Brock Model Research Articles

Reducing uncertainty in cancer risk estimation for patients with indeterminate pulmonary nodules using an integrated deep learning model

ObjectivePatients with indeterminate pulmonary nodules (IPN) with an intermediate to a high probability of lung cancer generally undergo invasive diagnostic procedures. Chest computed tomography image and clinical data have been in estimating the pretest probability of lung cancer. In this study, we apply a deep learning network to integrate multi-modal data from CT images and clinical data (including blood-based biomarkers) to improve lung cancer diagnosis. Our goal is to reduce uncertainty and to avoid morbidity, mortality, over- and undertreatment of patients with IPNs. MethodWe use a retrospective study design with cross-validation and external-validation from four different sites. We introduce a deep learning framework with a two-path structure to learn from CT images and clinical data. The proposed model can learn and predict with single modality if the multi-modal data is not complete. We use 1284 patients in the learning cohort for model development. Three external sites (with 155, 136 and 96 patients, respectively) provided patient data for external validation. We compare our model to widely applied clinical prediction models (Mayo and Brock models) and image-only methods (e.g., Liao et al. model). ResultsOur co-learning model improves upon the performance of clinical-factor-only (Mayo and Brock models) and image-only (Liao et al.) models in both cross-validation of learning cohort (e.g., AUC: 0.787 (ours) vs. 0.707–0.719 (baselines), results reported in validation fold and external-validation using three datasets from University of Pittsburgh Medical Center (e.g., 0.918 (ours) vs. 0.828–0.886 (baselines)), Detection of Early Cancer Among Military Personnel (e.g., 0.712 (ours) vs. 0.576–0.709 (baselines)), and University of Colorado Denver (e.g., 0.847 (ours) vs. 0.679–0.746 (baselines)). In addition, our model achieves better re-classification performance (cNRI 0.04 to 0.20) in all cross- and external-validation sets compared to the Mayo model. ConclusionsLung cancer risk estimation in patients with IPNs can benefit from the co-learning of CT image and clinical data. Learning from more subjects, even though those only have a single modality, can improve the prediction accuracy. An integrated deep learning model can achieve reasonable discrimination and re-classification performance.

The diagnostic value of machine-learning-based model for predicting the malignancy of solid nodules in multiple pulmonary nodules

Objective: To examine the efficiacy of a machine learning diagnostic model specifically for solid nodules in multiple pulmonary nodules constructed by combining patient clinical information and CT features. Methods: Totally 446 solid nodules resected from 287 patients with multiple pulmonary nodules in Department of Thoracic Surgery, Peking University People's Hospital from January 2010 to December 2018 were included. There were 117 males and 170 females, aging (61.4±9.9) yeras (range: 33 to 84 years). The nodules were randomly divided into training set (228 patients with 357 nodules) and test set (59 patients with 89 nodules) by a ratio of 4∶1. The extreme gradient boosting (XGBoost) algorithm was used to generate a predictive model (PKU-ML model) on the training set. The accuracy was verified on the test set and compared with previous published models. Finally, an independent single solid nodule set (155 patients, 95 males, aging (62.3±8.3) years (range: 37 to 77 years)) was used to evaluate the accuracy of the model for predictive value of single solid nodules. Area of receiver operating characteristic curve (AUC) was used to evaluate diagnostic values of models. Results: In the training set, the AUC of the PKU-ML model was 0.883 (95%CI: 0.849 to 0.917). In the test set, the performance of the PKU-ML model (AUC=0.838, 95%CI: 0.754 to 0.921) was better than the models designed for single pulmonary nodules (Brock model: AUC=0.709, 95%CI: 0.603 to 0.816, P=0.04; Mayo model: AUC=0.756, 95%CI: 0.656 to 0.856, P=0.01; VA model: AUC=0.674, 95%CI: 0.561 to 0.787, P<0.01), similar with PKUPH model (AUC=0.750, 95%CI: 0.649 to 0.851, P=0.07). In the independent single solid nodules set, the PKU-ML model also achieved good performance (AUC=0.786, 95%CI: 0.701 to 0.872). Conclusion: The machine learning based PKU-ML model can better predict the malignancy of solid nodules in multiple pulmonary nodules, and also achieved a good performance in predicting the malignancy of single solid pulmonary nodules compared to mathematical models.

