Enhancing a deep learning model for pulmonary nodule malignancy risk estimation in chest CT with uncertainty estimation.

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.

Prediction Variability to Identify Reduced AI Performance in Cancer Diagnosis at MRI and CT.

Background A priori identification of patients at risk of artificial intelligence (AI) failure in diagnosing cancer would contribute to the safer clinical integration of diagnostic algorithms. Purpose To evaluate AI prediction variability as an uncertainty quantification (UQ) metric for identifying cases at risk of AI failure in diagnosing cancer at MRI and CT across different cancer types, data sets, and algorithms. Materials and Methods Multicenter data sets and publicly available AI algorithms from three previous studies that evaluated detection of pancreatic cancer on contrast-enhanced CT images, detection of prostate cancer on MRI scans, and prediction of pulmonary nodule malignancy on low-dose CT images were analyzed retrospectively. Each task's algorithm was extended to generate an uncertainty score based on ensemble prediction variability. AI accuracy percentage and partial area under the receiver operating characteristic curve (pAUC) were compared between certain and uncertain patient groups in a range of percentile thresholds (10%-90%) for the uncertainty score using permutation tests for statistical significance. The pulmonary nodule malignancy prediction algorithm was compared with 11 clinical readers for the certain group (CG) and uncertain group (UG). Results In total, 18 022 images were used for training and 838 images were used for testing. AI diagnostic accuracy was higher for the cases in the CG across all tasks (P < .001). At an 80% threshold of certain predictions, accuracy in the CG was 21%-29% higher than in the UG and 4%-6% higher than in the overall test data sets. The lesion-level pAUC in the CG was 0.25-0.39 higher than in the UG and 0.05-0.08 higher than in the overall test data sets (P < .001). For pulmonary nodule malignancy prediction, accuracy of AI was on par with clinicians for cases in the CG (AI results vs clinician results, 80% [95% CI: 76, 85] vs 78% [95% CI: 70, 87]; P = .07) but worse for cases in the UG (AI results vs clinician results, 50% [95% CI: 37, 64] vs 68% [95% CI: 60, 76]; P < .001). Conclusion An AI-prediction UQ metric consistently identified reduced performance of AI in cancer diagnosis. © RSNA, 2023 Supplemental material is available for this article. See also the editorial by Babyn in this issue.

Prior CT Improves Deep Learning for Malignancy Risk Estimation of Screening-detected Pulmonary Nodules.

Background Prior chest CT provides valuable temporal information (eg, changes in nodule size or appearance) to accurately estimate malignancy risk. Purpose To develop a deep learning (DL) algorithm that uses a current and prior low-dose CT examination to estimate 3-year malignancy risk of pulmonary nodules. Materials and Methods In this retrospective study, the algorithm was trained using National Lung Screening Trial data (collected from 2002 to 2004), wherein patients were imaged at most 2 years apart, and evaluated with two external test sets from the Danish Lung Cancer Screening Trial (DLCST) and the Multicentric Italian Lung Detection Trial (MILD), collected in 2004-2010 and 2005-2014, respectively. Performance was evaluated using area under the receiver operating characteristic curve (AUC) on cancer-enriched subsets with size-matched benign nodules imaged 1 and 2 years apart from DLCST and MILD, respectively. The algorithm was compared with a validated DL algorithm that only processed a single CT examination and the Pan-Canadian Early Lung Cancer Detection Study (PanCan) model. Results The training set included 10 508 nodules (422 malignant) in 4902 trial participants (mean age, 64 years ± 5 [SD]; 2778 men). The size-matched external test sets included 129 nodules (43 malignant) and 126 nodules (42 malignant). The algorithm achieved AUCs of 0.91 (95% CI: 0.85, 0.97) and 0.94 (95% CI: 0.89, 0.98). It significantly outperformed the DL algorithm that only processed a single CT examination (AUC, 0.85 [95% CI: 0.78, 0.92; P = .002]; and AUC, 0.89 [95% CI: 0.84, 0.95; P = .01]) and the PanCan model (AUC, 0.64 [95% CI: 0.53, 0.74; P < .001]; and AUC, 0.63 [95% CI: 0.52, 0.74; P < .001]). Conclusion A DL algorithm using current and prior low-dose CT examinations was more effective at estimating 3-year malignancy risk of pulmonary nodules than established models that only use a single CT examination. Clinical trial registration nos. NCT00047385, NCT00496977, NCT02837809 © RSNA, 2023 Supplemental material is available for this article. See also the editorial by Horst and Nishino in this issue.

