基于改进深度卷积神经网络的病理图像有丝分裂检测算法研究

MITOSIS DETECTION ALGORITHM IN PATHOLOGICAL IMAGE BASED ON IMPROVE DEEP CONVOLUTION NEURAL NETWORK

针对病理图像中有丝分裂核的形态多变,而难区分、难检测的问题,提出一种基于改进深度卷积神经网络的计算机辅助有丝分裂检测算法。对原病理图像利用人工标签分割出有丝分裂核和非有丝分裂核小块作为候选集;利用ZCA白化方法对其进行预处理;对处理后的图像利用改进的卷积神经网络(CNN)逐层提取数据的高维特征;利用Softmax分类器进行分类。并用GPU进行加速实验以ICPR2012大赛数据集为例,通过实验其综合评价指标F-measure达到了0. 921 5。结果表明所提算法优于传统算法,可以更好地应用到有丝分裂检测中。

A computer aided mitosis detection algorithm based on improved deep convolution neural network is proposed to solve the problem that the morphology of mitotic nuclei in pathological images is changeable and difficult to detect. The nuclei mitotic and non-mitotic nuclei of original pathological images were segmented by artificial tags as candidate sets. ZCA whitening method was used to preprocess the images. The improved convolution neural network （CNN） was used to extract the high-dimensional features of the data after processing the images. The Softmax classifier was used for classification. Using GPU to accelerate, taking the ICPR2012 contest data set as an example, through the experiment, the comprehensive evaluation index F-measure reached 0.9215. The results show that the proposed algorithm is superior to the traditional algorithm and can be better applied to mitosis detection.