Segmentation of lung parenchyma in CT images using CNN trained with the clustering algorithm generated dataset

Abstract

BackgroundLung segmentation constitutes a critical procedure for any clinical-decision supporting system aimed to improve the early diagnosis and treatment of lung diseases. Abnormal lungs mainly include lung parenchyma with commonalities on CT images across subjects, diseases and CT scanners, and lung lesions presenting various appearances. Segmentation of lung parenchyma can help locate and analyze the neighboring lesions, but is not well studied in the framework of machine learning.MethodsWe proposed to segment lung parenchyma using a convolutional neural network (CNN) model. To reduce the workload of manually preparing the dataset for training the CNN, one clustering algorithm based method is proposed firstly. Specifically, after splitting CT slices into image patches, the k-means clustering algorithm with two categories is performed twice using the mean and minimum intensity of image patch, respectively. A cross-shaped verification, a volume intersection, a connected component analysis and a patch expansion are followed to generate final dataset. Secondly, we design a CNN architecture consisting of only one convolutional layer with six kernels, followed by one maximum pooling layer and two fully connected layers. Using the generated dataset, a variety of CNN models are trained and optimized, and their performances are evaluated by eightfold cross-validation. A separate validation experiment is further conducted using a dataset of 201 subjects (4.62 billion patches) with lung cancer or chronic obstructive pulmonary disease, scanned by CT or PET/CT. The segmentation results by our method are compared with those yielded by manual segmentation and some available methods.ResultsA total of 121,728 patches are generated to train and validate the CNN models. After the parameter optimization, our CNN model achieves an average F-score of 0.9917 and an area of curve up to 0.9991 for classification of lung parenchyma and non-lung-parenchyma. The obtain model can segment the lung parenchyma accurately for 201 subjects with heterogeneous lung diseases and CT scanners. The overlap ratio between the manual segmentation and the one by our method reaches 0.96.ConclusionsThe results demonstrated that the proposed clustering algorithm based method can generate the training dataset for CNN models. The obtained CNN model can segment lung parenchyma with very satisfactory performance and have the potential to locate and analyze lung lesions.

Identification of COPD From Multi-View Snapshots of 3D Lung Airway Tree via Deep CNN

“…However, a real diagnosis model should be able to distinguish COPD from other airway diseases (e.g., asthma) and differentiate the severity of COPD (e.g., Global Initiative for Chronic Obstructive Lung Disease (GOLD) stage 2-4). Fourth, we only utilize the information of airway, but the lung parenchyma, patterns of LAA, air trapping, pulmonary blood vessels and medical record (e.g., demographic information, biomarkers, and some known medical diseases) are not included in the deep CNN model [4], [64]. Inclusion of the above information will likely improve the prediction performance.…”

Automatic Detection of Pulmonary Nodules using Three-dimensional Chain Coding and Optimized Random Forest

“…As well as, the third hyperparameter, the maximum number of depths of an individual tree is selected by finding the highest accuracy using a range from 2 to 20. As we can see in Figure 10c, the FR reached the highest accuracy (0.9285) at the depths (18); thus, we selected that number as the third hypermeter.…”

“…With the recent emergence of machine learning techniques, segmentation using machine-learning-algorithms became famous for lung cancer diagnosis. Multi-class pixel-wise segmentation using deep neural networks, such as SegNet [17] and CNN [18], are examples of deep-learning-based segmentation schemes. Due to the use of multiple network layers, these segmentations are in favor of unique feature learning [19].…”

Quantitative Assessment of Parenchymal Involvement Using 3D Lung Model in Adolescent With Covid-19 Interstitial Pneumonia

“…Moreover, since CT represents an important imaging biomarker and is pivotal to guide pharmacological management and improve ventilation strategies, further implementation of semi-automatic and/or fully automatic (AI-based) algorithms for image processing (12,13) might be beneficial in order to rapidly and systematically provide accurate data about the extent of lung disease in these patients.…”