Diagnostic performance and prognostic value of CT-defined visceral pleural invasion in early-stage lung adenocarcinomas.

OBJECTIVES: To analyze the diagnostic performance and prognostic value of CT-defined visceral pleural invasion (CT-VPI) in early-stage lung adenocarcinomas. METHODS: Among patients with clinical stage I lung adenocarcinomas, half of patients were randomly selected for a diagnostic study, in which five thoracic radiologists determined the presence of CT-VPI. Probabilities for CT-VPI were obtained using deep learning (DL). Areas under the receiver operating characteristic curve (AUCs) and binary diagnostic measures were calculated and compared. Inter-rater agreement was assessed. For all patients, the prognostic value of CT-VPI by two radiologists and DL (using high-sensitivity and high-specificity cutoffs) was investigated using Cox regression. RESULTS: In 681 patients (median age, 65 years [interquartile range, 58-71]; 382 women), pathologic VPI was positive in 130 patients. For the diagnostic study (n = 339), the pooled AUC of five radiologists was similar to that of DL (0.78 vs. 0.79; p = 0.76). The binary diagnostic performance of radiologists was variable (sensitivity, 45.3-71.9%; specificity, 71.6-88.7%). Inter-rater agreement was moderate (weighted Fleiss κ, 0.51; 95%CI: 0.43-0.55). For overall survival (n = 680), CT-VPI by radiologists (adjusted hazard ratio [HR], 1.27 and 0.99; 95%CI: 0.84-1.92 and 0.63-1.56; p = 0.26 and 0.97) or DL (HR, 1.44 and 1.06; 95%CI: 0.86-2.42 and 0.67-1.68; p = 0.17 and 0.80) was not prognostic. CT-VPI by an attending radiologist was prognostic only in radiologically solid tumors (HR, 1.82; 95%CI: 1.07-3.07; p = 0.03). CONCLUSION: The diagnostic performance and prognostic value of CT-VPI are limited in clinical stage I lung adenocarcinomas. This feature may be applied for radiologically solid tumors, but substantial reader variability should be overcome. CLINICAL RELEVANCE STATEMENT: Although the diagnostic performance and prognostic value of CT-VPI are limited in clinical stage I lung adenocarcinomas, this parameter may be applied for radiologically solid tumors with appropriate caution regarding inter-reader variability. KEY POINTS: • Use of CT-defined visceral pleural invasion in clinical staging should be cautious, because prognostic value of CT-defined visceral pleural invasion remains unexplored. • Diagnostic performance and prognostic value of CT-defined visceral pleural invasion varied among radiologists and deep learning. • Role of CT-defined visceral pleural invasion in clinical staging may be limited to radiologically solid tumors.

Artificial Intelligence for Assessment of Endotracheal Tube Position on Chest Radiographs: Validation in Patients From Two Institutions.

BACKGROUND. Timely and accurate interpretation of chest radiographs obtained to evaluate endotracheal tube (ETT) position is important for facilitating prompt adjustment if needed. OBJECTIVE. The purpose of our study was to evaluate the performance of a deep learning (DL)-based artificial intelligence (AI) system for detecting ETT presence and position on chest radiographs in three patient samples from two different institutions. METHODS. This retrospective study included 539 chest radiographs obtained immediately after ETT insertion from January 1 to March 31, 2020, in 505 patients (293 men, 212 women; mean age, 63 years) from institution A (sample A); 637 chest radiographs obtained from January 1 to January 3, 2020, in 302 patients (157 men, 145 women; mean age, 66 years) in the ICU (with or without an ETT) from institution A (sample B); and 546 chest radiographs obtained from January 1 to January 20, 2020, in 83 patients (54 men, 29 women; mean age, 70 years) in the ICU (with or without an ETT) from institution B (sample C). A commercial DL-based AI system was used to identify ETT presence and measure ETT tip-to-carina distance (TCD). The reference standard for proper ETT position was TCD between greater than 3 cm and less than 7 cm, determined by human readers. Critical ETT position was separately defined as ETT tip below the carina or TCD of 1 cm or less. ROC analysis was performed. RESULTS. AI had sensitivity and specificity for identification of ETT presence of 100.0% and 98.7% (sample B) and 99.2% and 94.5% (sample C). AI had sensitivity and specificity for identification of improper ETT position of 72.5% and 92.0% (sample A), 78.9% and 100.0% (sample B), and 83.7% and 99.1% (sample C). At a threshold y-axis TCD of 2 cm or less, AI had sensitivity and specificity for critical ETT position of 100.0% and 96.7% (sample A), 100.0% and 100.0% (sample B), and 100.0% and 99.2% (sample C). CONCLUSION. AI identified improperly positioned ETTs on chest radiographs obtained after ETT insertion as well as on chest radiographs obtained of patients in the ICU at two institutions. CLINICAL IMPACT. Automated AI identification of improper ETT position on chest radiographs may allow earlier repositioning and thereby reduce complications.

Usual Interstitial Pneumonia: Associations With Complications After Percutaneous Transthoracic Needle Lung Biopsy.