Developing an understanding of artificial intelligence lung nodule risk prediction using insights from the Brock model

ObjectivesTo determine if predictions of the Lung Cancer Prediction convolutional neural network (LCP-CNN) artificial intelligence (AI) model are analogous to the Brock model.MethodsIn total, 10,485 lung nodules in 4660 participants from the National Lung Screening Trial (NLST) were analysed. Both manual and automated nodule measurements were inputted into the Brock model, and this was compared to LCP-CNN. The performance of an experimental AI model was tested after ablating imaging features in a manner analogous to removing predictors from the Brock model. First, the nodule was ablated leaving lung parenchyma only. Second, a sphere of the same size as the nodule was implanted in the parenchyma. Third, internal texture of both nodule and parenchyma was ablated.ResultsAutomated axial diameter (AUC 0.883) and automated equivalent spherical diameter (AUC 0.896) significantly improved the accuracy of Brock when compared to manual measurement (AUC 0.873), although not to the level of the LCP-CNN (AUC 0.936). Ablating nodule and parenchyma texture (AUC 0.915) led to a small drop in AI predictive accuracy, as did implanting a sphere of the same size as the nodule (AUC 0.889). Ablating the nodule leaving parenchyma only led to a large drop in AI performance (AUC 0.717).ConclusionsFeature ablation is a feasible technique for understanding AI model predictions. Nodule size and morphology play the largest role in AI prediction, with nodule internal texture and background parenchyma playing a limited role. This is broadly analogous to the relative importance of morphological factors over clinical factors within the Brock model.Key Points• Brock lung cancer risk prediction accuracy was significantly improved using automated axial or equivalent spherical measurements of lung nodule diameter, when compared to manual measurements.• Predictive accuracy was further improved by using the Lung Cancer Prediction convolutional neural network, an artificial intelligence-based model which obviates the requirement for nodule measurement.• Nodule size and morphology are important factors in artificial intelligence lung cancer risk prediction, with nodule texture and background parenchyma contributing a small, but measurable, role.

Cancer Risk Estimation Combining Lung Screening CT with Clinical Data Elements.

To develop a model to estimate lung cancer risk using lung cancer screening CT and clinical data elements (CDEs) without manual reading efforts. Two screening cohorts were retrospectively studied: the National Lung Screening Trial (NLST; participants enrolled between August 2002 and April 2004) and the Vanderbilt Lung Screening Program (VLSP; participants enrolled between 2015 and 2018). Fivefold cross-validation using the NLST dataset was used for initial development and assessment of the co-learning model using whole CT scans and CDEs. The VLSP dataset was used for external testing of the developed model. Area under the receiver operating characteristic curve (AUC) and area under the precision-recall curve were used to measure the performance of the model. The developed model was compared with published risk-prediction models that used only CDEs or imaging data alone. The Brock model was also included for comparison by imputing missing values for patients without a dominant pulmonary nodule. A total of 23 505 patients from the NLST (mean age, 62 years ± 5 [standard deviation]; 13 838 men, 9667 women) and 147 patients from the VLSP (mean age, 65 years ± 5; 82 men, 65 women) were included. Using cross-validation on the NLST dataset, the AUC of the proposed co-learning model (AUC, 0.88) was higher than the published models predicted with CDEs only (AUC, 0.69; P < .05) and with images only (AUC, 0.86; P < .05). Additionally, using the external VLSP test dataset, the co-learning model had a higher performance than each of the published individual models (AUC, 0.91 [co-learning] vs 0.59 [CDE-only] and 0.88 [image-only]; P < .05 for both comparisons). The proposed co-learning predictive model combining chest CT images and CDEs had a higher performance for lung cancer risk prediction than models that contained only CDE or only image data; the proposed model also had a higher performance than the Brock model.Keywords: Computer-aided Diagnosis (CAD), CT, Lung, Thorax Supplemental material is available for this article. © RSNA, 2021.

Prediction Model for Lung Cancer in High-Risk Nodules Being Considered for Resection: Development and Validation in a Chinese Population.

