Using FPGA to Accelerate Monte Carlo Superposition Based Radiation Dose Calculation

Radiation Therapy (RT) is a major modality for treating cancer by delivering radiation dose to cancer cells. The Monte Carlo Superposition (MCS) method provides a fast but accurate algorithm for radiation dose calculation. However, it is still too slow to meet the ever-increasing speed requirement. This thesis seeks to accelerate the Monte Carlo Superposition algorithm by developing a cost-effective hardware solution based on a FPGA platform. For a hardware design, data representation is critical for system performance. This thesis proposes a general method for finding fixed point data representation for floating point algorithms. One random number generator and the two main steps of the Monte Carlo Superposition algorithm: in-phantom and Multi-Leaf Collimator (MLC) raytracing, are implemented. Performance analysis and those already completed works show that a good speedup of the hardware design over the original software design can be achieved.