BACKGROUND. Changes in lung parenchyma elasticity in usual interstitial pneumonia (UIP) may increase the risk for complications after percutaneous transthoracic needle biopsy (PTNB) of the lung. OBJECTIVE. The purpose of this article was to investigate the association of UIP findings on CT with complications after PTNB, including pneumothorax, pneumothorax requiring chest tube insertion, and hemoptysis. METHODS. This retrospective single-center study included 4187 patients (mean age, 63.8 ± 11.9 [SD] years; 2513 men, 1674 women) who underwent PTNB between January 2010 and December 2015. Patients were categorized into a UIP group and non-UIP group by review of preprocedural CT. In the UIP group, procedural CT images were reviewed to assess for traversal of UIP findings by needle. Multivariable logistic regression analyses were performed to identify associations between the UIP group and needle traversal with postbiopsy complications, controlling for a range of patient, lesion, and procedural characteristics. RESULTS. The UIP and non-UIP groups included 148 and 4039 patients, respectively; in the UIP group, traversal of UIP findings by needle was observed in 53 patients and not observed in 95 patients. The UIP group, in comparison with the non-UIP group, had a higher frequency of pneumothorax (35.1% vs 17.9%, p < .001) and pneumothorax requiring chest tube placement (6.1% vs 1.5%, p = .001) and lower frequency of hemoptysis (2.0% vs 6.1%, p = .03). In multivariable analyses, the UIP group with traversal of UIP findings by needle, relative to the non-UIP group, showed independent associations with pneumothorax (OR, 5.25; 95% CI, 2.94-9.37; p < .001) and pneumothorax requiring chest tube placement (OR, 9.55; 95% CI, 3.74-24.38; p < .001). The UIP group without traversal of UIP findings by needle, relative to the non-UIP group, was not independently associated with pneumothorax (OR, 1.18; 95% CI, 0.71-1.97; p = .51) or pneumothorax requiring chest tube placement (OR, 1.08; 95% CI, 0.25-4.72; p = .92). The UIP group, with or without traversal of UIP findings by needle, was not independently associated with hemoptysis. No patient experienced air embolism or procedure-related death. CONCLUSION. Needle traversal of UIP findings is a risk factor for pneumothorax and pneumothorax requiring chest tube placement after PTNB. CLINICAL IMPACT. When performing PTNB in patients with UIP, radiologists should plan a needle trajectory that does not traverse UIP findings, when possible.

Effect of Human-AI Interaction on Detection of Malignant Lung Nodules on Chest Radiographs.

Background The factors affecting radiologists' diagnostic determinations in artificial intelligence (AI)-assisted image reading remain underexplored. Purpose To assess how AI diagnostic performance and reader characteristics influence detection of malignant lung nodules during AI-assisted reading of chest radiographs. Materials and Methods This retrospective study consisted of two reading sessions from April 2021 to June 2021. Based on the first session without AI assistance, 30 readers were assigned into two groups with equivalent areas under the free-response receiver operating characteristic curve (AUFROCs). In the second session, each group reinterpreted radiographs assisted by either a high or low accuracy AI model (blinded to the fact that two different AI models were used). Reader performance for detecting lung cancer and reader susceptibility (changing the original reading following the AI suggestion) were compared. A generalized linear mixed model was used to identify the factors influencing AI-assisted detection performance, including readers' attitudes and experiences of AI and Grit score. Results Of the 120 chest radiographs assessed, 60 were obtained in patients with lung cancer (mean age, 67 years ± 12 [SD]; 32 male; 63 cancers) and 60 in controls (mean age, 67 years ± 12; 36 male). Readers included 20 thoracic radiologists (5-18 years of experience) and 10 radiology residents (2-3 years of experience). Use of the high accuracy AI model improved readers' detection performance to a greater extent than use of the low accuracy AI model (area under the receiver operating characteristic curve, 0.77 to 0.82 vs 0.75 to 0.75; AUFROC, 0.71 to 0.79 vs 0.7 to 0.72). Readers who used the high accuracy AI showed a higher susceptibility (67%, 224 of 334 cases) to changing their diagnosis based on the AI suggestions than those using the low accuracy AI (59%, 229 of 386 cases). Accurate readings at the first session, correct AI suggestions, high accuracy Al, and diagnostic difficulty were associated with accurate AI-assisted readings, but readers' characteristics were not. Conclusion An AI model with high diagnostic accuracy led to improved performance of radiologists in detecting lung cancer on chest radiographs and increased radiologists' susceptibility to AI suggestions. © RSNA, 2023 Supplemental material is available for this article.

AI Improves Nodule Detection on Chest Radiographs in a Health Screening Population: A Randomized Controlled Trial.

