Efficacy of fast screening MR in children and adolescents with suspected intracranial tumors.

PURPOSE Our purpose was to determine the sensitivity, specificity, and receiver operator characteristic (ROC) curve of a fast screening MR protocol in children and adolescents with suspected intracranial tumors. METHODS One hundred forty-one patients (mean age, 9.7 years; range, 2 months to 23.5 years) with suspected brain tumor were entered in a case-control study. Eighty-seven patients had intracranial tumors (31 suprasellar/hypothalamic, 27 supratentorial, 26 infratentorial, and three pineal) and 54 patients in the control group had other disorders. Two neuroradiologists reviewed blindly a detailed three-sequence conventional protocol (acquisition time, 8 minutes 27 seconds) and a two-sequence fast screening MR protocol (acquisition time, 4 minutes 44 seconds). RESULTS Sensitivity and specificity of the fast screening protocol for intracranial tumors was 100% and 92.6%, respectively. The areas under the ROC curves were 0.966 for the fast screening and 0.980 for the conventional MR protocol. No diagnostic performance difference was found between the ROC curves using the Az index. A kappa statistic of .93 for both examinations indicated excellent interobserver agreement. Additional MR sequences and other neuroimaging studies were not deemed necessary to exclude the presence of an intracranial tumor. CONCLUSION A fast dual-plane brain MR protocol may be adequate to screen children and adolescents thought to have an intracranial tumor. The less than 5 minute acquisition time allows a complete examination (including preparation) to be performed in 10 to 15 minutes. Future studies are recommended before this time-efficient neuroimaging examination is incorporated into clinical practice.