Improving the grading accuracy of astrocytic neoplasms noninvasively by combining timing information with cerebral blood flow: A multi-Ti arterial spin-labeling MR imaging study

Abstract

BACKGROUND AND PURPOSE: Systematic and accurate glioma grading has clinical significance.Wepresent the utility of multi-TI arterial spin-labeling imaging and provide the bolus arrival time maps for grading astrocytomas. MATERIALS AND METHODS: Forty-Three patients with astrocytomas (21 men; mean age, 51 years) were recruited. The classification abilities of conventional MR imaging features, normalized CBF value derived from multi-TI arterial spin-labeling imaging, normalized bolus arrival time, and normalized CBF derived from single-TI arterial spin-labeling were compared in patients with World Health Organization (WHO) grade II, III, and IV astrocytomas. RESULTS: The normalized CBF value derived from multi-TI arterial spin-labeling imaging was higher in patients with higher grade astrocytoma malignancies compared with patients with lower grade astrocytomas, while the normalized bolus arrival time showed the opposite tendency. The normalized CBF value derived from the multi-TI arterial spin-labeling imaging showed excellent performance with areas under the receiver operating characteristic curve of 0.813 (WHO II versus III), 0.964 (WHO II versus IV), 0.872 (WHO III versus IV), and 0.883 (low-grade-versus-high-grade gliomas). The normalized CBF value derived from single-TI arterial spin-labeling imaging could statistically differentiate the WHO II and IV groups (area under the receiver operating characteristic curve0.826). The normalized bolus arrival time effectively identified the WHO grades II and III with an area under the receiver operating characteristic curve of 0.836. Combining the normalized CBF value derived from multi-TI arterial spin-labeling imaging and normalized bolus arrival time improved the diagnostic accuracy from 65.10% to 72.10% compared with the normalized CBF value derived from multi-TI arterial spin-labeling imaging being applied independently. The combination of multi-TI arterial spin-labeling imaging and conventional MR imaging had the best performance, with a diagnostic accuracy of 81.40%. CONCLUSIONS: Multi-TI arterial spin-labeling imaging can evaluate perfusion dynamics by combining normalized bolus arrival time and normalized CBF values derived from multiple TIs. It is superior to single-TI arterial spin-labeling imaging and conventional MR imaging features when applied independently and can improve the diagnostic accuracy when combined with conventional MR imaging for grading astrocytomas.