Purpose: Cone beam CT (CBCT) has been widely used in image guided radiation therapy. Patient cancer risk due to radiation exposure and degraded image quality due to x-ray scatter are issues affecting its application. We have proposed the use of a random beam blocker for simultaneous radiation exposure and scatter reduction. The randomness of blocking pattern is favoured by Compressive Sensing-type reconstruction. This study reports our experimental studies to validate this idea.
Methods: We constructed a blocker template using 3D printing and attached several tungsten BBs of diameter 1mm with a random pattern. The filling factor of the tungsten blocker (ratio of blocked area over total area) was 50%. The blocker is attached to the kV x-ray tube on a Varian TrueBeam linear accelerator. The x-ray spot to the blocker distance is 19cm and to the imager distance is 150cm. Hence, the blocker forms blocked circular regions in the detector with a diameter 0.8cm. The blocker rotates continuously during CBCT data acquisition. The pattern of the tungsten blockers is randomly changed at each projection. We performed CBCT scan of a homogeneous GAMMEX phantom. After acquiring the projection data, our in-house algorithm identified each blocked region and measured a scatter value. The measured scatter value was interpolated to the entire imager area, which was subtracted from the measured data in the unblocked region for scatter removal. The scatter removed data were used to reconstruct the CBCT image using an iterative TV-based algorithm.
Results: We successfully constructed the blocker apparatus. The use of the beam blocker significantly reduced scatter artefacts and reduced root-mean-square error from 94HU to 25HU. It is estimated to reduce the x-ray dose by 50%.
Conclusion: We have demonstrated the feasibility of using a random beam blocker for simultaneous imaging dose and scatter reduction.