Statistical analysis of multi-b factor diffusion weighted images can help distinguish between vasogenic and tumor-infiltrated edema.

ABSTRACT

PURPOSE: The diffusion model can be transformed into a multicompartment model by means of multi-b factor diffusion-weighted sequences. We adapted a method of statistical analysis of these images and evaluated its performance to distinguish tumor-infiltrated edema from vasogenic edema. MATERIALS AND METHODS: Forty-nine patients with infiltrating tumors (38 patients low to high-grade gliomas) or vasogenic edema (11 patients metastases, abscess, extra-axial lesions) were studied by multi-b factor diffusion-weighted imaging. Comparison of histological results and morphological and perfusion MRI defined 69 characteristic volumes of interest in the peritumoral edema of 69 distinct infiltrating lesions (40) or lesions inducing vasogenic edema (29). RESULTS: The factorial analysis had a sensitivity of 92.9% and a specificity of 90.6% between tumor-infiltrated and vasogenic edema. Simplified interpretation confined to values of the high and mean diffusivity compartments had a sensitivity of 87.5% and a specificity of 89.2% between strictly tumor-infiltrated edema and vasogenic edema with the advantage of simplified interpretation based on two-color parametric mapping. CONCLUSION: Discrimination between tumor-infiltrated edema and vasogenic edema can be achieved by means of a 90-s multi-b factor diffusion-weighted sequence and factorial analysis. Simplified visual and quantitative interpretation of the results should also allow integration of multi-b factor analysis into routine neuroradiology practice.