BackgroundDetermining benign and malignant nodules before surgery is very difficult when managing patients with pulmonary nodules, which further makes it difficult to choose an appropriate treatment. This study aimed to develop a lung cancer risk prediction model for predicting the nature of the nodule in patients’ lungs and deciding whether to perform a surgical intervention.MethodsThis retrospective study included patients with pulmonary nodules who underwent lobectomy or sublobectomy at Tianjin Medical University General Hospital between 2017 and 2020. All subjects were further divided into training and validation sets. Multivariable logistic regression models with backward selection based on the Akaike information criterion were used to identify independent predictors and develop prediction models.ResultsTo build and validate the model, 503 and 260 malignant and benign nodules were used. Covariates predicting lung cancer in the current model included female sex, age, smoking history, nodule type (pure ground-glass and part-solid), nodule diameter, lobulation, margin (smooth, or spiculated), calcification, intranodular vascularity, pleural indentation, and carcinoembryonic antigen. The final model of this study showed excellent discrimination and calibration with a concordance index (C-index) of 0.914 (0.890–0.939). In an independent sample used for validation, the C-index for the current model was 0.876 (0.825–0.927) compared with 0.644 (0.559–0.728) and 0.681 (0.605–0.757) for the Mayo and Brock models. The decision curve analysis showed that the current model had higher discriminatory power for malignancy than the Mayo and the Brock models.ConclusionsThe current model can be used in estimating the probability of lung cancer in nodules requiring surgical intervention. It may reduce unnecessary procedures for benign nodules and prompt diagnosis and treatment of malignant nodules.

The Probability of Lung Cancer in Patients With Incidentally Detected Pulmonary Nodules: Clinical Characteristics and Accuracy of Prediction Models

The frequency of cancer and accuracy of prediction models have not been studied in large, population-based samples of patients with incidental pulmonary nodules measuring > 8mm in diameter. How does the frequency of cancer vary by size and smoking history among patients with incidental nodules? How accurate are two widely used models for identifying cancer in these patients? We assembled a retrospective cohort of individuals with incidental nodules measuring > 8mm in diameter identified by chest CT imaging between 2006 and 2016. We used a validated natural language processing algorithm to identify nodules and their characteristics by scanning the text of dictated radiology reports. We reported patient and nodule characteristics stratified by the presence or absence of a lung cancer diagnosis within 27months of nodule identification and estimated the area under the receiver operating characteristic curve (AUC) to compare the accuracy of the Mayo Clinic and Brock models for identifying cancer. The sample included 23,780 individuals with a nodule measuring > 8mm, including 2,356 patients (9.9%) with a lung cancer diagnosis within 27months of nodule identification. Cancer was diagnosed in 5.4%of never smokers, 12.2%of former smokers, and 17.7%of current smokers. Cancer was diagnosed in 5.7%of patients with nodules measuring 9 to 15mm, 12.1%of patients with nodules > 15 to 20mm, and 18.4%of patients with nodules > 20 to 30mm. In the full sample, the Mayo Clinic model (AUC, 0.747; 95%CI, 0.737-0.757) was more accurate than the Brock model (AUC, 0.713; 95%CI, 0.702-0.724; P< .0001). When restricted to ever smokers, the Mayo Clinic model was still more accurate. Both models overestimated the probability of cancer. Almost 10%of patients with an incidental pulmonary nodule measuring > 8mm in diameter will receive a lung cancer diagnosis. Existing prediction models have only fair accuracy and overestimate the probability of cancer.

A simple assessment of lung nodule location for reduction in unnecessary invasive procedures.

