Region-of-Interest Imaging in Cone Beam 3-D CT

Imaging a sectional region within an object with a detector just big enough to cover the sectional region-of-interest is analyzed. With otherwise complete (for small objects) cone beam scanning configurations, some of the Radon data are corrupted by unwanted measurements from outside the region-of-interest, and furthermore the region of corrupted data is in such a shape that the use of a priori information on the finite support of the object alone is not sufficient to recover the missing data. We show that with some suitable choice of scanning configuration and with an innovative method of data combination, all the Radon data can be obtained accurately. The algorithm is mathematically exact, and requires no iterations and no additional measurements. The method can be applied to inspect portions of large industrial objects in industrial imaging, as well as to image portions of human bodies in medical diagnosis.