Background The impact of artificial intelligence (AI)-based computer-aided detection (CAD) software has not been prospectively explored in real-world populations. Purpose To investigate whether commercial AI-based CAD software could improve the detection rate of actionable lung nodules on chest radiographs in participants undergoing health checkups. Materials and Methods In this single-center, pragmatic, open-label randomized controlled trial, participants who underwent chest radiography between July 2020 and December 2021 in a health screening center were enrolled and randomized into intervention (AI group) and control (non-AI group) arms. One of three designated radiologists with 13-36 years of experience interpreted each radiograph, referring to the AI-based CAD results for the AI group. The primary outcome was the detection rate, that is, the number of true-positive radiographs divided by the total number of radiographs, of actionable lung nodules confirmed on CT scans obtained within 3 months. Actionable nodules were defined as solid nodules larger than 8 mm or subsolid nodules with a solid portion larger than 6 mm (Lung Imaging Reporting and Data System, or Lung-RADS, category 4). Secondary outcomes included the positive-report rate, sensitivity, false-referral rate, and malignant lung nodule detection rate. Clinical outcomes were compared between the two groups using univariable logistic regression analyses. Results A total of 10 476 participants (median age, 59 years [IQR, 50-66 years]; 5121 men) were randomized to an AI group (n = 5238) or non-AI group (n = 5238). The trial met the predefined primary outcome, demonstrating an improved detection rate of actionable nodules in the AI group compared with the non-AI group (0.59% [31 of 5238 participants] vs 0.25% [13 of 5238 participants], respectively; odds ratio, 2.4; 95% CI: 1.3, 4.7; P = .008). The detection rate for malignant lung nodules was higher in the AI group compared with the non-AI group (0.15% [eight of 5238 participants] vs 0.0% [0 of 5238 participants], respectively; P = .008). The AI and non-AI groups showed similar false-referral rates (45.9% [56 of 122 participants] vs 56.0% [56 of 100 participants], respectively; P = .14) and positive-report rates (2.3% [122 of 5238 participants] vs 1.9% [100 of 5238 participants]; P = .14). Conclusion In health checkup participants, artificial intelligence-based software improved the detection of actionable lung nodules on chest radiographs. © RSNA, 2023 Supplemental material is available for this article. See also the editorial by Auffermann in this isssue.

A Deep Learning Model Using Chest Radiographs for Prediction of 30-Day Mortality in Patients With Community-Acquired Pneumonia: Development and External Validation.

BACKGROUND. Chest radiography is an essential tool for diagnosing community-acquired pneumonia (CAP), but it has an uncertain prognostic role in the care of patients with CAP. OBJECTIVE. The purpose of this study was to develop a deep learning (DL) model to predict 30-day mortality from diagnosis among patients with CAP by use of chest radiographs to validate the performance model in patients from different time periods and institutions. METHODS. In this retrospective study, a DL model was developed from data on 7105 patients from one institution from March 2013 to December 2019 (3:1:1 allocation to training, validation, and internal test sets) to predict the risk of all-cause mortality within 30 days after CAP diagnosis by use of patients' initial chest radiographs. The DL model was evaluated in a cohort of patients diagnosed with CAP during emergency department visits at the same institution from January 2020 to March 2020 (temporal test cohort [n = 947]) and in two additional cohorts from different institutions (external test cohort A [n = 467], January 2020 to December 2020; external test cohort B [n = 381], March 2019 to October 2021). AUCs were compared between the DL model and an established risk prediction tool based on the presence of confusion, blood urea nitrogen level, respiratory rate, blood pressure, and age 65 years or older (CURB-65 score). The combination of CURB-65 score and DL model was evaluated with a logistic regression model. RESULTS. The AUC for predicting 30-day mortality was significantly larger (p < .001) for the DL model than for CURB-65 score in the temporal test set (0.77 vs 0.67). The larger AUC for the DL model than for CURB-65 score was not significant (p > .05) in external test cohort A (0.80 vs 0.73) or external test cohort B (0.80 vs 0.72). In the three cohorts, the DL model, in comparison with CURB-65 score, had higher (p < .001) specificity (range, 61-69% vs 44-58%) at the sensitivity of CURB-65 score. The combination of DL model and CURB-65 score, in comparison with CURB-65 score, yielded a significant increase in AUC in the temporal test cohort (0.77, p < .001) and external test cohort B (0.80, p = .04) and a nonsignificant increase in AUC in external test cohort A (0.80, p = .16). CONCLUSION. A DL-based model consisting of initial chest radiographs was predictive of 30-day mortality among patients with CAP with improved performance over CURB-65 score. CLINICAL IMPACT. The DL-based model may guide clinical decision-making in the care of patients with CAP.

Deep Learning for Detecting Pneumothorax on Chest Radiographs after Needle Biopsy: Clinical Implementation.

