High-performance hardware 3D medical imaging using wavelets in the residue number system

Abstract

Discrete wavelet transform (DWT) is used in medical imaging to solve many problems. The practical DWT implementation requires a lot of time. We propose parallelizing computations by using the non-positional residue number system (RNS) to improve the performance of medical imaging systems. We use RNS moduli of a special type which allows us to reduce the computational complexity of the arithmetic operations used for 3D tomographic imaging. In mathematical and hardware modeling, all calculations are performed with fixed-point numbers. Hardware implementation is carried out in Xilinx Vivado 2019.2 for the target device FPGA Kintex UltraScale xcku115-flvf1924-3-e and allows to increase the performance of medical imaging devices by 1.57 - 2.70 times