A Breast Mass Segmentation Method Based on FFC-MSPCNN

Aiming at the problems of the traditional Pulse-coupled neural network (PCNN) for medical image processing field, including many parameters and undetermined iteration numbers, this paper proposes a Fast Fire-controlled FC-MSPCNN (FFC-MSPCNN) to achieve breast masses segmentation. The proposed method further optimizes and improves the weighting matrix, link strength and dynamic threshold amplitude than previous PCNN models. The parameters are set reasonably and the iteration numbers is obviously reduced within an effect firing period. What is more, we also add a variable parameter K to adjust dynamic threshold. According to a larger number of related experiments, our proposed method has good evaluation results than other compared algorithms, and obviously decrease the randomness and unpredictability of firing neurons.