Background Accurate detection of pneumothorax on chest radiographs, the most common complication of percutaneous transthoracic needle biopsies (PTNBs), is not always easy in practice. A computer-aided detection (CAD) system may help detect pneumothorax. Purpose To investigate whether a deep learning-based CAD system can improve detection performance for pneumothorax on chest radiographs after PTNB in clinical practice. Materials and Methods A CAD system for post-PTNB pneumothorax detection on chest radiographs was implemented in an institution in February 2020. This retrospective cohort study consecutively included chest radiographs interpreted with CAD assistance (CAD-applied group; February 2020 to November 2020) and those interpreted before implementation (non-CAD group; January 2018 to January 2020). The reference standard was defined by consensus reading by two radiologists. The diagnostic accuracy for pneumothorax was compared between the two groups using generalized estimating equations. Matching was performed according to whether the radiograph reader and PTNB operator were the same using the greedy method. Results A total of 676 radiographs from 655 patients (mean age: 67 years ± 11; 390 men) in the CAD-applied group and 676 radiographs from 664 patients (mean age: 66 years ± 12; 400 men) in the non-CAD group were included. The incidence of pneumothorax was 18.2% (123 of 676 radiographs) in the CAD-applied group and 22.5% (152 of 676 radiographs) in the non-CAD group (P = .05). The CAD-applied group showed higher sensitivity (85.4% vs 67.1%), negative predictive value (96.8% vs 91.3%), and accuracy (96.8% vs 92.3%) than the non-CAD group (all P < .001). The sensitivity for a small amount of pneumothorax improved in the CAD-applied group (pneumothorax of <10%: 74.5% vs 51.4%, P = .009; pneumothorax of 10%-15%: 92.7% vs 70.2%, P = .008). Among patients with pneumothorax, 34 of 655 (5.0%) in the non-CAD group and 16 of 664 (2.4%) in the CAD-applied group (P = .009) required subsequent drainage catheter insertion. Conclusion A deep learning-based computer-aided detection system improved the detection performance for pneumothorax on chest radiographs after lung biopsy. © RSNA, 2022 See also the editorial by Schiebler and Hartung in this issue.

Value of a deep learning-based algorithm for detecting Lung-RADS category 4 nodules on chest radiographs in a health checkup population: estimation of the sample size for a randomized controlled trial.

OBJECTIVE: To explore the value of a deep learning-based algorithm in detecting Lung CT Screening Reporting and Data System category 4 nodules on chest radiographs from an asymptomatic health checkup population. METHODS: Data from an annual retrospective cohort of individuals who underwent chest radiographs for health checkup purposes and chest CT scanning within 3 months were collected. Among 3073 individuals, 118 with category 4 nodules on CT were selected. A reader performance test was performed using those 118 radiographs and randomly selected 51 individuals without any nodules. Four radiologists independently evaluated the radiographs without and with the results of the algorithm; and sensitivities/specificities were compared. The sample size needed to confirm the difference in detection rates was calculated, i.e., the number of true-positive radiographs divided by the total number of radiographs. RESULTS: The sensitivity of the radiologists substantially increased aided by the algorithm (38.8% [183/472] to 45.1% [213/472]; p < .001) without significant change in specificity (94.1% [192/204] vs. 92.2% [188/204]; p = .22). Pooled radiologists detected more nodules with the algorithm (32.0% [156/488] vs. 38.9% [190/488]; p < .001), without alteration of false-positive rates (0.09 [62/676], both). Pooled detection rates for the annual cohort were 1.49% (183/12,292) and 1.73% (213/12,292) without and with the algorithm, respectively. A sample size of 41,776 in each arm would be required to demonstrate significant detection rate difference with < 5% type I error and > 80% power. CONCLUSION: Although readers substantially increased sensitivity in detecting nodules on chest radiographs from a health checkup population aided by the algorithm, detection rate difference was only 0.24%, requiring a sample size >80,000 for a randomized controlled trial. KEY POINTS: • Aided by a deep learning algorithm, pooled radiologists improved their sensitivity in detecting Lung-RADS category 4 nodules on chest radiographs from a health checkup population (38.8% [183/472] to 45.1% [213/472]; p < .001), without increasing false-positive rate. • The prevalence of the Lung-RADS category 4 nodules was 3.8% (118/3073) on the population, resulting in only 0.24% increase of the detection rate for the radiologists with assistance of the algorithm. • To confirm the significant detection rate increase by a randomized controlled trial, a sample size of 84,000 would be required.

Artificial intelligence system for identification of false-negative interpretations in chest radiographs.

OBJECTIVES: To investigate the efficacy of an artificial intelligence (AI) system for the identification of false negatives in chest radiographs that were interpreted as normal by radiologists. METHODS: We consecutively collected chest radiographs that were read as normal during 1 month (March 2020) in a single institution. A commercialized AI system was retrospectively applied to these radiographs. Radiographs with abnormal AI results were then re-interpreted by the radiologist who initially read the radiograph ("AI as the advisor" scenario). The reference standards for the true presence of relevant abnormalities in radiographs were defined by majority voting of three thoracic radiologists. The efficacy of the AI system was evaluated by detection yield (proportion of true-positive identification among the entire examination) and false-referral rate (FRR, proportion of false-positive identification among all examinations). Decision curve analyses were performed to evaluate the net benefits of applying the AI system. RESULTS: A total of 4208 radiographs from 3778 patients (M:F = 1542:2236; median age, 56 years) were included. The AI system identified initially overlooked relevant abnormalities with a detection yield and an FRR of 2.4% and 14.0%, respectively. In the "AI as the advisor" scenario, radiologists detected initially overlooked relevant abnormalities with a detection yield and FRR of 1.2% and 0.97%, respectively. In a decision curve analysis, AI as an advisor scenario exhibited a positive net benefit when the cost-to-benefit ratio was below 1:0.8. CONCLUSION: An AI system could identify relevant abnormalities overlooked by radiologists and could enable radiologists to correct their false-negative interpretations by providing feedback to radiologists. KEY POINTS: • In consecutive chest radiographs with normal interpretations, an artificial intelligence system could identify relevant abnormalities that were initially overlooked by radiologists. • The artificial intelligence system could enable radiologists to correct their initial false-negative interpretations by providing feedback to radiologists when overlooked abnormalities were present.

