Semi-automatic volume measurement for orbital fat and total extraocular muscles based on Cube FSE-flex sequence in patients with thyroid-associated ophthalmopathy

Semi-automatic magnetic resonance imaging based orbital fat volumetry: reliability and correlation with computed tomography

“…As this limits the causal role of orbital diseases, other reasons for measurement inaccuracies could be related to the MRI protocol or segmentation method. The accuracy of volumetric measurement depends on an optimal protocol for the correct identification of intensity difference between the region of interest and its surrounding tissues 16 . Many MRI orbital protocols have been introduced for the volumetric measurement of orbital tissues.…”

“…In this approach, the regions of interest have to be defined on every single slice, making it very time demanding and impractical 31 . Due to the irregular morphology of the orbital fat, this method is associated with high inter-observer variability and is prone to bias according to the plane orientation of the scan 16 . Other studies address this problem by segmenting all but the orbital fat tissue (EOM, globe and optic nerve) and subtract these structures (which are easier to trace) from the whole orbital volume 2,15,32 .…”

“…The resulting volume, however, will also contain other tissues such as the lacrimal gland, vessels and connective tissue, which are difficult to distinguish from the fat, causing an overestimation of the true fat volume. More advanced methods such as regional growing and multidimensional threshold algorithms try to overcome these problems through a semi-automatic segmentation process but require more dedicated MRI protocols 9,16,31 . A standard MRI protocol was selected for this study to facilitate availability and make retrospective analysis more feasible.…”

“…Magnetic resonance imaging (MRI) provides excellent soft-tissue contrast, with better delineation of the orbital tissues and MRI can yield images in any plane. With the increasing use of MRI for evaluating GO, various protocols are described for safe and accurate measurement of adipose tissue with MRI 7,9,[14][15][16] . Initially, only determination of two-dimensional measurements was feasible, but technological advances and softwareassisted volumetric measurement facilitated the process for three-dimensional analysis 7,9,17 .…”

“…Different segmentation techniques have been developed to define the tissue borders and allow quantitative analysis. Due to the irregular morphology of orbital fat, the manual segmentation (i. e. tracing the tissue on each consecutive scan) is time-consuming and associated with high inter-operator variation 16,18 . Semi-automatic segmentation techniques try to address these issues, as the software will define the tissue outlines depending on reference tissue intensity.…”