Volume estimation of glioblastomas from MRI : comparison of volumetric segmentation and ellipsoid formula
MetadataShow full item record
Background: Tumour volume estimation of glioblastoma (GBM) is necessary for assessing the extent of disease, detecting growth, and for evaluating both neurosurgical and oncological treatment results. The ellipsoid formula offers a quick and simple method for volume estimation that is widely used. While semi-automatic segmentation presumably offer superior accuracy of measurements, it is rather time-consuming. We sought to explore agreement between pre-treatment GBM volumes estimated by semi-automatic segmentation and estimation based on the ellipsoid formula. Methods: GBM volumes in 165 consecutive patients were estimated from pre-treatment contrast enhanced T1-weighted 1.5 or 3 Tesla MRIs by semi-automatic volume segmentation (using 3D Slicer version 4.5.0). Volumes were subsequently compared to volumes obtained by the ellipsoid formula. Results: GBM volume estimation based on the ellipsoid formula had good agreement with semi-automatic segmentation on a group level (Cronbach’s alpha of 0.97, indicating a near perfect agreement). However, on an individual level, we found that large differences in volume measurement by the two methods were common. In tumours located in the midline the volume difference between the two methods was significantly greater than in non-midline tumours. Conclusions: Since agreement between these methods was good on a group level, comparisons across studies or meta-analyses of studies using these two techniques is presumably feasible. It also implies that the rather quick and simple ellipsoid formula may be sufficient for studies with a large number of participants, especially if tumour volume or derivatives such as extent of resection is not the primary endpoint. However estimating tumour volume by use of the ellipsoid formula may be too inaccurate where precise tumour volume assessment on an individual level is required, e.g. assessment of growth or response to treatment.