Magnetic resonance imaging: a valid imaging alternative for assessment of orbital fractures

Abstract

INTRODUCTION:Orbital fractures are a common form of facial trauma often accompanied by serious morbidity manifesting as both functional visual disturbance and oculo-facial deformity. Typically, their diagnosis is confirmed with standard multidetector computed tomographic (CT) scans. These scans also have proven utility in the fabrication of biomodels to assist surgical reconstruction of the resultant bony defects.The major drawback of CT is the associated ionizing radiation dose. Paediatric patients and patients receiving multiple CT scans within a short window of time are particularly vulnerable to the effects of ionizing radiation which include most notably an increased risk of malignancy as well as cataract formation.Magnetic resonance imaging (MRI) has emerged as a proposed alternative to CT with the major advantage of no associated ionizing radiation. Its efficacy in imaging the intact bony orbit has previously been demonstrated via rendering virtual 3D biomodels of both CT and MRI and comparing them. No previous research of this nature has examined the traumatised bony orbit. Additionally, no previous research has demonstrated the ability to 3D print accurate biomodels of the bony orbit from MRI.AIM:The aim of this research is to investigate the efficacy of MRI as compared to CT firstly for imaging the traumatised bony orbit for the purpose of diagnosis and secondarily for fabrication of orbital biomodels.METHODS:A retrospective case series design was used, utilizing data obtained from a previous study. The data consisted of eleven patients with suspected unilateral bony orbital trauma who received a CT and MRI scan.DICOM data was imported to AMIRA 3D visualization software where 3D models were manually generated and the fracture boundaries manually defined. These 3D models were imported into Rapidform reverse engineering software where they were aligned and the fractures compared for spatial congruency (deviation of borders and overall size as projected to a 2D plane).Additionally, 3D models were physically printed in ABS plastic and Synthes 0.7mm orbital fan plates were shaped to fit CT and MRI models by a single Maxillofacial Surgeon. A single set of plates was bent to the MRI models and two sets of plates were bent to the CT models to quantify operator error. These plates were optically scanned and analysed using Rapidform software for spatial congruency (surface and border deviation).RESULTS:Of the eleven participants, two did not demonstrate orbital fractures on scans and were excluded. Three participants had ZMC fractures with orbital involvement. Five participants had single unilateral orbital blowout fractures and one participant had two, ipsilateral orbital blowout fractures.For the analysis of fractures, a decision was made to analyse each fracture individually giving a total of ten fractures available due to two fractures in one participant. A high degree of congruence was demonstrated between orbital blowout fractures (n=7) seen on CT and MRI (average deviation between fracture bounds of 0.85mm). This was not reproduced for ZMC fractures (n= 3, average deviation between fracture bounds of 2.14mm).For physical plate bending, given the inaccuracies of ZMC fractures these were excluded. Additionally, it was not practical to treat the two ipsilateral fractures separately. Thus, six orbits were available. Marginally greater surface deviation was seen between MRI and the two CT groups (0.50 and 0.54mm) than between the CT groups (0.38mm) however a Wilcoxon signed ranks test did not reveal any significant difference.Operator time throughout the study was significant with manual segmentation of orbits from MRI taking approximately 8 hours on average (as opposed to one hour for CT).CONCLUSION:The results of this study suggest MRI can define orbital bony trauma with clinically acceptable accuracy. Its use in clinical settings will depend on its increasing availability and affordability. Additionally, the issue of segmentation time will need to be addressed. Further developments in the field such as 7T MRI and the application of machine learning may provide a way forward.