With the wide utilization of high-resolution computed tomography (HRCT) in the lung cancer screening, patients detected with pulmonary ground-glass nodules (GGNs) have increased over time and account for a large proportion of all thoracic diseases. Because of its less invasiveness and fast recovery, video-assisted thoracoscopic surgery (VATS) is currently the first choice of surgical approach to lung nodule resection. However, GGNs are usually difficult to recognize during VATS, and failure of nodule localization would result in conversion to thoracotomy or extended lung resection. In order to cope with this problem, a series of approaches for pulmonary nodule localization have developed in the last few years. This article aims to summarize the reported methods of lung nodule localization and analyze its corresponding pros and cons, in order to help thoracic surgeons to choose appropriate localization method in different clinical conditions.
With the development of thin section axial computed tomography scan, the detection rate of pulmonary ground-glass nodules (GGN) continues increasing. GGN has a special natural growth history: pure ground-glass nodules (PGGN) smaller than 10 mm can hold steady for a long term, surgery resection is unnecessary, patients need regular follow up. Larger part solid ground-glass nodules (PSN) with a solid component can be malignant early stage lung cancer, which requires early surgery intervention. Establishment of a standard definition of GGN growth, investments in the long term natural growth history of GGN, validation of the clinical, radiology and genetic risk factors would be beneficial for the management of GGN patients.
Objective To explore the independent risk factors for tumor invasiveness of ground-glass nodules and establish a tumor invasiveness prediction model. Methods A retrospective analysis was performed in 389 patients with ground-glass nodules admitted to the Department of Thoracic Surgery in the First Hospital of Lanzhou University from June 2018 to May 2021 with definite pathological findings, including clinical data, imaging features and tumor markers. A total of 242 patients were included in the study according to inclusion criteria, including 107 males and 135 females, with an average age of 57.98±9.57 years. CT data of included patients were imported into the artificial intelligence system in DICOM format. The artificial intelligence system recognized, automatically calculated and output the characteristics of pulmonary nodules, such as standard diameter, solid component size, volume, average CT value, maximum CT value, minimum CT value, central CT value, and whether there were lobulation, burr sign, pleural depression and blood vessel passing. The patients were divided into two groups: a preinvasive lesions group (atypical adenomatoid hyperplasia/adenocarcinoma in situ) and an invasive lesions group (minimally invasive adenocarcinoma/invasive adenocarcinoma). Univariate and multivariate analyses were used to screen the independent risk factors for tumor invasiveness of ground-glass nodules and then a prediction model was established. The receiver operating characteristic (ROC) curve was drawn, and the critical value was calculated. The sensitivity and specificity were obtained according to the Yorden index. Results Univariate and multivariate analyses showed that central CT value, Cyfra21-1, solid component size, nodular nature and burr of the nodules were independent risk factors for the diagnosis of tumor invasiveness of ground-glass nodules. The optimum critical value of the above indicators between preinvasive lesions and invasive lesions were –309.00 Hu, 3.23 ng/mL, 8.65 mm, respectively. The prediction model formula for tumor invasiveness probability was logit (P)=0.982–(3.369×nodular nature)+(0.921×solid component size)+(0.002×central CT value)+(0.526×Cyfra21-1)–(0.0953×burr). The areas under the curve obtained by plotting the ROC curve using the regression probabilities of regression model was 0.908. The accuracy rate was 91.3%. Conclusion The logistic regression model established in this study can well predict the tumor invasiveness of ground-glass nodules by CT and tumor markers with high predictive value.
ObjectiveTo establish a machine learning model based on computed tomography (CT) radiomics for preoperatively predicting invasive degree of lung ground-glass nodules (GGNs). MethodsWe retrospectively analyzed the clinical data of GGNs patients whose solid component less than 3 cm in the Department of Thoracic Surgery of Shanghai Pulmonary Hospital from March 2021 to July 2021 and the First Hospital of Lanzhou University from January 2019 to May 2022. The lesions were divided into pre-invasiveness and invasiveness according to postoperative pathological results, and the patients were randomly divided into a training set and a test set in a ratio of 7∶3. Radiomic features (1 317) were extracted from CT images of each patient, the max-relevance and min-redundancy (mRMR) was used to screen the top 100 features with the most relevant categories, least absolute shrinkage and selection operator (LASSO) was used to select radiomic features, and the support vector machine (SVM) classifier was used to establish the prediction model. We calculated the area under the curve (AUC), sensitivity, specificity, accuracy, negative predictive value, positive predictive value to evaluate the performance of the model, drawing calibration and decision curves of the prediction model to evaluate the accuracy and clinical benefit of the model, analyzed the performance in the training set and subgroups with different nodule diameters, and compared the prediction performance of this model with Mayo and Brock models. Two primary thoracic surgeons were required to evaluate the invasiveness of GGNs to investigate the clinical utility of the model. ResultsA total of 400 patients were divided into the training set (n=280) and the test set (n=120) according to the admission criteria. There were 267 females and 133 males with an average age of 52.4±12.7 years. Finally, 8 radiomic features were screened out from the training set data to build SVM model. The AUC, sensitivity and specificity of the model in the training and test sets were 0.91, 0.89, 0.75 and 0.86, 0.92, 0.60, respectively. The model showed good prediction performance in the training set 0-10 mm, 10-20 mm and the test set 0-10 mm, 10-20 mm subgroups, with AUC values of 0.82, 0.88, 0.84, 0.72, respectively. The AUC of SVM model was significantly better than that of Mayo model (0.73) and Brock model (0.73). With the help of this model, the AUC value, sensitivity, specificity and accuracy of thoracic surgeons A and B in distinguishing invasive or non-invasive adenocarcinoma were significantly improved. ConclusionThe SVM model based on radiomics is helpful to distinguish non-invasive lesions from invasive lesions, and has stable predictive performance for GGNs of different sizes and has better prediction performance than Mayo and Brock models. It can help clinicians to more accurately judge the invasiveness of GGNs, to make more appropriate diagnosis and treatment decisions, and achieve accurate treatment.
The subtype of lung cancer that presents as subsolid nodules on imaging exhibits unique biological behavior and favorable prognosis. Recently, the American Association for Thoracic Surgery (AATS) issued "The 2023 American Associationfor Thoracic Surgery (AATS) expert consensus document: Management of subsolid lung nodules". This consensus, based on the latest literature and current clinical experience, proposes updated strategies for managing subsolid nodules. It emphasizes the correlation between imaging findings and pathological classification, individualized follow-up and surgical management strategies for subsolid nodules, and multimodal treatment approaches for multiple subsolid pulmonary nodules.