We present new mid-IR observations of the quadruply lensed quasar Q2237+0305 taken with CanariCam on the Gran Telescopio Canarias. Mid-IR emission by hot dust, unlike the optical and near-IR emission from the accretion disk, is unaffected by the interstellar medium (extinction/scattering) or stellar microlensing. We compare these “true” ratios to the (stellar) microlensed flux ratios observed in the optical/near-IR to constrain the structure of the quasar accretion disk. We find a half-light radius of {R}1/2={3.4}-2.1+5.3\sqrt{< M> /0.3{M}ȯ } lt-day at {λ }{rest}=1736 \mathringA and an exponent for the temperature profile R\propto {λ }p of p=0.79+/- 0.55, where p=4/3 for a standard thin-disk model. If we assume that the differences in the mid-IR flux ratios measured over the years are due to microlensing variability, we find a lower limit for the size of the mid-IR-emitting region of {R}1/2≳ 200 \sqrt{< M> /0.3{M}ȯ } lt-day. We also test for the presence of substructure/satellites by comparing the observed mid-IR flux ratios with those predicted from smooth lens models. We can explain the differences if the surface density fraction in satellites near the lensed images is α ={0.033}-0.019+0.046 for a singular isothermal ellipsoid plus external shear mass model or α ={0.013}-0.008+0.019 for a mass model combining ellipsoidal NFW and de Vaucouleurs profiles in an external shear.