基于全卷积神经网络的肺纤维化合并肺肿瘤CT图像的分割方法

CT image segmentationalgorithm of pulmonary fibrosis with lung tumor based on total convolution neural network

为提高CT图像分割提取图像特征的分割效果,设计基于全卷积神经网络的肺纤维化合并肺肿瘤CT图像的分割方法。肺部CT影像经过膨胀、腐蚀、孔洞填充、开运算、闭运算、掩模运算得到消除器官的肺实质图像,并提取ROI。通过改进全卷积神经网络结构,制定全卷积神经网络对于输入特征图的选取标准,完成CT图像分割算法的研究。选取IOU、Dice系数、精准率与召回率作为图像分割的评价指标。实验结果表明,经过对不同分割方法评价指标的比较,本研究设计的方法具有更理想的分割结果。

In order to improve the CT image segmentation effects and extraction of image features,we designed a CT image segmentation method based on full convolutional neural network for pulmonary fibrosis complicated with lung tumor.The lung CT images were processed by expansion,erosion,hole filling,opening operation,closing operation,and mask operation to obtain lung parenchymal images of the eliminated organs,and ROI was extracted.By improving the structure of the fully convolutional neural network and developing the selection criteria of the fully convolutional neural network for inputting feature maps,the research on the CT image segmentation algorithm was completed.IOU,Dice coefficient,precision rate and recall rate were selected as the evaluation indexes of image segmentation.The experimental results show that comparing to the evaluation indexes of different segmentation methods,the algorithm of the study has better segmentation performance.