Convolutional Neural Network-Based Diagnostic Model for a Solid, Indeterminate Solitary Pulmonary Nodule or Mass on Computed Tomography

PurposeTo establish a non-invasive diagnostic model based on convolutional neural networks (CNNs) to distinguish benign from malignant lesions manifesting as a solid, indeterminate solitary pulmonary nodule (SPN) or mass (SPM) on computed tomography (CT).MethodA total of 459 patients with solid indeterminate SPNs/SPMs on CT were ultimately included in this retrospective study and assigned to the train (n=366), validation (n=46), and test (n=47) sets. Histopathologic analysis was available for each patient. An end-to-end CNN model was proposed to predict the natural history of solid indeterminate SPN/SPMs on CT. Receiver operating characteristic curves were plotted to evaluate the predictive performance of the proposed CNN model. The accuracy, sensitivity, and specificity of diagnoses by radiologists alone were compared with those of diagnoses by radiologists by using the CNN model to assess its clinical utility.ResultsFor the CNN model, the AUC was 91% (95% confidence interval [CI]: 0.83–0.99) in the test set. The diagnostic accuracy of radiologists with the CNN model was significantly higher than that without the model (89 vs. 66%, P<0.01; 87 vs. 61%, P<0.01; 85 vs. 66%, P=0.03, in the train, validation, and test sets, respectively). In addition, while there was a slight increase in sensitivity, the specificity improved significantly by an average of 42% (the corresponding improvements in the three sets ranged from 43, 33, and 42% to 82, 78, and 84%, respectively; P<0.01 for all).ConclusionThe CNN model could be a valuable tool in non-invasively differentiating benign from malignant lesions manifesting as solid, indeterminate SPNs/SPMs on CT.

A Comparative Study of Radiomics and Deep-Learning Based Methods for Pulmonary Nodule Malignancy Prediction in Low Dose CT Images

ObjectivesBoth radiomics and deep learning methods have shown great promise in predicting lesion malignancy in various image-based oncology studies. However, it is still unclear which method to choose for a specific clinical problem given the access to the same amount of training data. In this study, we try to compare the performance of a series of carefully selected conventional radiomics methods, end-to-end deep learning models, and deep-feature based radiomics pipelines for pulmonary nodule malignancy prediction on an open database that consists of 1297 manually delineated lung nodules.MethodsConventional radiomics analysis was conducted by extracting standard handcrafted features from target nodule images. Several end-to-end deep classifier networks, including VGG, ResNet, DenseNet, and EfficientNet were employed to identify lung nodule malignancy as well. In addition to the baseline implementations, we also investigated the importance of feature selection and class balancing, as well as separating the features learned in the nodule target region and the background/context region. By pooling the radiomics and deep features together in a hybrid feature set, we investigated the compatibility of these two sets with respect to malignancy prediction.ResultsThe best baseline conventional radiomics model, deep learning model, and deep-feature based radiomics model achieved AUROC values (mean ± standard deviations) of 0.792 ± 0.025, 0.801 ± 0.018, and 0.817 ± 0.032, respectively through 5-fold cross-validation analyses. However, after trying out several optimization techniques, such as feature selection and data balancing, as well as adding context features, the corresponding best radiomics, end-to-end deep learning, and deep-feature based models achieved AUROC values of 0.921 ± 0.010, 0.824 ± 0.021, and 0.936 ± 0.011, respectively. We achieved the best prediction accuracy from the hybrid feature set (AUROC: 0.938 ± 0.010).ConclusionThe end-to-end deep-learning model outperforms conventional radiomics out of the box without much fine-tuning. On the other hand, fine-tuning the models lead to significant improvements in the prediction performance where the conventional and deep-feature based radiomics models achieved comparable results. The hybrid radiomics method seems to be the most promising model for lung nodule malignancy prediction in this comparative study.

Development and Analysis of a Predictive Nomogram Assessing Cancer Risk in a Chinese Cohort of Patients Presenting with Pulmonary Nodules

