Abstract

The efficacy of proton magnetic resonance imaging (MRI) was evaluated and compared with that of B-scan ultrasound in the detection and differentiation of diabetic vitreous hemorrhage. Although conventional spin-echo MRI could not locate vitreous hemorrhages, gradient-recalled-echo (GRE) MRI readily did so. The aberrant signals appeared to originate from the interfacing between hemorrhages and the vitreous, and possibly also from the paramagnetic effect of the ferrous ion. The information provided by boundary/susceptibility detection, unique to the GRE sequence, is useful in delineating the extent of vitreous hemorrhage and hemolysis. However, for the diagnosis and follow up of diabetic vitreous hemorrhages, MRI appears no more informative than B-scan ultrasonography.