BackgroundCT screening for lung cancer results in a significant mortality reduction but is complicated by invasive procedures performed for evaluation of the many detected benign nodules. The purpose of this study was to evaluate measures of nodule location within the lung as predictors of malignancy.MethodsWe analyzed images and data from 3,483 participants in the National Lung Screening Trial (NLST). All nodules (4–20 mm) were characterized by 3D geospatial location using a Cartesian coordinate system and evaluated in logistic regression analysis. Model development and probability cutpoint selection was performed in the NLST testing set. The Geospatial test was then validated in the NLST testing set, and subsequently replicated in a new cohort of 147 participants from The Detection of Early Lung Cancer Among Military Personnel (DECAMP) Consortium.ResultsThe Geospatial Test, consisting of the superior-inferior distance (Z distance), nodule diameter, and radial distance (carina to nodule) performed well in both the NLST validation set (AUC 0.85) and the DECAMP replication cohort (AUC 0.75). A negative Geospatial Test resulted in a less than 2% risk of cancer across all nodule diameters. The Geospatial Test correctly reclassified 19.7% of indeterminate nodules with a diameter over 6mm as benign, while only incorrectly classifying 1% of cancerous nodules as benign. In contrast, the parsimonious Brock Model applied to the same group of nodules correctly reclassified 64.5% of indeterminate nodules as benign but resulted in misclassification of a cancer as benign in 18.2% of the cases. Applying the Geospatial test would result in reducing invasive procedures performed for benign lesions by 11.3% with a low rate of misclassification (1.3%). In contrast, the Brock model applied to the same group of patients results in decreasing invasive procedures for benign lesion by 39.0% but misclassifying 21.1% of cancers as benign.ConclusionsUtilizing information about geospatial location within the lung improves risk assessment for indeterminate lung nodules and may reduce unnecessary procedures.Trial RegistrationNCT00047385, NCT01785342.

Development and Validation of Machine Learning-based Model for the Prediction of Malignancy in Multiple Pulmonary Nodules: Analysis from Multicentric Cohorts.

Nodule evaluation is challenging and critical to diagnose multiple pulmonary nodules (MPNs). We aimed to develop and validate a machine learning-based model to estimate the malignant probability of MPNs to guide decision-making. A boosted ensemble algorithm (XGBoost) was used to predict malignancy using the clinicoradiologic variables of 1,739 nodules from 520 patients with MPNs at a Chinese center. The model (PKU-M model) was trained using 10-fold cross-validation in which hyperparameters were selected and fine-tuned. The model was validated and compared with solitary pulmonary nodule (SPN) models, clinicians, and a computer-aided diagnosis (CADx) system in an independent transnational cohort and a prospective multicentric cohort. The PKU-M model showed excellent discrimination [area under the curve; AUC (95% confidence interval (95% CI)), 0.909 (0.854-0.946)] and calibration (Brier score, 0.122) in the development cohort. External validation (583 nodules) revealed that the AUC of the PKU-M model was 0.890 (0.859-0.916), higher than those of the Brock model [0.806 (0.771-0.838)], PKU model [0.780 (0.743-0.817)], Mayo model [0.739 (0.697-0.776)], and VA model [0.682 (0.640-0.722)]. Prospective comparison (200 nodules) showed that the AUC of the PKU-M model [0.871 (0.815-0.915)] was higher than that of surgeons [0.790 (0.711-0.852), 0.741 (0.662-0.804), and 0.727 (0.650-0.788)], radiologist [0.748 (0.671-0.814)], and the CADx system [0.757 (0.682-0.818)]. Furthermore, the model outperformed the clinicians with an increase of 14.3% in sensitivity and 7.8% in specificity. After its development using machine learning algorithms, validation using transnational multicentric cohorts, and prospective comparison with clinicians and the CADx system, this novel prediction model for MPNs presented solid performance as a convenient reference to help decision-making.

Validation of the BRODERS classifier (Benign versus aggRessive nODule Evaluation using Radiomic Stratification), a novel HRCT-based radiomic classifier for indeterminate pulmonary nodules.

Implementation of low-dose chest computed tomography (CT) lung cancer screening and the ever-increasing use of cross-sectional imaging are resulting in the identification of many screen- and incidentally detected indeterminate pulmonary nodules. While the management of nodules with low or high pre-test probability of malignancy is relatively straightforward, those with intermediate pre-test probability commonly require advanced imaging or biopsy. Noninvasive risk stratification tools are highly desirable. We previously developed the BRODERS classifier (Benign versus aggRessive nODule Evaluation using Radiomic Stratification), a conventional predictive radiomic model based on eight imaging features capturing nodule location, shape, size, texture and surface characteristics. Herein we report its external validation using a dataset of incidentally identified lung nodules (Vanderbilt University Lung Nodule Registry) in comparison to the Brock model. Area under the curve (AUC), as well as sensitivity, specificity, negative and positive predictive values were calculated. For the entire Vanderbilt validation set (n=170, 54% malignant), the AUC was 0.87 (95% CI 0.81-0.92) for the Brock model and 0.90 (95% CI 0.85-0.94) for the BRODERS model. Using the optimal cut-off determined by Youden's index, the sensitivity was 92.3%, the specificity was 62.0%, the positive (PPV) and negative predictive values (NPV) were 73.7% and 87.5%, respectively. For nodules with intermediate pre-test probability of malignancy, Brock score of 5-65% (n=97), the sensitivity and specificity were 94% and 46%, respectively, the PPV was 78.4% and the NPV was 79.2%. The BRODERS radiomic predictive model performs well on an independent dataset and may facilitate the management of indeterminate pulmonary nodules.

