Magnetic resonance imaging-based detection of retinal hemorrhages in a multicenter cohort of abusive head trauma

Abstract Background In pediatric abusive head trauma (AHT), retinal hemorrhages are a key diagnostic feature. Detection by fundoscopy may be delayed or limited, whereas magnetic resonance imaging (MRI) enables non-invasive, objective assessment on routine brain sequences. Objective To evaluate the diagnostic utility of different MRI sequences—particularly susceptibility-weighted imaging (SWI), T2*-weighted (T2*w), and morphological sequences—in detecting retinal hemorrhages in AHT. Materials and methods In this retrospective multicenter study (2006–2015), 57 well-documented AHT cases from three German institutions were analyzed. A subgroup consisted of “confession cases.” MRI scans were reviewed for retinal hemorrhages across SWI, T2*w, T1-weighted, T2-weighted (T2w), and fluid-attenuated inversion recovery (FLAIR) sequences by blinded expert readers. Fundoscopy results served as the gold standard. Sensitivities were calculated for each sequence, and “confession” versus “non-confession” cases were compared. Results Fundoscopy detected retinal hemorrhages in 44 of 56 evaluable cases (78.6%). MRI identified retinal hemorrhages most frequently on gradient recalled echo sequences, with SWI showing higher sensitivity compared to T2*w (76.9% vs. 47.8%). T2w imaging showed markedly lower sensitivity (30.3%) but detected retinal hemorrhages on one eye missed on T2*w imaging in two cases. In three cases, MRI detected retinal hemorrhages not reported on fundoscopy. No statistically significant differences were found between “confession” and “non-confession” cases across all parameters considered ( P >0.05). Conclusion MRI—particularly SWI and, to a lesser extent, T2*w imaging—may offer a useful tool of detecting retinal hemorrhages in AHT when fundoscopy is limited. T2w imaging may provide complementary information in selected cases. Graphical abstract