Background: The objective of this study is to evaluate the sensitivity and specificity of DWI in differentiating brain abscesses from other intracranial cystic lesions.
Methods: One hundred fifteen patients with 147 cystic lesions (mean age, 26.4 year) were prospectively studied with DWI on a 1.5-T magnetic resonance imaging. Lesions appearing hyperintense on DWI with the ADC values of lower than 0.9 +/- 0.13 x 10(-3) mm(2)/s (mean +/- SD) were considered as brain abscess, whereas hypointense lesions on DWI with the ADC values 2.2 +/- 0.9 x 10(-3) mm(2)/s were categorized as nonabscess cystic lesions.
Results: Ninety-three of 97 brain abscess lesions were hyperintense on DWI, with significantly low (P = .0001) ADC value (0.87 +/- 0.05 x 10(-3) mm(2)/s) (mean +/- SEM), compared with 48 nonabscess lesions (2.89 +/- 0.05 x 10(-3) mm(2)/s). Four of 97 brain abscess lesions in 65 patients were false negative, and 2 of 50 nonabscess lesions in 50 patients were false positive for the diagnosis of brain abscess. The ADC value of the tumor cysts (2.9 +/- 0.05 x 10(-3) mm(2)/s) was significantly lower (P = .02) compared with benign cysts and neurocysticercosis (3.2 +/- 0.05 x 10(-3) mm(2)/s) among nonabscess group. The sensitivity of DWI for the differentiation of brain abscesses from nonabscesses was 96%; specificity, 96%; positive predictive value, 98%; negative predictive value, 92%; and accuracy of the test, 96%.
Conclusions: Diffusion-weighted imaging has high sensitivity and specificity for the differentiation of brain abscess from other nonabscess intracranial cystic lesions.