In indoor applications, computed tomography is the process of transforming a network of intersecting attenuation measurements into a spatially resolved two-dimensional concentration map. In this study the Low Third Derivative method (LTD) was numerically evaluated and optimized for different conditions. A modified version of the LTD algorithm (LTD m ) was proposed and evaluated against the original version. Eight test maps were reconstructed under different conditions, such as weight ratio, pixel resolution, beam density and measurement noise. Performance of both LTD algorithms was found to be intimately related to the number of peaks and complexity in the test map and the steepness of the peaks. The LTD m algorithm improved the quality, especially for concentration maps including steep gradients and regions with very low concentrations. The LTD m method heavily lessened aliasing distortions and efficiently minimized the effects of noise.