3-hydroxymorphinan enhances mitochondrial biogenesis and adipocyte browning through AMPK-dependent pathway.

3-hydroxymorphinan (3-HM), a metabolite of dextromethorphan, has previously been reported to have anti-inflammatory, anti-oxidative stress, and neuroprotective effects. However, its effect on energy metabolism in adipocytes remains unclear. Herein, we investigated 3-hydroxymorphinan (3-HM) effects on mitochondrial biogenesis, oxidative stress, and lipid accumulation in 3T3-L1 adipocytes. Further, we explored 3-HM-associated molecular mechanisms. Mouse adipocyte 3T3-L1 cells were treated with 3-HM, and various protein expression levels were determined by western blotting analysis. Mitochondria accumulation and lipid accumulation were measured by staining methods. Cell toxicity was assessed by cell viability assay. We found that treatment of 3T3-L1 adipocytes with 3-HM increased expression of brown adipocyte markers, such as uncoupling protein-1 (UCP-1) and peroxisome proliferator-activated receptor-gamma coactivator 1-alpha (PGC-1α). 3-HM promotes mitochondrial biogenesis and its-mediated gene expression. Additionally, 3-HM treatment suppressed mitochondrial ROS generation and superoxide along with improved mitochondrial complex I activity. We found that treatment of 3-HM enhanced AMPK phosphorylation. siRNA-mediated suppression of AMPK reversed all these changes in 3T3-L1 adipocytes. In sum, 3-HM promotes mitochondrial biogenesis and browning and attenuates oxidative stress and lipid accumulation in 3T3-L1 adipocytes via AMPK signaling. Thus, 3-HM-mediated AMPK activation can be considered a therapeutic approach for treating obesity and related diseases.

Deep Learning for Detection of Pulmonary Metastasis on Chest Radiographs.

Background A computer-aided detection (CAD) system may help surveillance for pulmonary metastasis at chest radiography in situations where there is limited access to CT. Purpose To evaluate whether a deep learning (DL)-based CAD system can improve diagnostic yield for newly visible lung metastasis on chest radiographs in patients with cancer. Materials and Methods A regulatory-approved CAD system for lung nodules was implemented to interpret chest radiographs from patients referred by the medical oncology department in clinical practice. In this retrospective diagnostic cohort study, chest radiographs interpreted with assistance from a CAD system after the implementation (January to April 2019, CAD-assisted interpretation group) and those interpreted before the implementation (September to December 2018, conventional interpretation group) of the CAD system were consecutively included. The diagnostic yield (frequency of true-positive detections) and false-referral rate (frequency of false-positive detections) of formal reports of chest radiographs for newly visible lung metastasis were compared between the two groups using generalized estimating equations. Propensity score matching was performed between the two groups for age, sex, and primary cancer. Results A total of 2916 chest radiographs from 1521 patients (1546 men, 1370 women; mean age, 62 years) and 5681 chest radiographs from 3456 patients (2941 men, 2740 women; mean age, 62 years) were analyzed in the CAD-assisted interpretation and conventional interpretation groups, respectively. The diagnostic yield for newly visible metastasis was higher in the CAD-assisted interpretation group (0.86%, 25 of 2916 [95% CI: 0.58, 1.3] vs 0.32%, 18 of 568 [95% CI: 0.20, 0.50%]; P = .004). The false-referral rate in the CAD-assisted interpretation group (0.34%, 10 of 2916 [95% CI: 0.19, 0.64]) was not inferior to that in the conventional interpretation group (0.25%, 14 of 5681 [95% CI: 0.15, 0.42]) at the noninferiority margin of 0.5% (95% CI of difference: -0.15, 0.35). Conclusion A deep learning-based computer-aided detection system improved the diagnostic yield for newly visible metastasis on chest radiographs in patients with cancer with a similar false-referral rate. © RSNA, 2021 Online supplemental material is available for this article.

Development and validation of a deep learning algorithm detecting 10 common abnormalities on chest radiographs.

