We compare statistical and analytical reconstruction methods for micro-CT systems employing noncontinuous, modular CCD detectors. By combining many cooled CCDs connected with straight tapering to columnar CsI crystals, such systems would be capable of overcoming some major limitations of CCD-based technology, ie. limited field of view, insufficient detection efficency at low exposures and blurring caused by fiber-optic coupling. When applied to such a modular setup, statistical reconstruction (SR) has the advantage that it simply ignores the missing projection data. Filtered Back Projection assumes a continuous detector and therefore requires interpolation to fill in the projection discontinuities. Both reconstruction methods result in images with significant artifacts when applied to a system with symmetric placement of gaps. This can be explained by the incompleteness of projection data inherent to this kind of configuration. When the gaps are placed asymmetrically, however, SR effectively removes the errors caused by detector discontinuities, whereas FBP still produces images polluted by streaks.