Vessel-Masked Perfusion Magnetic Resonance Imaging With Histogram Analysis Improves Diagnostic Accuracy for the Grading of Glioma.

OBJECTIVE: Dynamic susceptibility contrast magnetic resonance imaging is widely used to assess glioma grade; histogram analyses are used for precise tumor perfusion evaluations. We evaluated the effect of vessel contamination in normalized cerebral blood volume (nCBV) to differentiate high- and low-grade gliomas.

METHODS: Thirty-four patients with gliomas underwent dynamic susceptibility contrast magnetic resonance imaging. Both traditional and vessel-masked nCBV maps were constructed. Histogram analyses of whole tumors and statistical comparisons were performed to compare traditional and vessel-masked images.

RESULTS: Mean values of all the histogram metrics were lower in vessel-masked images than in traditional images. Receiver operating characteristic curve analyses for every histogram metric showed a higher area under the curve for vessel-masked images than for traditional images. The integrated discrimination improvement showed that the vessel-masked images were superior to the traditional images significantly for predicting the glioma grading.

CONCLUSIONS: Vessel-masked nCBV maps can prevent overestimations of CBV measurements and can improve diagnostic accuracy for glioma grading.