A Lung Dense Deep Convolution Neural Network for Robust Lung Parenchyma Segmentation

Abstract

Lung parenchyma segmentation is the prerequisite for an automatic diagnosis system to analyze lung CT (computed tomography) images. However, traditional lung segmentation algorithms have poor adaptability and are not effectively robust regarding lung databases with blood vessels and small voids which can interfere the segmentation. The main work of this paper is as follows: Firstly, a lung dense deep convolutional neural network (LDDNet) is proposed, which adopts some popular optimizer methods, such as dense block, batch normalization (BN) and dropout. The performance of LDDNet is tested on the public lung database LIDC-IDRI which contains many cases of interference for segmentation. Secondly, the labeled with blood vessels and small voids are not contained by the public ground-truth masks of the LIDC-IDRI database, therefore these regions are labeled by us with LabelMe software. Thirdly, for the aim of exploring the effect of image preprocessing on segmenting lung CT images with deep neural network, contrast enhancing, median filtering and Laplacian filtering are used to preprocess the image as comparative experiments. Finally, dataset is classified into four classes by the geometrical shapes to test the performance of LDDNet. The accuracy of the segmentation experiment reaches over 99% and the four classes can all reach over 95%. Additionally, blood vessels and small voids are segmented out from the lung parenchyma which is not achieved by other methods. Experimental results confirm that the proposed LDDNet can segment the lung parenchymal area more accurately and has better robustness in comparison with other neural networks and most of the traditional methods. INDEX TERMS Deep dense neural network, lung segmentation, robustness.

A Systematic Review of Automated Segmentation Methods and Public Datasets for the Lung and its Lobes and Findings on Computed Tomography Images

“…Even though these architectural modifications can bring improvements to the medical imaging segmentation performance, it has been shown that pre-training in natural images does not necessarily translate to better training in medical images. Regarding other types of methods, they are still being proposed and providing competitive performance in all included target structures [73,84,246]. Due to the deep learning requirements for high quality and varied annotated data, traditional techniques are starting to be used again in deep learning pipelines for regularization and semi-supervised learning [234].…”

A Method of Lung Organ Segmentation in CT Images Based on Multiple Residual Structures and an Enhanced Spatial Attention Mechanism

“…Convolutional neural networks (CNNs) are the core of deep learning-based machine vision and can effectively characterize images by directly applying visual laws. In the early days of semantic segmentation of medical images based on deep learning, deep convolutional neural networks (DCNN) [8,9] and full convolutional neural networks (FCN) [10][11][12] were widely used and achieved better results than traditional segmentation algorithms. However, DCNN and FCN algorithms have their own limitations.…”

A Pulmonary Vascular Extraction Algorithm from Chest CT/CTA Images

“…Jin et al [10] has proposed an easy-to-implement pulmonary parenchyma segmentation algorithm for chest CT images with the presence of borderline pulmonary nodules by integrating an improved two-dimensional convex packet algorithm with region growth and morphology for pulmonary contour and internal structure. Chen et al [11] proposed a dense deep convolutional network (LDDNet) for the pulmonary, which used some common optimizer methods such as dense block, batch normalization, and undersampling operations. Xiao et al [12] combined the threshold iterative segmentation method with the fractal geometry method for detecting the pulmonary depression boundary to perform the initial segmentation of the pulmonary parenchyma and then completed the accurate segmentation of the pulmonary parenchyma by convex packet repair to complete the accurate segmentation of pulmonary parenchyma.…”