Images obtained from CT scanners are often used to estimate the radiation dose delivered to a target organ in oncology applications. Since a patient needs to be positioned in a similar manner as in a therapy machine, a portion of the patient is often positioned outside the scan field-of-view (SFOV). Projection truncation problem also occurs frequently in conventional CT scans and causes imaging artifacts that lead to suboptimal image quality. In this paper, we propose a reconstruction algorithm that enables an adequate estimation of the object outside the SFOV. We make use of the fact that the total attenuation of each ideal projection in a parallel sampling geometry remains constant over views. We use the magnitudes and slopes of the projection samples at the location of truncation to estimate water cylinders that can best fit to the projection data outside the SFOV. To improve the robustness of the algorithm, continuity constraints are placed on the fitting parameters. Extensive phantom and patient experiments were conducted to test the robustness and accuracy of the proposed algorithm.