External validation and comparison of the Brock model and Lung-RADS for the baseline lung cancer CT screening using data from the Korean Lung Cancer Screening Project.

To validate and compare the performance of the Brock model and Lung CT Screening Reporting and Data System (Lung-RADS) on nodules detected by baseline CT screening. We performed a secondary analysis of the Korean Lung Cancer Screening Project (K-LUCAS; ClinicalTrials.gov , NCT03394703), a nationwide, multicenter, prospective cohort study. From April 2017 to December 2018, low-dose CT screening was performed on high-risk subjects. Discrimination and calibration of Brock models 2a and 2b (i.e., full model without and with spiculation, respectively) were assessed, and discrimination was compared with that of Lung-RADS, which utilized subjective assessment categories 2b (b stands for benign) and 4X. Of the 13,150 subjects, 4578 were eligible (median age 62 years; 4458 men; 9929 nodules including 40 lung cancers). Areas under the receiver operating characteristic curve were 0.96 (IQR 0.92-0.99) for Brock model 2a, 0.96 (IQR 0.92-0.99) for Brock model 2b, and 0.95 (IQR 0.91-0.99) for Lung-RADS (p = 0.32 and p = 0.34, respectively). At an equivalent cutoff of 5%, Brock model 2b (sensitivity 87.5% [35/40]; specificity 93.6% [9259/9889]; positive predictive value [PPV] 5.3% [35/665]; negative predictive value [NPV] 99.9% [9259/9264]) and Lung-RADS (sensitivity 87.5% [35/40]; specificity 93.3% [9222/9889]; PPV 5.0% [35/702]; NPV 99.9% [9222/9227]) performed similarly well (all p > 0.05). The calibration performance of both Brock models 2a and 2b was poor (both p < 0.001). Lung-RADS, when reinforced with visual assessment-based categories, has a similar diagnostic performance to the Brock model for baseline CT scans. • Brock model 2b and Lung CT Screening Reporting and Data System (Lung-RADS) demonstrated a similar discrimination performance for lung cancer in the baseline CT screening (areas under the receiver operating characteristic curve 0.96 vs. 0.95; p = 0.34). • When visual assessment-based categories were removed from Lung-RADS, specificity and positive predictive value were lower than those of Brock model 2b (p = 0.001 and p = 0.02, respectively). • The Brock model showed poor calibration (p < 0.001).

Assessment of Integrated Classifier’s Ability to Distinguish Benign From Malignant Lung Nodules: Extended Analyses and 2-Year Follow-Up Results of the PANOPTIC (Pulmonary Nodule Plasma Proteomic Classifier) Trial

Assessment of Integrated Classifier’s Ability to Distinguish Benign From Malignant Lung Nodules: Extended Analyses and 2-Year Follow-Up Results of the PANOPTIC (Pulmonary Nodule Plasma Proteomic Classifier) Trial

A prediction model to evaluate the pretest risk of malignancy in solitary pulmonary nodules: evidence from a large Chinese southwestern population.

