Development of a combined radiomics and CT feature-based model for differentiating malignant from benign subcentimeter solid pulmonary nodules. European radiology experimental BACKGROUND:We aimed to develop a combined model based on radiomics and computed tomography (CT) imaging features for use in differential diagnosis of benign and malignant subcentimeter (≤ 10 mm) solid pulmonary nodules (SSPNs). METHODS:A total of 324 patients with SSPNs were analyzed retrospectively between May 2016 and June 2022. Malignant nodules (n = 158) were confirmed by pathology, and benign nodules (n = 166) were confirmed by follow-up or pathology. SSPNs were divided into training (n = 226) and testing (n = 98) cohorts. A total of 2107 radiomics features were extracted from contrast-enhanced CT. The clinical and CT characteristics retained after univariate and multivariable logistic regression analyses were used to develop the clinical model. The combined model was established by associating radiomics features with CT imaging features using logistic regression. The performance of each model was evaluated using the area under the receiver-operating characteristic curve (AUC). RESULTS:Six CT imaging features were independent predictors of SSPNs, and four radiomics features were selected after a dimensionality reduction. The combined model constructed by the logistic regression method had the best performance in differentiating malignant from benign SSPNs, with an AUC of 0.942 (95% confidence interval 0.918-0.966) in the training group and an AUC of 0.930 (0.902-0.957) in the testing group. The decision curve analysis showed that the combined model had clinical application value. CONCLUSIONS:The combined model incorporating radiomics and CT imaging features had excellent discriminative ability and can potentially aid radiologists in diagnosing malignant from benign SSPNs. RELEVANCE STATEMENT:The model combined radiomics features and clinical features achieved good efficiency in predicting malignant from benign SSPNs, having the potential to assist in early diagnosis of lung cancer and improving follow-up strategies in clinical work. KEY POINTS:• We developed a pulmonary nodule diagnostic model including radiomics and CT features. • The model yielded the best performance in differentiating malignant from benign nodules. • The combined model had clinical application value and excellent discriminative ability. • The model can assist radiologists in diagnosing malignant from benign pulmonary nodules. 10.1186/s41747-023-00400-6

Small solitary pulmonary nodules (< or =1 cm) detected at population-based CT screening for lung cancer: Reliable high-resolution CT features of benign lesions. Takashima Shodayu,Sone Shusuke,Li Feng,Maruyama Yuichiro,Hasegawa Minoru,Matsushita Tsuyoshi,Takayama Fumiyoshi,Kadoya Masumi AJR. American journal of roentgenology OBJECTIVE:We assessed thin-section CT features specific to benignity in solitary pulmonary nodules of 1 cm or smaller that were detected at population-based CT screening for lung cancer. MATERIALS AND METHODS:Two reviewers independently made qualitative (presence or absence of lobulation, spiculation, air bronchogram, cavity, satellite lesions, pleural tag, concave margins, polygonal shape, and peripheral subpleural lesion) and quantitative (lesion size, percentage of ground-glass opacity areas, and two- and three-dimensional ratios of lesion) assessments in CT images of 72 nodules (25 lung cancers, seven atypical adenomatous hyperplasias, and 40 benign lesions). Optimal criteria specific to benignity were studied. RESULTS:The prevalence of polygonal shape (p = 0.005 and p = 0.019, reviewer 1 and reviewer 2), peripheral subpleural lesion (p = 0.011 and p = 0.033), a predominantly solid lesion (p < 0.001 and p < 0.001), and three-dimensional ratios (p < 0.001 and p < 0.001) were greater in benign lesions than in malignancies. The prevalence of a predominantly solid lesion (p < 0.001 and p < 0.001) was greater in benign lesions than in atypical adenomatous hyperplasias, and the prevalence of a peripheral subpleural lesion (p = 0.004 and p = 0.012) was greater in atypical adenomatous hyperplasias than in malignancies. Using a single CT feature, polygonal shape and a three-dimensional ratio of greater than 1.78 showed 100% specificity for both reviewers. Among all combinations of CT findings specific to benignity, a combined criterion of a predominantly solid lesion and peripheral subpleural lesion or polygonal shape or the three-dimensional ratio attained the highest sensitivity (63% and 60%) for both reviewers. CONCLUSION:A combined criterion of CT features was optimal for predicting benign pulmonary lesions. 10.2214/ajr.180.4.1800955