Multi-source data approach for personalized outcome prediction in lung cancer screening: update from the NELSON trial.

Trials show that low-dose computed tomography (CT) lung cancer screening in long-term (ex-)smokers reduces lung cancer mortality. However, many individuals were exposed to unnecessary diagnostic procedures. This project aims to improve the efficiency of lung cancer screening by identifying high-risk participants, and improving risk discrimination for nodules. This study is an extension of the Dutch-Belgian Randomized Lung Cancer Screening Trial, with a focus on personalized outcome prediction (NELSON-POP). New data will be added on genetics, air pollution, malignancy risk for lung nodules, and CT biomarkers beyond lung nodules (emphysema, coronary calcification, bone density, vertebral height and body composition). The roles of polygenic risk scores and air pollution in screen-detected lung cancer diagnosis and survival will be established. The association between the AI-based nodule malignancy score and lung cancer will be evaluated at baseline and incident screening rounds. The association of chest CT imaging biomarkers with outcomes will be established. Based on these results, multisource prediction models for pre-screening and post-baseline-screening participant selection and nodule management will be developed. The new models will be externally validated. We hypothesize that we can identify 15-20% participants with low-risk of lung cancer or short life expectancy and thus prevent ~140,000 Dutch individuals from being screened unnecessarily. We hypothesize that our models will improve the specificity of nodule management by 10% without loss of sensitivity as compared to assessment of nodule size/growth alone, and reduce unnecessary work-up by 40-50%.

Deep Learning for Malignancy Risk Estimation of Pulmonary Nodules Detected at Low-Dose Screening CT.

Background Accurate estimation of the malignancy risk of pulmonary nodules at chest CT is crucial for optimizing management in lung cancer screening. Purpose To develop and validate a deep learning (DL) algorithm for malignancy risk estimation of pulmonary nodules detected at screening CT. Materials and Methods In this retrospective study, the DL algorithm was developed with 16 077 nodules (1249 malignant) collected -between 2002 and 2004 from the National Lung Screening Trial. External validation was performed in the following three -cohorts -collected between 2004 and 2010 from the Danish Lung Cancer Screening Trial: a full cohort containing all 883 nodules (65 -malignant) and two cancer-enriched cohorts with size matching (175 nodules, 59 malignant) and without size matching (177 -nodules, 59 malignant) of benign nodules selected at random. Algorithm performance was measured by using the area under the receiver operating characteristic curve (AUC) and compared with that of the Pan-Canadian Early Detection of Lung Cancer (PanCan) model in the full cohort and a group of 11 clinicians composed of four thoracic radiologists, five radiology residents, and two pulmonologists in the cancer-enriched cohorts. Results The DL algorithm significantly outperformed the PanCan model in the full cohort (AUC, 0.93 [95% CI: 0.89, 0.96] vs 0.90 [95% CI: 0.86, 0.93]; P = .046). The algorithm performed comparably to thoracic radiologists in cancer-enriched cohorts with both random benign nodules (AUC, 0.96 [95% CI: 0.93, 0.99] vs 0.90 [95% CI: 0.81, 0.98]; P = .11) and size-matched benign nodules (AUC, 0.86 [95% CI: 0.80, 0.91] vs 0.82 [95% CI: 0.74, 0.89]; P = .26). Conclusion The deep learning algorithm showed excellent performance, comparable to thoracic radiologists, for malignancy risk estimation of pulmonary nodules detected at screening CT. This algorithm has the potential to provide reliable and reproducible malignancy risk scores for clinicians, which may help optimize management in lung cancer screening. © RSNA, 2021 Online supplemental material is available for this article. See also the editorial by Tammemägi in this issue.

Increased respiratory morbidity in individuals with interstitial lung abnormalities.