Lung cancer is the leading cause of cancer death and there have been clinical prediction models. This study aimed to evaluate the diagnostic performance of published models and create new models to evaluate the probability of malignant solitary pulmonary nodules (SPNs) in Chinese population. We consecutively enrolled 2061 patients with SPNs from West China Hospital between January 2008 and December 2016, each SPN was pathologically confirmed. First, four published prediction models, Mayo clinic model, Veterans Affairs (VA) model, Brock model and People's Hospital of Peking University (PEH) model were validated in our patients. Then, utilizing logistic regression, decision tree and random forest (RF), we developed three new models and internally validated them. Area under the receiver operating characteristic curve (AUC) values of four published models were as follows: Mayo 0.705 (95% CI 0.658-0.752, n = 726), VA 0.64 6 (95% CI 0.598-0.695, n = 800), Brock 0.575 (95% CI 0.502-0.648, n = 550) and PEH 0.675 (95% CI 0.627-0.723, n = 726). Logistic regression model, decision tree model and RF model were developed, AUC values of these models were 0.842 (95% CI 0.778-0.906), 0.734 (95% CI 0.647-0.821), 0.851 (95% CI 0.789-0.914), respectively. The four published lung cancer prediction models do not apply to our population, and we have established new models that can be used to predict the probability of malignant SPNs.

Diagnosis of Invasive Lung Adenocarcinoma Based on Chest CT Radiomic Features of Part-Solid Pulmonary Nodules: A Multicenter Study.

Background Solid components of part-solid nodules (PSNs) at CT are reflective of invasive adenocarcinoma, but studies describing radiomic features of PSNs and the perinodular region are lacking. Purpose To develop and to validate radiomic signatures diagnosing invasive lung adenocarcinoma in PSNs compared with the Brock, clinical-semantic features, and volumetric models. Materials and Methods This retrospective multicenter study (https://ClinicalTrials.gov, NCT03872362) included 291 patients (median age, 60 years; interquartile range, 55-65 years; 191 women) from January 2013 to October 2017 with 297 PSN lung adenocarcinomas split into training (n = 229) and test (n = 68) data sets. Radiomic features were extracted from the different regions (gross tumor volume [GTV], solid, ground-glass, and perinodular). Random-forest models were trained using clinical-semantic, volumetric, and radiomic features, and an online nodule calculator was used to compute the Brock model. Performances of models were evaluated using standard metrics such as area under the curve (AUC), accuracy, and calibration. The integrated discrimination improvement was applied to assess model performance changes after the addition of perinodular features. Results The radiomics model based on ground-glass and solid features yielded an AUC of 0.98 (95% confidence interval [CI]: 0.96, 1.00) on the test data set, which was significantly higher than the Brock (AUC, 0.83 [95% CI: 0.72, 0.94]; P = .007), clinical-semantic (AUC, 0.90 [95% CI: 0.83, 0.98]; P = .03), volumetric GTV (AUC, 0.87 [95% CI: 0.78, 0.96]; P = .008), and radiomics GTV (AUC, 0.88 [95% CI: 0.80, 0.96]; P = .01) models. It also achieved the best accuracy (93% [95% CI: 84%, 98%]). Both this model and the model with added perinodular features showed good calibration, whereas adding perinodular features did not improve the performance (integrated discrimination improvement, -0.02; P = .56). Conclusion Separating ground-glass and solid CT radiomic features of part-solid nodules was useful in diagnosing the invasiveness of lung adenocarcinoma, yielding a better predictive performance than the Brock, clinical-semantic, volumetric, and radiomics gross tumor volume models. Online supplemental material is available for this article. See also the editorial by Nishino in this issue. Published under a CC BY 4.0 license.

Performance of Lung Nodule Management Algorithms for Lung-RADS Category 4 Lesions

To test the performance of the American College of Chest Physicians (ACCP) and British Thoracic Society (BTS) algorithms to stratify high-risk nodules identified at lung cancer screening. Patients with Lung-RADS category 4 nodules identified on lung cancer screening computed tomography (CT) between March 2014 and August 2018 were identified, and a subset of 150 were randomly selected. Nodule characteristics and, if available, fluorodeoxyglucose (FDG) uptake on positron emission tomography (PET)-CT scan were recorded. Radiologists blinded to final diagnosis and downstream testing performed five-point visual assessment score for probability of nodule malignancy; their accuracies are averaged below. Probabilities of malignancy according to Brock and Herder models were calculated. ACCP and BTS algorithms were applied to the nodules. Final diagnosis of malignancy was made in 65/150 (43%) of patients. The sensitivity, specificity and accuracy for nodule malignancy were: radiologist visual score (92%, 85%, 88%); BTS (76%, 91%, 85%); ACCP (63%, 89%, 78%); and Brock calculator (77%, 71%, 73%). The sensitivity, specificity, and accuracy for nodule malignancy in patients with FDG PET-CT scan (n = 78) were: FDG uptake (91%, 64%, 83%); Herder probability (91%, 68%, 83%); radiologist visual score (93%, 69%, 86%); BTS (84%, 64%, 78%); Brock probability (82%, 50%, 72%); and ACCP (68%, 59%, 65%). Thoracic radiologist visual analysis yielded the greatest accuracy for nodule triage in the entire cohort. BTS performed better than ACCP guidelines and both performed better than the Brock model alone.