Clinical and Computed Tomography Characteristics of Inflammatory Solid Pulmonary Nodules with Morphology Suggesting Malignancy. Academic radiology RATIONALE AND OBJECTIVES:To investigate the clinical and computed tomography characteristics of inflammatory solid pulmonary nodules (SPNs) with morphology suggesting malignancy, hereinafter referred to as atypical inflammatory SPNs (AI-SPNs). MATERIALS AND METHODS:The CT data of 515 patients with SPNs who underwent surgical resection were retrospectively analyzed. These patients were divided into inflammatory and malignant groups and their clinical and imaging features were compared. Binary logistic regression analysis was performed to identify the independent factors for diagnosing AI-SPNs. An external validation cohort included 133 consecutive patients to test the model's predictive efficiency. RESULTS:Univariate analysis showed that age < 62 years, male sex, maximum spiculation length > 9 mm, polygonal shapes, three-planar ratio > 1.48, Lung window/mediastinal window (L/M) ratio > 1.13, pleural tag type I, satellite lesions, and halo sign were more frequent in AI-SPNs, whereas pleural tag type III, bronchial truncation, and perifocal fibrosis were more common in malignant SPNs (M-SPNs) (all P < 0.05). Binary logistic regression showed age < 62 years, male sex, polygonal shape, three-planar ratio > 1.48, L/M ratio > 1.13, pleural tag type I, satellite lesions, halo sign, and absence of bronchial truncation were independent factors for diagnosing AI-SPNs (AUC, sensitivity, specificity, and accuracy of 0.951, 83.30%, 92.30%, and 87.20%, respectively). In the external validation cohort, the AUC, sensitivity, specificity, and accuracy were 0.969, 90.47%, 90.00%, and 90.23%, respectively. CONCLUSION:AI-SPNs and M-SPNs exhibited different clinical and imaging characteristics. A good understanding of these differences may help reduce diagnostic errors in AI-SPNs and enable to choose an optimal treatment strategy. 10.1016/j.acra.2024.09.016

Enhancing a deep learning model for pulmonary nodule malignancy risk estimation in chest CT with uncertainty estimation. European radiology OBJECTIVE:To investigate the effect of uncertainty estimation on the performance of a Deep Learning (DL) algorithm for estimating malignancy risk of pulmonary nodules. METHODS AND MATERIALS:In this retrospective study, we integrated an uncertainty estimation method into a previously developed DL algorithm for nodule malignancy risk estimation. Uncertainty thresholds were developed using CT data from the Danish Lung Cancer Screening Trial (DLCST), containing 883 nodules (65 malignant) collected between 2004 and 2010. We used thresholds on the 90th and 95th percentiles of the uncertainty score distribution to categorize nodules into certain and uncertain groups. External validation was performed on clinical CT data from a tertiary academic center containing 374 nodules (207 malignant) collected between 2004 and 2012. DL performance was measured using area under the ROC curve (AUC) for the full set of nodules, for the certain cases and for the uncertain cases. Additionally, nodule characteristics were compared to identify trends for inducing uncertainty. RESULTS:The DL algorithm performed significantly worse in the uncertain group compared to the certain group of DLCST (AUC 0.62 (95% CI: 0.49, 0.76) vs 0.93 (95% CI: 0.88, 0.97); p < .001) and the clinical dataset (AUC 0.62 (95% CI: 0.50, 0.73) vs 0.90 (95% CI: 0.86, 0.94); p < .001). The uncertain group included larger benign nodules as well as more part-solid and non-solid nodules than the certain group. CONCLUSION:The integrated uncertainty estimation showed excellent performance for identifying uncertain cases in which the DL-based nodule malignancy risk estimation algorithm had significantly worse performance. CLINICAL RELEVANCE STATEMENT:Deep Learning algorithms often lack the ability to gauge and communicate uncertainty. For safe clinical implementation, uncertainty estimation is of pivotal importance to identify cases where the deep learning algorithm harbors doubt in its prediction. KEY POINTS:• Deep learning (DL) algorithms often lack uncertainty estimation, which potentially reduce the risk of errors and improve safety during clinical adoption of the DL algorithm. • Uncertainty estimation identifies pulmonary nodules in which the discriminative performance of the DL algorithm is significantly worse. • Uncertainty estimation can further enhance the benefits of the DL algorithm and improve its safety and trustworthiness. 10.1007/s00330-024-10714-7

Evaluation of Prognosis in Patients with Lung Adenocarcinoma with Atypical Solid Nodules on Thin-Section CT Images. Radiology. Cardiothoracic imaging Purpose To evaluate the clinicopathologic characteristics and prognosis of patients with clinical stage IA lung adenocarcinoma with atypical solid nodules (ASNs) on thin-section CT images. Materials and Methods Data from patients with clinical stage IA lung adenocarcinoma who underwent resection between January 2005 and December 2012 were retrospectively reviewed. According to their manifestations on thin-section CT images, nodules were classified as ASNs, subsolid nodules (SSNs), and typical solid nodules (TSNs). The clinicopathologic characteristics of the ASNs were investigated, and the differences across the three groups were analyzed. The Kaplan-Meier method and multivariable Cox analysis were used to evaluate survival differences among patients with ASNs, SSNs, and TSNs. Results Of the 254 patients (median age, 58 years [IQR, 53-66]; 152 women) evaluated, 49 had ASNs, 123 had SSNs, and 82 had TSNs. Compared with patients with SSNs, those with ASNs were more likely to have nonsmall adenocarcinoma ( < .001), advanced-stage adenocarcinoma ( = .004), nonlepidic growth adenocarcinoma ( < .001), and middle- or low-grade differentiation tumors ( < .001). Compared with patients with TSNs, those with ASNs were more likely to have no lymph node involvement ( = .009) and epidermal growth factor receptor mutation positivity ( = .018). Average disease-free survival in patients with ASNs was significantly longer than that in patients with TSNs ( < .001) but was not distinguishable from that in patients with SSNs ( = .051). Conclusion ASNs were associated with better clinical outcomes than TSNs in patients with clinical stage IA lung adenocarcinoma. Adenocarcinoma, Atypical Solid Nodules, CT, Disease-free Survival, Lung, Prognosis, Pulmonary . Published under a CC BY 4.0 license. 10.1148/ryct.220234