Spot decomposition in a novel pencil beam scanning proton computed tomography

Document Type


Publication Date



Range uncertainty is one of the most critical obstacles in proton therapy. Proton computed tomography (pCT) can potentially directly reconstruct relative stopping power (RSP) within 1%. We report the very first proton imaging technique based on spot decomposition within each spot for increasing efficiency of pCT using pencil beam scanning (PBS) technique. A 14cm-diameter cylinder water phantom was used in our simulation, embedded with three groups of cylinders (bone, muscle, and adipose, respectively). Each group of cylinders contains three different sizes (2cm, 1cm, and 3mm in diameter). A TOPAS Monte-Carlo model was developed simulating the on-board pCT image acquisition on a PBS gantry. A phase scorer was used to simulate a multi-layer pixelated proton residual energy detector. Each proton spot was divided into 19 overlapping sub-spots, and residual energy statistics were calculated for each sub-spot and were assigned to the nearest detector pixel. One hundred eighty projections were generated with 882 spots on each 10x10cm projection, using 200 MeV protons. Spot spacing and size (1-sigma) were 5.4mm and 6-8mm, respectively, on the detector. A direct path of proton transport route was assumed in an FDK-based reconstruction. pCT imaging dose was calculated, and the accuracy of RSP reconstruction was analyzed. Total dose to the phantom was 0.4mGy. The reconstructed mean RSP were 0.999(-0.1%), 0.969(-0.03%), 1.028(-0.2%), and 1.703(-0.9%) for the water, adipose, muscle, and bone, respectively. The STDs of the reconstructed RSP were <1.0%. This study has provided a new framework and offered an efficient approach for pCT acquisition and reconstruction.