External validation of a convolutional neural network artificial intelligence tool to predict malignancy in pulmonary nodules

BackgroundEstimation of the risk of malignancy in pulmonary nodules detected by CT is central in clinical management. The use of artificial intelligence (AI) offers an opportunity to improve risk prediction....

Radiologic-Pathologic Correlation for Nondiagnostic CT-Guided Lung Biopsies Performed for the Evaluation of Lung Cancer.

OBJECTIVE. For nondiagnostic CT-guided lung biopsies, we tested whether radiologicpathologic correlation could identify patients who may benefit from repeat biopsy. MATERIALS AND METHODS. In this retrospective study, 1525 lung biopsies were performed between July 2013 and June 2017, 243 of which were nondiagnostic. Of these 243 lung biopsies, 98 were performed to evaluate for lung malignancy; 17 were excluded because of insufficient follow-up, leaving a total of 81 cases. The Brock and Herder models were used to calculate risk; in addition, cases were independently blindly reviewed by two thoracic radiologists who assigned a score from 1 (probably benign) to 5 (probably malignant). The final diagnosis was established by pathology results or benignancy was established if the lesion resolved or remained stable for at least 2 years. RESULTS. Of the 81 nondiagnostic lung biopsies, initial pathology results included 33 cases of inflammation, 28 cases of normal lung tissue or insufficient sample, 10 cases of organizing pneumonia, and 10 cases of atypical cells. 42% (34/81) of cases were eventually determined to be malignant (negative predictive value [NPV] of 58%). Pathology results of organizing pneumonia had the lowest rate of malignancy (2/10 = 20%), and pathology results of atypical cells had the highest rate of malignancy (5/10 = 50%, p = 0.51). Within this highly selected cohort, the Brock and Herder models were not predictive of malignancy, with areas under the ROC curve (AUCs) of 0.52 and 0.52, respectively. Evaluation by thoracic radiologists yielded AUCs of 0.85 and 0.77. When radiologist-assigned scores of 1 and 2 were considered as benign, the NPV was 90% and 95%. CONCLUSION. Review of nondiagnostic lung biopsies for radiologic-pathologic concordance by thoracic radiologists can triage patients who may benefit from repeat biopsy.

Comparison of Veterans Affairs, Mayo, Brock classification models and radiologist diagnosis for classifying the malignancy of pulmonary nodules in Chinese clinical population.

Several classification models based on Western population have been developed to help clinicians to classify the malignancy probability of pulmonary nodules. However, the diagnostic performance of these Western models in Chinese population is unknown. This paper aimed to compare the diagnostic performance of radiologist evaluation of malignancy probability and three classification models (Mayo Clinic, Veterans Affairs, and Brock University) in Chinese clinical pulmonology patients. This single-center retrospective study included clinical patients from Tianjin Medical University Cancer Institute and Hospital with new, CT-detected pulmonary nodules in 2013. Patients with a nodule with diameter of 4-25 mm, and histological diagnosis or 2-year follow-up were included. Analysis of area under the receiver operating characteristic curve (AUC), decision curve analysis (DCA) and threshold of decision analysis was used to evaluate the diagnostic performance of radiologist diagnosis and the three classification models, with histological diagnosis or 2-year follow-up as the reference. In total, 277 patients (286 nodules) were included. Two hundred and seven of 286 nodules (72.4%) in 203 patients were malignant. AUC of the Mayo model (0.77; 95% CI: 0.72-0.82) and Brock model (0.77; 95% CI: 0.72-0.82) were similar to radiologist diagnosis (0.78; 95% CI: 0.73-0.83; P=0.68, P=0.71, respectively). The diagnostic performance of the VA model (AUC: 0.66) was significantly lower than that of radiologist diagnosis (P=0.003). A three-class classifying threshold analysis and DCA showed that the radiologist evaluation had higher discriminatory power for malignancy than the three classification models. In a cohort of Chinese clinical pulmonology patients, radiologist evaluation of lung nodule malignancy probability demonstrated higher diagnostic performance than Mayo, Brock, and VA classification models. To optimize nodule diagnosis and management, a new model with more radiological characteristics could be valuable.

