Apparent diffusion coefficient of human brain tumors at MR imaging.

PURPOSE To determine if apparent diffusion coefficient (ADC) can be used to differentiate brain tumors at magnetic resonance (MR) imaging. MATERIALS AND METHODS Institutional review board approval or informed patient consent was not required. MR images were reviewed retrospectively in 275 patients with brain tumors: 147 males and 128 females 1-81 years old, treated between September 1997 and July 2003. Regions of interest were placed manually in tumor regions on MR images, and ADC was calculated with a five-point regression method at b values of 0, 250, 500, 750, and 1000 sec/mm2. ADC values were average values in tumor. All brain tumor subgroups were analyzed. Logistic discriminant analysis was performed by using ADC, age, and patient sex as independent variables to discriminate among tumor groups. RESULTS A significant negative correlation existed between ADC and astrocytic tumors of World Health Organization grades 2-4 (grade 2 vs grades 3 and 4, accuracy of 91.3% [P < .01]; grade 3 vs 4, accuracy of 82.4% [P < .01]). ADC of dysembryoplastic neuroepithelial tumors (DNTs) was higher than that of astrocytic grade 2 tumors (accuracy, 100%) and other glioneuronal tumors. ADC of malignant lymphomas was lower than that of glioblastomas and metastatic tumors (accuracy, 83.6%; P < .01). ADC of primitive neuroectodermal tumors (PNETs) was lower than that of ependymomas (accuracy, 100%). ADC of meningiomas was lower than that of schwannomas (accuracy, 92.4%; P < .01). ADC of craniopharyngiomas was higher than that of pituitary adenomas (accuracy, 85.2%; P < .05). ADC of epidermoid tumors was lower than that of chordomas (accuracy, 100%). In meningiomas, ADC was not indicative of malignancy grade or histologic subtype. CONCLUSION ADC is useful for differentiation of some human brain tumors, particularly DNT, malignant lymphomas versus glioblastomas and metastatic tumors, and ependymomas versus PNETs.