BACKGROUND: Interstitial lung abnormalities (ILA) are common in participants of lung cancer screening trials and broad population-based cohorts. They are associated with increased mortality, but less is known about disease specific morbidity and healthcare utilisation in individuals with ILA. METHODS: We included all participants from the screening arm of the Danish Lung Cancer Screening Trial with available baseline CT scan data (n = 1990) in this cohort study. The baseline scan was scored for the presence of ILA and patients were followed for up to 12 years. Data about all hospital admissions, primary healthcare visits and medicine prescriptions were collected from the Danish National Health Registries and used to determine the participants' disease specific morbidity and healthcare utilisation using Cox proportional hazards models. RESULTS: The 332 (16.7%) participants with ILA were more likely to be diagnosed with one of several respiratory diseases, including interstitial lung disease (HR: 4.9, 95% CI: 1.8-13.3, p = 0.008), COPD (HR: 1.7, 95% CI: 1.2-2.3, p = 0.01), pneumonia (HR: 2.0, 95% CI: 1.4-2.7, p <  0.001), lung cancer (HR: 2.7, 95% CI: 1.8-4.0, p <  0.001) and respiratory failure (HR: 1.8, 95% CI: 1.1-3.0, p = 0.03) compared with participants without ILA. These findings were confirmed by increased hospital admission rates with these diagnoses and more frequent prescriptions for inhalation medicine and antibiotics in participants with ILA. CONCLUSIONS: Individuals with ILA are more likely to receive a diagnosis and treatment for several respiratory diseases, including interstitial lung disease, COPD, pneumonia, lung cancer and respiratory failure during long-term follow-up.

Ensuring Basic Competence in Thoracentesis.

BACKGROUND: Trocar pigtail catheter thoracentesis (TPCT) is a common procedure often performed by junior physicians. Simulation-based training may effectively train physicians in the procedure prior to performing it on patients. An assessment tool with solid validity evidence is necessary to ensure sufficient procedural competence. OBJECTIVES: Our study objectives were (1) to collect evidence of validity for a newly developed pigtail catheter assessment tool (Thoracentesis Assessment Tool [ThorAT]) developed for the evaluation of TPCT performance and (2) to establish a pass/fail score for summative assessment. METHODS: We assessed the validity evidence for the ThorAT using the recommended framework for validity by Messick. Thirty-four participants completed two consecutive procedures and their performance was assessed by two blinded, independent raters using the ThorAT. We compared performance scores to test whether the assessment tool was able to discern between the two groups, and a pass/fail score was established. RESULTS: The assessment tool was able to discriminate between the two groups in terms of competence level. Experienced physicians received significantly higher test scores than novices in both the first and second procedure. A pass/fail score of 25.2 points was established, resulting in 4 (17%) passing novices and 1 (9%) failing experienced participant in the first procedure. In the second procedure 9 (39%) novices passed and 2 (18%) experienced participants failed. CONCLUSIONS: This study provides a tool for summative assessment of competence in TPCT. Strong validity evidence was gathered from five sources of evidence. A simulation-based training program using the ThorAT could ensure competence before performing thoracentesis on patients.

Implementation of lung cancer CT screening in the Nordic countries.

INTRODUCTION: We review the current knowledge of CT screening for lung cancer and present an expert-based, joint protocol for the proper implementation of screening in the Nordic countries. MATERIALS AND METHODS: Experts representing all the Nordic countries performed literature review and concensus for a joint protocol for lung cancer screening. RESULTS AND DISCUSSION: Areas of concern and caution are presented and discussed. We suggest to perform CT screening pilot studies in the Nordic countries in order to gain experience and develop specific and safe protocols for the implementation of such a program.

Results of the Randomized Danish Lung Cancer Screening Trial with Focus on High-Risk Profiling.

