Research Of Digital X-Ray Tomosynthesis Reconstruction

Tomosynthesis is a three-dimensional imaging technique by computer synthesis of projections based on traditional X-ray imaging system. The characteristic of this technique is the X-ray source and detector moving synchronously during projection acquisition and it can reconstruct an arbitrary slice from several projections in limited angle. Traditional X-ray imaging superimposes all the slices together which makes it difficult to distinguish objects in detail, but tomosynthesis can remove the influence from other slices effectively to reconstruct an arbitrary slice. Tomosynthesis has the advantage of less requirement of projection number, flexible in acquisition, fast scanning speed and low radiation does which can cover the shortage of traditional X-ray radiation and computed tomography.This paper studied on Shift and Add, Filtered Back Projection and Algebraic Reconstruction Technique deeply, compared the characteristics of these algorithms. We found analytical reconstruction algorithm was less effective than iteration reconstruction technique when three-dimensional Fourier space sampling was incomplete through static phantom simulation. In order to overcome the disadvantage in slow operation speed of iteration reconstruction technique we optimized the algorithm, reduced the space overhead and improved the reconstruction speed effectively.Two kinds of digital dynamic phantom were given by which original reconstruction technique were tested, and the results showed a significant motion artifact existed. To reduce the motion artifacts, a new multi-source scanning configuration was proposed for tomosynthesis in dynamic reconstruction using two dynamic digital phantoms. By quantitative comparison of single-source and multi-source reconstruction results, performance of multi-source tomosynthesis in dynamic reconstruction was evaluate.The results showed that multi-source tomosynthesis could reduce motion artifacts effectively, thus improving the accuracy of image exponentially beter than traditional scanning configuration. The paper confirmed the effectiveness of multi-source tomosynthesis in dynamic reconstruction which were practical and meaningful to improve cardiac imaging and respiratory imaging.