There is a clear need for technology that enables accurate, high-resolution, three-dimensional (3D) measurement of intricate dose distributions associated with modern radiation treatments. A potential candidate has emerged in the form of water-equivalent "3D gel dosimetry" utilizing optical-computed-tomography (optical-CT). In a previous paper we presented basic physical characterization of an in-house prototype optical-CT scanning system. The present paper builds on that work by investigating sources of optical artifacts and geometric distortion in optical-CT scanning. Improvements in scanner design are described. Correction strategies were developed to compensate for reflection and refraction, imperfections in the water-bath, signal drift, and other effects. Refraction and reflection were identified as the principal factors causing inaccurate reconstruction of absolute attenuation coefficients. A correction specific to a given flask was developed utilizing prescans of the flask when filled with water-bath fluid, thereby isolating the refractive and reflective components for that flask. Residual artifacts were corrected by fitting a theoretical model to the well-behaved portion of these prescans and extrapolating to regions of lost data, enabling reconstruction of absolute optical-CT attenuation coefficients to within 4% of corresponding spectrophotometer values. Needle phantoms are introduced to quantify geometric distortion under a range of conditions. Radial distortion of reconstructed needle positions was reduced to < 0.3 mm (0.27% of the field of view) through adjustment of the water-bath refractive index. Geometric distortion in polymer gel due to radiation-induced refractive index changes was found to be negligible under the conditions examined. The influence of scattered light on reconstructed attenuation coefficients was investigated by repeat optical-CT scans while varying the aperture of a scatter-rejecting collimator. Significant depression of reconstructed attenuation coefficients was observed, particularly under conditions of poor scatter rejection collimation. The general conclusion is that the first-generation optical-CT technique can be made insensitive to geometrical distortion, but can be susceptible to scatter effects. For accurate reconstruction of absolute attenuation coefficients, correction strategies are essential.