RATIONALE: As of April 2015, participants in the Danish Lung Cancer Screening Trial had been followed for at least 5 years since their last screening. OBJECTIVES: Mortality, causes of death, and lung cancer findings are reported to explore the effect of computed tomography (CT) screening. METHODS: A total of 4,104 participants aged 50-70 years at the time of inclusion and with a minimum 20 pack-years of smoking were randomized to have five annual low-dose CT scans (study group) or no screening (control group). MEASUREMENTS AND MAIN RESULTS: Follow-up information regarding date and cause of death, lung cancer diagnosis, cancer stage, and histology was obtained from national registries. No differences between the two groups in lung cancer mortality (hazard ratio, 1.03; 95% confidence interval, 0.66-1.6; P = 0.888) or all-cause mortality (hazard ratio, 1.02; 95% confidence interval, 0.82-1.27; P = 0.867) were observed. More cancers were found in the screening group than in the no-screening group (100 vs. 53, respectively; P < 0.001), particularly adenocarcinomas (58 vs. 18, respectively; P < 0.001). More early-stage cancers (stages I and II, 54 vs. 10, respectively; P < 0.001) and stage IIIa cancers (15 vs. 3, respectively; P = 0.009) were found in the screening group than in the control group. Stage IV cancers were nonsignificantly more frequent in the control group than in the screening group (32 vs. 23, respectively; P = 0.278). For the highest-stage cancers (T4N3M1, 21 vs. 8, respectively; P = 0.025), this difference was statistically significant, indicating an absolute stage shift. Older participants, those with chronic obstructive pulmonary disease, and those with more than 35 pack-years of smoking had a significantly increased risk of death due to lung cancer, with nonsignificantly fewer deaths in the screening group. CONCLUSIONS: No statistically significant effects of CT screening on lung cancer mortality were found, but the results of post hoc high-risk subgroup analyses showed nonsignificant trends that seem to be in good agreement with the results of the National Lung Screening Trial. Clinical trial registered with www.clinicaltrials.gov (NCT00496977).

Ground-Glass Opacity Lung Nodules in the Era of Lung Cancer CT Screening: Radiology, Pathology, and Clinical Management.

The advent of computed tomography screening for lung cancer will increase the incidence of ground-glass opacity (GGO) nodules detected and referred for diagnostic evaluation and management. GGO nodules remain a diagnostic challenge; therefore, a more systematic approach is necessary to ensure correct diagnosis and optimal management. Here we present the latest advances in the radiologic imaging and pathology of GGO nodules, demonstrating that radiologic features are increasingly predictive of the pathology of GGO nodules. We review the current guidelines from the Fleischner Society, the National Comprehensive Cancer Network, and the British Thoracic Society. In addition, we discuss the management and follow-up of GGO nodules in the light of experience from screening trials. Minimally invasive tissue biopsies and the marking of GGO nodules for surgery are new and rapidly developing fields that will yield improvements in both diagnosis and treatment. The standard-of-care surgical treatment of early lung cancer is still minimally invasive lobectomy with systematic lymph node dissection. However, recent research has shown that some GGO lesions may be treated with sublobar resections; these findings may expand the surgical treatment options available in the future.

Predictive Accuracy of the PanCan Lung Cancer Risk Prediction Model -External Validation based on CT from the Danish Lung Cancer Screening Trial.

OBJECTIVES: Lung cancer risk models should be externally validated to test generalizability and clinical usefulness. The Danish Lung Cancer Screening Trial (DLCST) is a population-based prospective cohort study, used to assess the discriminative performances of the PanCan models. METHODS: From the DLCST database, 1,152 nodules from 718 participants were included. Parsimonious and full PanCan risk prediction models were applied to DLCST data, and also coefficients of the model were recalculated using DLCST data. Receiver operating characteristics (ROC) curves and area under the curve (AUC) were used to evaluate risk discrimination. RESULTS: AUCs of 0.826-0.870 were found for DLCST data based on PanCan risk prediction models. In the DLCST, age and family history were significant predictors (p = 0.001 and p = 0.013). Female sex was not confirmed to be associated with higher risk of lung cancer; in fact opposing effects of sex were observed in the two cohorts. Thus, female sex appeared to lower the risk (p = 0.047 and p = 0.040) in the DLCST. CONCLUSIONS: High risk discrimination was validated in the DLCST cohort, mainly determined by nodule size. Age and family history of lung cancer were significant predictors and could be included in the parsimonious model. Sex appears to be a less useful predictor. KEY POINTS: • High accuracy in logistic modelling for lung cancer risk stratification of nodules. • Lung cancer risk prediction is primarily based on size of pulmonary nodules. • Nodule spiculation, age and family history of lung cancer are significant predictors. • Sex does not appear to be a useful risk predictor.

