Evaluation of a Physical Based Approach of Scattered Radiation Correction in Cone Beam CT for Non-Destructive Testing Applications

Cone Beam Computerized Tomography (CBCT) enables threedimensional imaging with isotropic resolution. X-ray scatter management is a challenging task for quantitative CBCT imaging : scattered radiation level is significantly high on cone beam systems compared to collimated fan beam systems. The effects of this scattered radiation are cupping artifacts, streaks, and quantification inaccuracies. At CEA-LETI, an original scatter management process without additional on-line acquisition has been developed, the API (Analytical transformation Plus Indexation based) method. The proposed method is composed of two steps: a scatter calibration is first performed through off-line acquisitions and is used to evaluate the level and a global shape of scattered radiation on on-line tomographic projections of the object. This global shape is adapted to the current acquisition with an analytical transformation issued from physical equations to evaluate the scattered radiation distribution on tomographic projections. This approach has been applied successfully in medical field. This paper presents in detail the API method and evaluates it in CBCT for Non-Destructive Testing (NDT). To evaluate API method, experimental radiographic projections of simple geometry objects were performed with a Thales Flaschscan flat panel detector and corrected from scatter by API method and by a beam stops method. Simulated projections of primary radiation were performed with the simulation software Sindbad, developed at CEA LETI. Reconstructions were then performed on experimental scatter corrected projections and on simulated projections. The API method provides results in good agreement with simulations, suppressing cupping artifact. The improvements due to the scatter correction by the API method have the same order of magnitude as those due to the correction by the beam stops method on experiments performed without antiscatter grid. The estimation of scatter is found to be more accurate with the API method than with the beam stops method on experiments performed with an antiscatter grid. API method improves NDT quantitatively and qualitatively without increasing on-line acquisition time. ECNDT 2006 Poster 39