RESUMO
PURPOSE: To quantitatively investigate the feasibility of MRI as a tool for assessing the spatial distribution of a convectively delivered agent using a canine prostate model. MATERIALS AND METHODS: Canine prostates (ex vivo, n = 3; in vivo, n = 12) were injected under several injection paradigms with a solution of gadolinium-DTPA for MR contrast and methylene blue as a grossly visible surrogate drug marker. Ex vivo and in vivo distributions were assessed at 1.5T and quantitatively compared. RESULTS: Measured distributions using MRI and methylene blue pathology photographs were analyzed using a Bland-Altman method. The fractional percentage volume covered (V frac) compared the measurements grossly: Pearson's correlation coefficients were R = 0.99 for ex vivo and R = 0.77 for in vivo (P < 0.05). The fractional percentage of area covered (A frac) demonstrated the high degree of spatial correlation between individual slices: R = 0.93 for ex vivo and R = 0.98 for in vivo (P < 0.05). There was no statistically observable bias in scale or offset between the measurements. CONCLUSION: Measured distributions using MRI and pathology were highly correlated and unbiased, indicating the potential of MRI as a tool for quantitative assessment of interstitial delivery of injected therapies in vivo.