We aimed to develop a deep learning algorithm detecting 10 common abnormalities (DLAD-10) on chest radiographs, and to evaluate its impact in diagnostic accuracy, timeliness of reporting and workflow efficacy.DLAD-10 was trained with 146 717 radiographs from 108 053 patients using a ResNet34-based neural network with lesion-specific channels for 10 common radiological abnormalities (pneumothorax, mediastinal widening, pneumoperitoneum, nodule/mass, consolidation, pleural effusion, linear atelectasis, fibrosis, calcification and cardiomegaly). For external validation, the performance of DLAD-10 on a same-day computed tomography (CT)-confirmed dataset (normal:abnormal 53:147) and an open-source dataset (PadChest; normal:abnormal 339:334) was compared with that of three radiologists. Separate simulated reading tests were conducted on another dataset adjusted to real-world disease prevalence in the emergency department, consisting of four critical, 52 urgent and 146 nonurgent cases. Six radiologists participated in the simulated reading sessions with and without DLAD-10.DLAD-10 exhibited area under the receiver operating characteristic curve values of 0.895-1.00 in the CT-confirmed dataset and 0.913-0.997 in the PadChest dataset. DLAD-10 correctly classified significantly more critical abnormalities (95.0% (57/60)) than pooled radiologists (84.4% (152/180); p=0.01). In simulated reading tests for emergency department patients, pooled readers detected significantly more critical (70.8% (17/24) versus 29.2% (7/24); p=0.006) and urgent (82.7% (258/312) versus 78.2% (244/312); p=0.04) abnormalities when aided by DLAD-10. DLAD-10 assistance shortened the mean±sd time-to-report critical and urgent radiographs (640.5±466.3 versus 3371.0±1352.5 s and 1840.3±1141.1 versus 2127.1±1468.2 s, respectively; all p<0.01) and reduced the mean±sd interpretation time (20.5±22.8 versus 23.5±23.7 s; p<0.001).DLAD-10 showed excellent performance, improving radiologists' performance and shortening the reporting time for critical and urgent cases.

Optimum diameter threshold for lung nodules at baseline lung cancer screening with low-dose chest CT: exploration of results from the Korean Lung Cancer Screening Project.

OBJECTIVES: To explore the optimum diameter threshold for solid nodules to define positive results at baseline screening low-dose CT (LDCT) and to compare two-dimensional and volumetric measurement of lung nodules for the diagnosis of lung cancers. METHODS: We included consecutive participants from the Korean Lung Cancer Screening project between 2017 and 2018. The average transverse diameter and effective diameter (diameter of a sphere with the same volume) of lung nodules were measured by semi-automated segmentation. Diagnostic performances for lung cancers diagnosed within 1 year after LDCT were evaluated using area under receiver-operating characteristic curves (AUCs), sensitivities, and specificities, with diameter thresholds for solid nodules ranging from 6 to 10 mm. The reduction of unnecessary follow-up LDCTs and the diagnostic delay of lung cancers were estimated for each threshold. RESULTS: Fifty-two lung cancers were diagnosed among 10,424 (10,141 men; median age 62 years) participants within 1 year after LDCT. Average transverse (0.980) and effective diameters (0.981) showed similar AUCs (p = .739). Elevating the average transverse diameter threshold from 6 to 9 mm resulted in a significantly increased specificity (91.7 to 96.7%, p < .001), a modest reduction in sensitivity (96.2 to 94.2%, p = .317), a 60.2% estimated reduction of unnecessary follow-up LDCTs, and a diagnostic delay in 1.9% of lung cancers. Elevating the threshold to 10 mm led to a significant reduction in sensitivity (86.5%, p = .025). CONCLUSIONS: Elevating the diameter threshold for solid nodules from 6 to 9 mm may lead to a substantial reduction in unnecessary follow-up LDCTs with a small proportion of diagnostic delay of lung cancers. KEY POINTS: • Elevation of the diameter threshold for solid nodules from 6 to 9 mm can substantially reduce unnecessary follow-up LDCTs with a small proportion of diagnostic delay of lung cancers. • The average transverse and effective diameters of lung nodules showed similar performances for the prediction of a lung cancer diagnosis.

Implementation of the cloud-based computerized interpretation system in a nationwide lung cancer screening with low-dose CT: comparison with the conventional reading system.

OBJECTIVES: We aimed to compare the CT interpretation before and after the implementation of a computerized system for lung nodule detection and measurements in a nationwide lung cancer screening program. METHODS: Our screening program started in April 2017, with 14 participating institutions. Initially, all CTs were interpreted using interpretation systems in each institution and manual nodule measurement (conventional system). A cloud-based CT interpretation system, equipped with semi-automated measurement and CAD (computer-aided detection) for lung nodules (cloud-based system), was implemented during the project. Positive rates and performances for lung cancer diagnosis based on the Lung-RADS version 1.0 were compared between the conventional and cloud-based systems. RESULTS: A total of 1821 (M:F = 1782:39, mean age 62.7 years, 16 confirmed lung cancers) and 4666 participants (M:F = 4560:106, mean age 62.8 years, 31 confirmed lung cancers) were included in the conventional and cloud-based systems, respectively. Significantly more nodules were detected in the cloud-based system (0.76 vs. 1.07 nodule/participant, p < .001). Positive rate did not differ significantly between the two systems (9.9% vs. 11.0%, p = .211), while their variability across institutions was significantly lower in the cloud-based system (coefficients of variability, 0.519 vs. 0.311, p = .018). The Lung-RADS-based sensitivity (93.8% vs. 93.5%, p = .979) and specificity (90.9% vs. 89.6%, p = .132) did not differ significantly between the two systems. CONCLUSION: Implementation of CAD and semi-automated measurement for lung nodules in a nationwide lung cancer screening program resulted in increased number of detected nodules and reduced variability in positive rates across institutions. KEY POINTS: • Computer-aided CT reading detected more lung nodules than radiologists alone in lung cancer screening. • Positive rate in lung cancer screening did not change with computer-aided reading. • Computer-aided CT reading reduced inter-institutional variability in lung cancer screening.

