumour length, total number of visible mesorectal nodes, and morphology/composition. For 11/15 staging variables the more experienced readers showed less benefit then the less-expert participants in terms of diagnostic accuracy when comparing the pre- and post-course results, though differences did not reach statistical significance in the vast majority of cases. Table 1 Continued
RkJQdWJsaXNoZXIy MjY0ODMw