Improved reconstructions and generalized filtered back projection for optical projection tomography

Optical projection tomography (OPT) is a noninvasive imaging technique that enables imaging of small specimens (<1?cm), such as organs or animals in early developmental stages. In this paper, we present a set of computational methods that can be applied to the acquired data sets in order to correct for (a) unknown background or illumination intensity distributions over the field of view, (b) intensity spikes in single CCD pixels (so-called "hot pixels"), and (c) refractive index mismatch between the media in which the specimens are embedded and the environment. We have tested these correction methods using a variety of samples and present results obtained from Parhyale hawaiensis embedded in glycerol and in sea water. Successful reconstructions of fluorescence and absorption OPT images have been obtained for weakly scattering specimens embedded in media with nonmatched refractive index, thus advancing OPT toward routine in vivo imaging.