Variability in interpretation of low-dose chest CT using computerized assessment in a nationwide lung cancer screening program: comparison of prospective reading at individual institutions and retrospective central reading.

OBJECTIVES: To evaluate the degree of variability in computer-assisted interpretation of low-dose chest CTs (LDCTs) among radiologists in a nationwide lung cancer screening (LCS) program, through comparison with a retrospective interpretation from a central laboratory. MATERIALS AND METHODS: Consecutive baseline LDCTs (n = 3353) from a nationwide LCS program were investigated. In the institutional reading, 20 radiologists in 14 institutions interpreted LDCTs using computer-aided detection and semi-automated segmentation systems for lung nodules. In the retrospective central review, a single radiologist re-interpreted all LDCTs using the same system, recording any non-calcified nodules ≥ 3 mm without arbitrary rejection of semi-automated segmentation to minimize the intervention of radiologist's discretion. Positive results (requiring additional follow-up LDCTs or diagnostic procedures) were initially classified by the lung CT screening reporting and data system (Lung-RADS) during the interpretation, while the classifications based on the volumetric criteria from the Dutch-Belgian lung cancer screening trial (NELSON) were retrospectively applied. Variabilities in positive rates were assessed with coefficients of variation (CVs). RESULTS: In the institutional reading, positive rates by the Lung-RADS ranged from 7.5 to 43.3%, and those by the NELSON ranged from 11.4 to 45.0% across radiologists. The central review exhibited higher positive rates by Lung-RADS (20.0% vs. 27.3%; p < .001) and the NELSON (23.1% vs. 37.0%; p < .001), and lower inter-institution variability (CV, 0.30 vs. 0.12, p = .003 by Lung-RADS; CV, 0.25 vs. 0.12, p = .014 by the NELSON) compared to the institutional reading. CONCLUSION: Considerable inter-institution variability in the interpretation of LCS results is caused by different usage of the computer-assisted system. KEY POINTS: • Considerable variability existed in the interpretation of screening LDCT among radiologists partly from the different usage of the computerized system. • A retrospective reading of low-dose chest CTs in the central laboratory resulted in reduced variability but an increased positive rate.

Prediction of visceral pleural invasion in lung cancer on CT: deep learning model achieves a radiologist-level performance with adaptive sensitivity and specificity to clinical needs.

OBJECTIVES: To develop and validate a preoperative CT-based deep learning model for the prediction of visceral pleural invasion (VPI) in early-stage lung cancer. METHODS: In this retrospective study, dataset 1 (for training, tuning, and internal validation) included 676 patients with clinical stage IA lung adenocarcinomas resected between 2009 and 2015. Dataset 2 (for temporal validation) included 141 patients with clinical stage I adenocarcinomas resected between 2017 and 2018. A CT-based deep learning model was developed for the prediction of VPI and validated in terms of discrimination and calibration. An observer performance study and a multivariable regression analysis were performed. RESULTS: The area under the receiver operating characteristic curve (AUC) of the model was 0.75 (95% CI, 0.67-0.84), which was comparable to those of board-certified radiologists (AUC, 0.73-0.79; all p > 0.05). The model had a higher standardized partial AUC for a specificity range of 90 to 100% than the radiologists (all p < 0.05). The high sensitivity cutoff (0.245) yielded a sensitivity of 93.8% and a specificity of 31.2%, and the high specificity cutoff (0.448) resulted in a sensitivity of 47.9% and a specificity of 86.0%. Two of the three radiologists provided highly sensitive (93.8% and 97.9%) but not specific (48.4% and 40.9%) diagnoses. The model showed good calibration (p > 0.05), and its output was an independent predictor for VPI (adjusted odds ratio, 1.07; 95% CI, 1.03-1.11; p < 0.001). CONCLUSIONS: The deep learning model demonstrated a radiologist-level performance. The model could achieve either highly sensitive or highly specific diagnoses depending on clinical needs. KEY POINTS: • The preoperative CT-based deep learning model demonstrated an expert-level diagnostic performance for the presence of visceral pleural invasion in early-stage lung cancer. • Radiologists had a tendency toward highly sensitive, but not specific diagnoses for the visceral pleural invasion.

Deep learning-based automated detection algorithm for active pulmonary tuberculosis on chest radiographs: diagnostic performance in systematic screening of asymptomatic individuals.

