作者: Jing Lian , Zhen Yang , Wenhao Sun , Yanan Guo , Li Zheng
DOI: 10.1016/J.NEUCOM.2018.12.007
关键词:
摘要: Abstract An image segmentation method of a modified simplified pulse-coupled neural network (MSPCNN) based on human visual system (HVS) is proposed for medical images. The successfully determines the stimulus input MSPCNN according to characteristics PCNN and HVS. In order accomplish goal, we attempt deduce sub-intensity range central neurons firing by introducing neighboring matrix Q calculating intensity distribution new MSPCNN(NMSPCNN), then reveal way how parameter Sint generates Sioij closer Besides, try substitute above into extract more suitable lesions contrast prevalent models, has higher accuracy rates lower computational complexity because setting method. Finally, good comparing with state-of-the-art methods better performance, presenting overall metric OEM MIAS 0.8784, DDSM 0.8606 gallstones 0.8585.