Accurate MRI Segmentation of Skull Base Meningiomas

Objective: Skull base meningiomas often have an irregular shape depending on their location. The volume calculation of these lesions can be inaccurate if approximated to a spheroid shape, which is commonly done. A more precise volume calculation should be mandatory to accurately estimate tumor growth and/or resection rate. We tested the accuracy of a semiautomatic software for volume segmentation of intracranial skull base meningiomas.

Methods: Tumor volumes were calculated on magnetic resonance images (T1+Gd sequences) in 60 patients with single intracranial skull base meningiomas using commercial lesion management software (Vue PACS Livewire, Carestream, Rochester, New York, United States). A comparison between manual volume MRI-segmentation (used as reference), the semiautomatic software, and spheroid approximation was performed.

Results: Good correlation between manual calculation and semiautomatic technique was demonstrated regardless of the size and shape of the meningioma. Comparison between the semiautomatic segmentation and the spheroidal approximation showed very variable results with a wide range of difference (from −68% to + 74%). Stratified by location, the anterior skull base region, sphenoidal and cavernous sinus meningiomas showed the largest difference between the two techniques.

Conclusion: The use of semiautomatic segmentation of volume calculation of skull base meningiomas is more accurate than the commonly used spheroid approximation and thus is more appropriate for the assessment of tumor growth and/or resection rate.