Correlator design using genetic algorithms for medical imaging

摘要

Pulse compression using weighted chirp pulses require a set of bulky correlators to implement the underlying matched filters, each of which consists of a large number of multipliers. In medical ultrasound imaging, 64 or 128 identical correlators are required depending on the system level and the number of multipliers of each correlator can be larger than 256. In order to reduce the hardware complexity, a multiplierless correlator was designed based on genetic algorithms. In our design, each multiplier coefficients is constrained to be a signed-power-of-two (SPOT) integer or a sum of two such integer numbers. The designed multiplierless correlators are compared with the correlators using the ideal coefficient values and proven to provide the aimed performance with greatly reduced hardware complexity.