Background: Lung cancer is a major global threat to public health for which a novel prognostic nomogram is urgently needed.Patients and methods: Here, we designed a novel prognostic nomogram using a training dataset consisting of 178 pulmonary nodules for design and 124nodules for external validation. The R ‘caret’ package was used to separate patients for design into two groups, including a training cohort (n=126) for model construction and an internal validation cohort (n=52). Optimal feature selection for this model was achieved using the least absolute shrinkage and selection operator regression (LASSO) model. C-index values, calibration plots, and decision curve analyses were used to gauge the discrimination, calibration, and clinical utility, respectively, of this predictive model. Validation was then performed with the validation cohort.Results: A predictive nomogram was successfully constructed incorporating hypertension status, plasma fibrinogen levels, serum uric acid (SUA) levels, triglyceride (TG) and high-density lipoprotein (HDL) levels, density, spicule sign, ground-glass opacity (GGO), and pulmonary nodule size. This model exhibited good discriminative ability, with a C-index value of 0.795 (95% CI: 0.720–0.870), and was well-calibrated. When we used the validation cohort to evaluate the model, the C-indexes were 0.886 (95% CI: 0.800–0.972) and 0.817 (95% CI: 0.747–0.897) for internal validation and external validation, respectively. Decision curve analyses indicated the clinical value of this predictive nomogram when used at a lung cancer possibility threshold of 9%.Conclusion: The nomogram constructed in this study, which incorporates hypertension status, plasma fibrinogen levels, SUA, TG, HDL, density, spicule sign, GGO status, and pulmonary nodule size was able to reliably predict lung cancer risk in this Chinese cohort of patients presenting with pulmonary nodules.

Solitary pulmonary nodule malignancy predictive models applicable to routine clinical practice: a systematic review

Background Solitary pulmonary nodule (SPN) is a common finding in routine clinical practice when performing chest imaging tests. The vast majority of these nodules are benign, and only a small proportion are malignant. The application of predictive models of nodule malignancy in routine clinical practice would help to achieve better diagnostic management of SPN. The present systematic review was carried out with the purpose of critically assessing studies aimed at developing predictive models of solitary pulmonary nodule (SPN) malignancy from SPN incidentally detected in routine clinical practice. Methods We performed a search of available scientific literature until October 2020 in Pubmed, SCOPUS and Cochrane Central databases. The inclusion criteria were observational studies carried out in low-risk population from 35 years old onwards aimed at constructing predictive models of malignancy of pulmonary solitary nodule detected incidentally in routine clinical practice. Studies had to be published in peer-reviewed journals, either in Spanish, Portuguese or English. Exclusion criteria were non-human studies, or predictive models based in high-risk populations, or models based on computational approaches. Exclusion criteria were non-human studies, or predictive models based in high-risk populations, or models based on computational approaches (such as radiomics). We used The Transparent Reporting of a multivariable Prediction model for Individual Prognosis Or Diagnosis (TRIPOD) statement, to describe the type of predictive model included in each study, and The Prediction model Risk Of Bias ASsessment Tool (PROBAST) to evaluate the quality of the selected articles. Results A total of 186 references were retrieved, and after applying the exclusion/inclusion criteria, 15 articles remained for the final review. All studies analysed clinical and radiological variables. The most frequent independent predictors of SPN malignancy were, in order of frequency, age, diameter, spiculated edge, calcification and smoking history. Variables such as race, SPN growth rate, emphysema, fibrosis, apical scarring and exposure to asbestos, uranium and radon were not analysed by the majority of the studies. All studies were classified as high risk of bias due to inadequate study designs, selection bias, insufficient population follow-up and lack of external validation, compromising their applicability for clinical practice. Conclusions The studies included have been shown to have methodological weaknesses compromising the clinical applicability of the evaluated SPN malignancy predictive models and their potential influence on clinical decision-making for the SPN diagnostic management. Systematic review registration PROSPERO CRD42020161559

HR-MPF: high-resolution representation network with multi-scale progressive fusion for pulmonary nodule segmentation and classification

Accurate segmentation and classification of pulmonary nodules are of great significance to early detection and diagnosis of lung diseases, which can reduce the risk of developing lung cancer and improve patient survival rate. In this paper, we propose an effective network for pulmonary nodule segmentation and classification at one time based on adversarial training scheme. The segmentation network consists of a High-Resolution network with Multi-scale Progressive Fusion (HR-MPF) and a proposed Progressive Decoding Module (PDM) recovering final pixel-wise prediction results. Specifically, the proposed HR-MPF firstly incorporates boosted module to High-Resolution Network (HRNet) in a progressive feature fusion manner. In this case, feature communication is augmented among all levels in this high-resolution network. Then, downstream classification module would identify benign and malignant pulmonary nodules based on feature map from PDM. In the adversarial training scheme, a discriminator is set to optimize HR-MPF and PDM through back propagation. Meanwhile, a reasonably designed multi-task loss function optimizes performance of segmentation and classification overall. To improve the accuracy of boundary prediction crucial to nodule segmentation, a boundary consistency constraint is designed and incorporated in the segmentation loss function. Experiments on publicly available LUNA16 dataset show that the framework outperforms relevant advanced methods in quantitative evaluation and visual perception.