Learning effect on perinatal post-mortem magnetic resonance imaging reporting: single reporter diagnostic accuracy of 200 cases

Abstract

OBJECTIVE: The objective of the study is to compare diagnostic accuracy of perinatal post-mortem magnetic resonance (PMMR) imaging against conventional autopsy, when reported by a single-blinded observer for all organ systems following a period of initial experience. METHODS: We compared pre-autopsy PMMR with conventional autopsy for the detection of (1) major pathological abnormalities related to the cause of death and (2) all diagnostic findings in five different body organ systems. PMMR was reported blinded to autopsy findings. RESULTS: In 201 cases, 123/146 (84.2%) of major abnormalities were identified by PMMR. Overall diagnostic accuracy of PMMR was 89.6% [95% confidence interval (CI): 84.3, 93.2%] across all cases, with high concordance 91.8% (95% CI: 89.9, 93.4%) across most organ systems. Our study showed higher concordance than single reporter statistics previously reported in neurological [92.2% vs 73.8%; diff 18.4% (95% CI: 11.0, 25.4%) p < 0.01] and thoracic systems [93.7% vs 81.2%; diff 12.5% (95% CI: 6.3, 18.4%) p < 0.01] and slightly better overall [91.8% vs 87.1%; diff 4.7% (95% CI: 2.1, 7.3%) p < 0.01]. CONCLUSION: The PMMR examinations can be reliably reported by a single radiologist, following a period of experience and training with this specific modality, with high-diagnostic accuracy for all organ systems. © 2017 John Wiley & Sons, Ltd.