Cancer Risk in Subsolid Nodules in the National Lung Screening Trial.

Background Subsolid pulmonary nodules, comprising pure ground-glass nodules (GGNs) and part-solid nodules (PSNs), have a high risk of indolent malignancy. Lung Imaging Reporting and Data System (Lung-RADS) nodule management guidelines are based on expert opinion and lack independent validation. Purpose To evaluate Lung-RADS estimates of the malignancy rates of subsolid nodules, using nodules from the National Lung Screening Trial (NLST), and to compare Lung-RADS to the NELSON trial classification as well as the Brock University calculator. Materials and Methods Subsets of GGNs and PSNs were selected from the NLST for this retrospective study. A thoracic radiologist reviewed the baseline and follow-up CT images, confirmed that they were true subsolid nodules, and measured the nodules. The primary outcome for each nodule was the development of malignancy within the follow-up period (median, 6.5 years). Nodules were stratified according to Lung-RADS, NELSON trial criteria, and the Brock model. For analyses, nodule subsets were weighted on the basis of frequency in the NLST data set. Nodule stratification models were tested by using receiver operating characteristic curves. Results A total of 622 nodules were evaluated, of which 434 nodules were subsolid. At baseline, 304 nodules were classified as Lung-RADS category 2, with a malignancy rate of 3%, which is greater than the 1% in Lung-RADS (P = .004). The malignancy rate for GGNs smaller than 10 mm (two of 129, 1.3%) was smaller than that for GGNs measuring 10-19 mm (11 of 153, 6%) (P = .01). The malignancy rate for Lung-RADS category 3 was 14% (13 of 67), which is greater than the reported 2% in Lung-RADS (P < .001). The Brock model predicted malignancy better than Lung-RADS and the NELSON trial scheme (area under the receiver operating characteristic curve = 0.78, 0.70, and 0.67, respectively; P = .02 for Brock model vs NELSON trial scheme). Conclusion Subsolid nodules classified as Lung Imaging Reporting and Data System (Lung-RADS) categories 2 and 3 have a higher risk of malignancy than reported. The Brock risk calculator performed better than measurement-based classification schemes such as Lung-RADS. © RSNA, 2019 Online supplemental material is available for this article. See also the editorial by Kauczor and von Stackelberg in this issue.

Pediatric Decision Making: Ross, Rawls, and Getting Children and Families Right

What process ought to guide decision making for pediatric patients? The prevailing view is that decision making should be informed and guided by the best interest of the child. A widely discussed structural model proposed by Buchanan and Brock focuses on parents as surrogate decision makers and examines best interests as guiding and-or intervention principles. Working from two recent articles by Ross on "constrained parental autonomy" in pediatric decision making (which is grounded in the Buchanan and Brock model), I discuss (supportively) features of Ross's effort vis-a-vis the best interest standard. I argue that any pediatric decision-making model that brackets or formally limits an engagement with the child patient assumes too much. Further, any model that under appreciates the place of parents and their autonomy, and the dynamic parent-child relationship, misses an opportunity to broaden the clinical encounter by considering questions of justice for the child (Rawls) and within a family (Ross). In this context, I focus on the child's emerging and ongoing emotional and intellectual development and autonomy¬-their capabilities and identifying primary goods.

External validation and recalibration of the Brock model to predict probability of cancer in pulmonary nodules using NLST data

IntroductionWe performed an external validation of the Brock model using the National Lung Screening Trial (NLST) data set, following strict guidelines set forth by the Transparent Reporting of a multivariable...