基于深度学习的胃癌病理切片图像分割算法研究

北京理工大学珠海学院2020届本科生毕业论文Research on Image Segmentation Algorithm of Gastric CancerPathological Slices Based on Deep LearningAbstractBased on the analysis and application of deep learning in medical image recognition at homeand abroad,this paper conducts a detailed study on the segmentation of cancer pathologicalslices,and proposes an automatic segmentation method ofgastric cancer pathological slicesbased on convolutional neural networks.This method is based on a full convolutional neu-ral network and incorporates a pyramid pooling module.The pyramid pooling model caneffectively improve the receptive field of the network,and it can better adapt to segmentedobjects of different sizes.And achieved a good segmentation effect.The experimental re-sults show that the average accuracy rate reaches 89%and MIoU reaches 65%.This methodhas been greatly improved compared with the traditional methods in the past.Keywords:Deep Learning Pyramid Pooling Model Medical Image Segmentation Gas-tric Cancer Pathological Section