Pediatric Radiology | 2021
Is orbital susceptibility-weighted imaging superior to standard brain gradient echo sequences for detecting retinal hemorrhages?
Abstract
Dear Editors, We would like to thank the authors Bhatia et al. [1] for the review concerning orbital susceptibility-weighted imaging in abusive head trauma. The article points out the relevance of identifying retinal hemorrhage because of its association with abusive head trauma. Abusive head trauma is the leading cause of fatal head injuries, and up to 30% of abused children can go unrecognized [2, 3]. Therefore, it is imperative to have an accurate method to identify abusive head trauma and prevent further aggressions. Ophthalmological cohorts report an incidence of 51–100% of retinal hemorrhages are detected by fundoscopy in abusive head trauma [4]. Furthermore, the combination of subdural hemorrhage and severe retinal hemorrhage in the absence of scalp swelling is highly specific for abusive head trauma, despite its low sensitivity [5]. Thus, the search for retinal hemorrhage is pivotal in suspected abusive head trauma. The gold standard for the diagnosis of retinal hemorrhage is dilated fundus exam [1]. In suspected abusive head trauma, fundoscopy should be performed 24–48 h from hospital admission in order to guarantee optimal accuracy [6]; however, as well put by Bhatia et al. [1], fundoscopy is not always possible in this short time window because of the clinical status or lack of availability of pediatric ophthalmologists. Since the detection of retinal hemorrhage in gradient echo images by Altinok et al. [7], the application of MRI in the detection of ocular signs of abusive head trauma has been researched [6, 8, 9]. Zuccoli et al. [8] compared the accuracy of standard brain susceptibility-weighted imaging (SWI) with slice thickness of 3 mm and a high-resolution orbital SWI sequence with slice thickness of 1–1.4 mm. The authors were able to detect 13 of 21 retinal hemorrhages with the standard brain SWI and 12 of 15 retinal hemorrhages with high-resolution orbital SWI (sensitivity of 62% and 80% for whole-brain and orbit-focused SWI, respectively, and not 75% and 83%, as stated by Bhatia et al. [1]). Although not performed, a hypothesis test can be applied with the available data: because of the paired nature of the data (different methods applied to the same patient), the McNemar test must be applied. In the sample, 10 patients had retinal hemorrhage characterized by both methods, 2 cases were detected only in the orbit-focused SWI, and 3 cases were false-negatives in both protocols, resulting in a P-value of 0.5. Therefore, it cannot be said that the high-resolution orbital SWI sequence is superior to the standard brain SWI. Other authors have investigated the sensitivity of brain gradient echo sequences for detecting retinal hemorrhage in head trauma. Beavers et al. [9], using a T2* sequence (1–5-mm slice thickness), and Thamburaj et al. [6], using a thin-slice SWI sequence (1.5-mm slice thickness), identified 50–61% sensitivity, similar to the whole-brain SWI protocol by Zuccoli et al. [8]. Therefore, there are still not enough data to demonstrate a superior accuracy of orbit-focused SWI in relation to standard brain SWI.