ABSTRACT
INTRODUCTION: The purpose of our study was to test the accuracy and applicability of decision rules utilizing apparent diffusion coefficient (ADC) ratios on accurate preoperative diagnosis of common pediatric cerebellar tumors across two institutions. METHODS: In this HIPAA-compliant, IRB-approved study, performed at two institutions, 140 pediatric cerebellar tumors were included. Two separate reviewers placed regions of interest on the solid components of 140 tumors (98 at site A and 42 at site B) and normal brain on the ADC maps. The third reviewer who was blinded to the histopathological diagnoses made the same measurements on 140 patients to validate the data. Tumor to normal brain ADC ratios were calculated. Receiver operator curve (ROC) analysis was performed to generate thresholds to discriminate tumors. Utility of decision rules based on these thresholds was tested. RESULTS: While ADC values of medulloblastomas were different between the sites, there was no difference among the ADC ratios of medulloblastomas, pilocytic astrocytomas, ependymomas, and atypical teratoid rhabdoid tumors between the sites. ADC ratio of ≥1.8 correctly discriminated pilocytic astrocytomas from ependymomas with a sensitivity of 0.83 and a specificity of 0.78. ADC ratio of <1.2 correctly discriminated ependymomas from embryonal tumors with a sensitivity of 0.87 and a specificity of 0.83. The proposed decision rules correctly discriminated 120 of the 140 tumors (85.71%). Age ≥2 years criterion correctly sorted medulloblastomas in 84.48% of patients and age <2 years correctly distinguished atypical teratoid rhabdoid tumors in 90.00% of patients with embryonal tumors. CONCLUSIONS: Decision rules based on ADC ratios are applicable across two institutions in the accurate preoperative diagnosis of common pediatric cerebellar tumors.
