Lung cancer is a leading cause of cancer-related death worldwide and in the USA. Low Dose Computed tomography (LDCT) is the primary method of detection and diagnosis of lung cancers. Radiomics provides further analysis using LDCT scans which provide an opportunity for early detection and diagnosis of lung cancers. The convolutional neural network (CNN), a powerful method for image classification and recognition, has opened an alternative path for tumor identification and detection from LDCT scans. Nodules have different shapes, boundaries or patterns. In this study, we created feature images from different texture features of nodules and then used a CNN to classify each of the feature images. We call this approach “Deep Radiomics”. Law’s 3-D texture images were used for our analysis. Ten Law’s texture images were generated and used to train an ensemble of CNNs. Texture provides information about how an image looks. The use of feature images as CNN input is a novel approach to enable the generation and extraction of new types of features and lends itself to ensemble generation. From the LDCT arm of the national lung cancer screening study (NLST) dataset, a subset of nodule positive and screen-detected lung cancer (SDLC) cases were used in our study. The best result obtained from this study was 79.32% accuracy and 0.88 AUC, which is an improvement in accuracy over using just image features or just original images as input to CNNs for classification.
Lung cancer has a high incidence and mortality rate. Early detection and diagnosis of lung cancers is best achieved with low-dose computed tomography (CT). Classical radiomics features extracted from lung CT images have been shown as able to predict cancer incidence and prognosis. With the advancement of deep learning and convolutional neural networks (CNNs), deep features can be identified to analyze lung CTs for prognosis prediction and diagnosis. Due to a limited number of available images in the medical field, the transfer learning concept can be helpful. Using subsets of participants from the National Lung Screening Trial (NLST), we utilized a transfer learning approach to differentiate lung cancer nodules versus positive controls. We experimented with three different pretrained CNNs for extracting deep features and used five different classifiers. Experiments were also conducted with deep features from different color channels of a pretrained CNN. Selected deep features were combined with radiomics features. A CNN was designed and trained. Combinations of features from pretrained, CNNs trained on NLST data, and classical radiomics were used to build classifiers. The best accuracy (76.79%) was obtained using feature combinations. An area under the receiver operating characteristic curve of 0.87 was obtained using a CNN trained on an augmented NLST data cohort.
Radiomics is the process of analyzing radiological images by extracting quantitative features for monitoring and diagnosis of various cancers. Analyzing images acquired from different medical centers is confounded by many choices in acquisition, reconstruction parameters and differences among device manufacturers. Consequently, scanning the same patient or phantom using various acquisition/reconstruction parameters as well as different scanners may result in different feature values. To further evaluate this issue, in this study, CT images from a physical radiomic phantom were used. Recent studies showed that some quantitative features were dependent on voxel size and that this dependency could be reduced or removed by the appropriate normalization factor. Deep features extracted from a convolutional neural network, may also provide additional features for image analysis. Using a transfer learning approach, we obtained deep features from three convolutional neural networks pre-trained on color camera images. An we examination of the dependency of deep features on image pixel size was done. We found that some deep features were pixel size dependent, and to remove this dependency we proposed two effective normalization approaches. For analyzing the effects of normalization, a threshold has been used based on the calculated standard deviation and average distance from a best fit horizontal line among the features’ underlying pixel size before and after normalization. The inter and intra scanner dependency of deep features has also been evaluated.