Smoking habits in the randomised Danish Lung Cancer Screening Trial with low-dose CT: final results after a 5-year screening programme.

BACKGROUND: We present the final results of the effect of lung cancer screening with low-dose CT on the smoking habits of participants in a 5-year screening trial. METHODS: The Danish Lung Cancer Screening Trial (DLCST) was a 5-year screening trial that enrolled 4104 subjects; 2052 were randomised to annual low-dose CT (CT group) and 2052 received no intervention (control group). Participants were current and ex-smokers (≥4 weeks abstinence from smoking) with a tobacco consumption of ≥20 pack years. Smoking habits were determined annually. Missing values for smoking status at the final screening round were handled using two different models. RESULTS: There were no statistically significant differences in annual smoking status between the CT group and control group. Overall the ex-smoker rates (CT + control group) significantly increased from 24% (baseline) to 37% at year 5 of screening (p<0.001). The annual point prevalence quit rate increased from 11% to 24% during the five screening rounds; the ex-smokers' relapse rate remained stable, around 11%, across the same period. CONCLUSIONS: Screening with low-dose CT had no extra effect on smoking status compared with the control group, but overall the screening programme probably promoted smoking cessation. CLINICAL TRIAL REGISTRATION: The DLCST is registered in Clinical Trials.gov Protocol Registration System (identification no. NCT00496977).

Management of incidental pulmonary nodules in CT examination.

Lung cancer is the leading cause of cancer-related death in Denmark and the world. The increase in CT examinations has led to an increase in detection of pulmonary nodules divided into solid and subsolid (including ground glass and part solid). Risk factors for malignancy include age, smoking, female gender, and specific ethnicities. Nodule traits like size, spiculation, upper-lobe location, and emphysema correlate with higher malignancy risk. Managing these potentially malignant nodules relies on evidence-based guidelines and risk stratification. These risk stratification models can standardize the approach for the management of incidental pulmonary findings, as argued in this review.

The Validity of a New Procedure-Based Definition of Cancer Status in Patients with Breast-, Lung- and Colorectal Cancer in the Danish National Patient Registry.

Background/Aim: The Danish National Patient Registry (DNPR) provides unique epidemiological insight, but often lacks granular data. We propose a procedure-based definition of cancer status in patients with breast-, lung- and colorectal cancer, which can be applied to administrative health databases. New definitions of cancer status are needed as mortality and morbidity are closely linked to cancer status, yet most studies only use duration since cancer diagnosis as a severity marker. The aim of the study was to validate a new pragmatic definition. Methods: Medical journals of 600 patients, with breast-, lung- and colorectal cancer from the Department of Oncology at Herlev-Gentofte Hospital were retrospectively reviewed. We defined active cancer as a cancer diagnosis, not followed by a potentially curative procedure within 6 months of the diagnosis. The remaining patients were characterized as having non-active cancer. This dichotomization was then compared to a cancer status assessment based on treatment received and paraclinical test such as their first post-procedural control scan. Based on this comparison, we calculated the positive predictive value (PPV) of our definitions of active and non-active cancer. Results: The calculated PPVs for active breast-, lung- and colorectal cancer were 87% (CI 95%: 0.74-0.99), 91% (CI 95%: 0.87-0.96) and 82% (CI 95%: 0.73-0.91). The PPVs for non-active breast-, lung- and colorectal cancer were 95% (CI 95%: 0.92-0.99), 91% (CI 95%: 0.82-0.99) and 73% (CI 95%: 0.66-0.81), respectively. Conclusion: We found an overall high PPV for both active and non-active cancer across all three types of cancer.

CT screening for lung cancer brings forward early disease. The randomised Danish Lung Cancer Screening Trial: status after five annual screening rounds with low-dose CT.

