Identifying tumor vascular permeability heterogeneity using reduced encoding techniques.

摘要

We test the hypothesis that the loss of spatial resolution to gain temporal resolution in clinical dynamic contrast enhanced (DCE) magnetic resonance mammography (MRM) causes partial volume effects that yield inaccurate permeability-surface area products (PS = K_p↔t) which results in erroneous diagnostic information and we offer a potential solution using reduced encoding techniques to solve this problem. We compared the PS obtained from DCE MRI at clinical MRI resolutions (2500 x 2500 μm resolution), to that obtained from resolutions analogous to histopathological in plane resolutions (938 x 938 μm and 469 x 469 μm resolution). Secondly, we determined the accuracy of PS obtained from Keyhole, R&barbelow;educed-encoding I&barbelow;maging by G&barbelow;eneralized-series R&barbelow;econstruction (RIGR), and T&barbelow;wo-reference RIGR (TRIGR) using high-resolution baseline data (469 x 469 μm resolution) and clinical resolution dynamic data (2500 x 2500 μm resolution). Lastly, we statistically correlated two-compartment model fitting parameters (tumor EES volume fraction, v_e, tumor plasma volume fraction, v_p, and PS) obtained from DCE MRI at all three resolutions to histopathologically determined tumor diagnosis.; In our model, female Sprague Dawley rats with N-ethyl-N-nitrosourea (ENU) induced mammary tumors imaged with fast T1-weighted gradient echo DCE MRI following a Gd-DTPA injection, there is a window of resolutions that detects similar PS “hot spots” compared to those obtained from the clinical imager resolution. The top five PS “hot spots” obtained from 469 μm resolution FFT are statistically different from those at 938 μm resolution FFT, p = 0.0014, and 2500 μm resolution FFT, p 0.0001. Keyhole when compared with a FFT of similar resolution does not detect PS “hot spots” of similar value, p = 0.0002. PS “hot spots” obtained from RIGR compared to those from FFT are statistically the same value, p = 0.2734, but do not statistically agree on the location of mapped values. The top five K_p↔t/VT “hot spots” and their corresponding v_e can statistically differentiate invasive ductal carcinoma from non-invasive papillary carcinoma for the 469 μm and 938 μm resolution, p = 0.0017 and p = 0.0047, respectively, but not for 2500 μm resolution, p = 0.9008.