BACKGROUND: The effects of low-dose CT screening on disease stage shift, mortality and overdiagnosis are unclear. Lung cancer findings and mortality rates are reported at the end of screening in the Danish Lung Cancer Screening Trial. METHODS: 4104 men and women, healthy heavy smokers/former smokers were randomised to five annual low-dose CT screenings or no screening. Two experienced chest radiologists read all CT scans and registered the location, size and morphology of nodules. Nodules between 5 and 15 mm without benign characteristics were rescanned after 3 months. Growing nodules (>25% volume increase and/or volume doubling time<400 days) and nodules >15 mm were referred for diagnostic workup. In the control group, lung cancers were diagnosed and treated outside the study by the usual clinical practice. RESULTS: Participation rates were high in both groups (screening: 95.5%; control: 93.0%; p<0.001). Lung cancer detection rate was 0.83% at baseline and mean annual detection rate was 0.67% at incidence rounds (p=0.535). More lung cancers were diagnosed in the screening group (69 vs. 24, p<0.001), and more were low stage (48 vs 21 stage I-IIB non-small cell lung cancer (NSCLC) and limited stage small cell lung cancer (SCLC), p=0.002), whereas frequencies of high-stage lung cancer were the same (21 vs 16 stage IIIA-IV NSCLC and extensive stage SCLC, p=0.509). At the end of screening, 61 patients died in the screening group and 42 in the control group (p=0.059). 15 and 11 died of lung cancer, respectively (p=0.428). CONCLUSION: CT screening for lung cancer brings forward early disease, and at this point no stage shift or reduction in mortality was observed. More lung cancers were diagnosed in the screening group, indicating some degree of overdiagnosis and need for longer follow-up.

Risk of Malignancy in Patients with Asthma-COPD Overlap Compared to Patients with COPD without Asthma.

Chronic inflammation such as asthma may lead to higher risks of malignancy, which may be inhibited by anti-inflammatory medicine such as inhaled corticosteroids (ICS). The aim of this study was to evaluate if patients with asthma-Chronic Obstructive Pulmonary Disease (COPD) overlap have a higher risk of malignancy than patients with COPD without asthma, and, secondarily, if inhaled corticosteroids modify such a risk in a nationwide multi-center retrospective cohort study of Danish COPD-outpatients with or without asthma. Patients with asthma-COPD overlap were propensity score matched (PSM) 1:2 to patients with COPD without asthma. The endpoint was cancer diagnosis within 2 years. Patients were stratified depending on prior malignancy within 5 years. ICS was explored as a possible risk modifier. We included 50,897 outpatients with COPD; 88% without prior malignancy and 20% with asthma. In the PSM cohorts, 26,003 patients without prior malignancy and 3331 patients with prior malignancy were analyzed. There was no association between asthma-COPD overlap and cancer with hazard ratio (HR) = 0.92, CI = 0.78-1.08, p = 0.31 (no prior malignancy) and HR = 1.04, CI = 0.85-1.26, and p = 0.74 (prior malignancy) as compared to patients with COPD without asthma. ICS did not seem to modify the risk of cancer. In conclusion, in our study, asthma-COPD overlap was not associated with an increased risk of cancer events.

Intrapleural fibrinolysis and DNase versus video-assisted thoracic surgery (VATS) for the treatment of pleural empyema (FIVERVATS): protocol for a randomised, controlled trial - surgery as first-line treatment.

INTRODUCTION: Pleural empyema is a frequent disease with a high morbidity and mortality. Current standard treatment includes antibiotics and thoracic ultrasound (TUS)-guided pigtail drainage. Simultaneously with drainage, an intrapleural fibrinolyticum can be given. A potential better alternative is surgery in terms of video-assisted thoracoscopic surgery (VATS) as first-line treatment. The aim of this study is to determine the difference in outcome in patients diagnosed with complex parapneumonic effusion (stage II) and pleural empyema (stage III) who are treated with either VATS surgery or TUS-guided drainage and intrapleural therapy (fibrinolytic (Alteplase) with DNase (Pulmozyme)) as first-line treatment. METHODS AND ANALYSIS: A national, multicentre randomised, controlled study. Totally, 184 patients with a newly diagnosed community acquired complicated parapneumonic effusion or pleural empyema are randomised to either (1) VATS procedure with drainage or (2) TUS-guided pigtail catheter placement and intrapleural therapy with Actilyse and DNase. The total follow-up period is 12 months. The primary endpoint is length of hospital stay and secondary endpoints include for example, mortality, need for additional interventions, consumption of analgesia and quality of life. ETHICS AND DISSEMINATION: All patients provide informed consent before randomisation. The research project is carried out in accordance with the Helsinki II Declaration, European regulations and Good Clinical Practice Guidelines. The Scientific Ethics Committees for Denmark and the Danish Data Protection Agency have provided permission. Information about the subjects is protected under the Personal Data Processing Act and the Health Act. The trial is registered at www. CLINICALTRIALS: gov, and monitored by the regional Good clinical practice monitoring unit. The results of this study will be published in peer-reviewed journals and presented at various national and international conferences. TRIAL REGISTRATION NUMBER: NCT04095676.

Assessment of fully automatic segmentation of pulmonary artery and aorta on noncontrast CT with optimal surface graph cuts.

PURPOSE: Accurate segmentation of the pulmonary arteries and aorta is important due to the association of the diameter and the shape of these vessels with several cardiovascular diseases and with the risk of exacerbations and death in patients with chronic obstructive pulmonary disease. We propose a fully automatic method based on an optimal surface graph-cut algorithm to quantify the full 3D shape and the diameters of the pulmonary arteries and aorta in noncontrast computed tomography (CT) scans. METHODS: The proposed algorithm first extracts seed points in the right and left pulmonary arteries, the pulmonary trunk, and the ascending and descending aorta by using multi-atlas registration. Subsequently, the centerlines of the pulmonary arteries and aorta are extracted by a minimum cost path tracking between the extracted seed points, with a cost based on a combination of lumen intensity similarity and multiscale medialness in three planes. The centerlines are refined by applying the path tracking algorithm to curved multiplanar reformatted scans and are then smoothed and dilated nonuniformly according to the extracted local vessel radius from the medialness filter. The resulting coarse estimates of the vessels are used as initialization for a graph-cut segmentation. Once the vessels are segmented, the diameters of the pulmonary artery (PA) and the ascending aorta (AA) and the P A : A A ratio are automatically calculated both in a single axial slice and in a 10 mm volume around the automatically extracted PA bifurcation level. The method is evaluated on noncontrast CT scans from the Danish Lung Cancer Screening Trial (DLCST). Segmentation accuracy is determined by comparing with manual annotations on 25 CT scans. Intraclass correlation (ICC) between manual and automatic diameters, both measured in axial slices at the PA bifurcation level, is computed on an additional 200 CT scans. Repeatability of the automated 3D volumetric diameter and P A : A A ratio calculations (perpendicular to the vessel axis) are evaluated on 118 scan-rescan pairs with an average in-between time of 3 months. RESULTS: We obtained a Dice segmentation overlap of 0.94 ± 0.02 for pulmonary arteries and 0.96 ± 0.01 for the aorta, with a mean surface distance of 0.62 ± 0.33 mm and 0.43 ± 0.07 mm, respectively. ICC between manual and automatic in-slice diameter measures was 0.92 for PA, 0.97 for AA, and 0.90 for the P A : A A ratio, and for automatic diameters in 3D volumes around the PA bifurcation level between scan and rescan was 0.89, 0.95, and 0.86, respectively. CONCLUSION: The proposed automatic segmentation method can reliably extract diameters of the large arteries in non-ECG-gated noncontrast CT scans such as are acquired in lung cancer screening.

Automatic airway segmentation from computed tomography using robust and efficient 3-D convolutional neural networks.

This paper presents a fully automatic and end-to-end optimised airway segmentation method for thoracic computed tomography, based on the U-Net architecture. We use a simple and low-memory 3D U-Net as backbone, which allows the method to process large 3D image patches, often comprising full lungs, in a single pass through the network. This makes the method simple, robust and efficient. We validated the proposed method on three datasets with very different characteristics and various airway abnormalities: (1) a dataset of pediatric patients including subjects with cystic fibrosis, (2) a subset of the Danish Lung Cancer Screening Trial, including subjects with chronic obstructive pulmonary disease, and (3) the EXACT'09 public dataset. We compared our method with other state-of-the-art airway segmentation methods, including relevant learning-based methods in the literature evaluated on the EXACT'09 data. We show that our method can extract highly complete airway trees with few false positive errors, on scans from both healthy and diseased subjects, and also that the method generalizes well across different datasets. On the EXACT'09 test set, our method achieved the second highest sensitivity score among all methods that reported good specificity.