OBJECTIVES: Performance of deep learning-based automated detection (DLAD) algorithms in systematic screening for active pulmonary tuberculosis is unknown. We aimed to validate DLAD algorithm for detection of active pulmonary tuberculosis and any radiologically identifiable relevant abnormality on chest radiographs (CRs) in this setting. METHODS: We performed out-of-sample testing of a pre-trained DLAD algorithm, using CRs from 19.686 asymptomatic individuals (ages, 21.3 ± 1.9 years) as part of systematic screening for tuberculosis between January 2013 and July 2018. Area under the receiver operating characteristic curves (AUC) for diagnosis of tuberculosis and any relevant abnormalities were measured. Accuracy measures including sensitivities, specificities, positive predictive values (PPVs), and negative predictive values (NPVs) were calculated at pre-defined operating thresholds (high sensitivity threshold, 0.16; high specificity threshold, 0.46). RESULTS: All five CRs from four individuals with active pulmonary tuberculosis were correctly classified as having abnormal findings by DLAD with specificities of 0.959 and 0.997, PPVs of 0.006 and 0.068, and NPVs of both 1.000 at high sensitivity and high specificity thresholds, respectively. With high specificity thresholds, DLAD showed comparable diagnostic measures with the pooled radiologists (p values > 0.05). For the radiologically identifiable relevant abnormality (n = 28), DLAD showed an AUC value of 0.967 (95% confidence interval, 0.938-0.996) with sensitivities of 0.821 and 0.679, specificities of 0.960 and 0.997, PPVs of 0.028 and 0.257, and NPVs of both 0.999 at high sensitivity and high specificity thresholds, respectively. CONCLUSIONS: In systematic screening for tuberculosis in a low-prevalence setting, DLAD algorithm demonstrated excellent diagnostic performance, comparable with the radiologists in the detection of active pulmonary tuberculosis. KEY POINTS: • Deep learning-based automated detection algorithm detected all chest radiographs with active pulmonary tuberculosis with high specificities and negative predictive values in systematic screening. • Deep learning-based automated detection algorithm had comparable diagnostic measures with the radiologists for detection of active pulmonary tuberculosis on chest radiographs. • For the detection of radiologically identifiable relevant abnormalities on chest radiographs, deep learning-based automated detection algorithm showed excellent diagnostic performance in systematic screening.

Automatic prediction of left cardiac chamber enlargement from chest radiographs using convolutional neural network.

OBJECTIVE: To develop deep learning-based cardiac chamber enlargement-detection algorithms for left atrial (DLCE-LAE) and ventricular enlargement (DLCE-LVE), on chest radiographs METHODS: For training and internal validation of DLCE-LAE and -LVE, 5,045 chest radiographs (CRs; 2,463 normal and 2,393 LAE) and 1,012 CRs (456 normal and 456 LVE) matched with the same-day echocardiography were collected, respectively. External validation was performed using 107 temporally independent CRs. Reader performance test was conducted using the external validation dataset by five cardiothoracic radiologists without and with the results of DLCE. Classification performance of DLCE was evaluated and compared with those of the readers and conventional radiographic features, including cardiothoracic ratio, carinal angle, and double contour. In addition, DLCE-LAE was tested on 5,277 CRs from a healthcare screening program cohort. RESULTS: DLCE-LAE showed areas under the receiver operating characteristics curve (AUROCs) of 0.858 on external validation. On reader performance test, DLCE-LAE showed better results than pooled radiologists (AUROC 0.858 vs. 0.651; p < .001) and significantly increased their performance when used as a second reader (AUROC 0.651 vs. 0.722; p < .001). DLCE-LAE also showed a significantly higher AUROC than conventional radiographic findings (AUROC 0.858 vs. 0.535-0.706; all ps < .01). In the healthcare screening cohort, DLCE-LAE successfully detected 71.0% (142/200) CRs with moderate-to-severe LAE (93.5% [29/31] of severe cases), while yielding 11.8% (492/4,184) false-positive rate. DLCE-LVE showed AUROCs of 0.966 and 0.594 on internal and external validation, respectively. CONCLUSION: DLCE-LAE outperformed and improved cardiothoracic radiologists' performance in detecting LAE and showed promise in screening individuals with moderate-to-severe LAE in a healthcare screening cohort. KEY POINTS: • Our deep learning algorithm outperformed cardiothoracic radiologists in detecting left atrial enlargement on chest radiographs. • Cardiothoracic radiologists improved their performance in detecting left atrial enlargement when aided by the algorithm. • On a healthcare-screening cohort, our algorithm detected 71.0% (142/200) radiographs with moderate-to-severe left atrial enlargement while yielding 11.8% (492/4,184) false-positive rate.

Deep learning computer-aided detection system for pneumonia in febrile neutropenia patients: a diagnostic cohort study.

BACKGROUND: Diagnosis of pneumonia is critical in managing patients with febrile neutropenia (FN), however, chest X-ray (CXR) has limited performance in the detection of pneumonia. We aimed to evaluate the performance of a deep learning-based computer-aided detection (CAD) system in pneumonia detection in the CXRs of consecutive FN patients and investigated whether CAD could improve radiologists' diagnostic performance when used as a second reader. METHODS: CXRs of patients with FN (a body temperature ≥ 38.3 °C, or a sustained body temperature ≥ 38.0 °C for an hour; absolute neutrophil count < 500/mm3) obtained between January and December 2017 were consecutively included, from a single tertiary referral hospital. Reference standards for the diagnosis of pneumonia were defined by consensus of two thoracic radiologists after reviewing medical records and CXRs. A commercialized, deep learning-based CAD system was retrospectively applied to detect pulmonary infiltrates on CXRs. For comparing performance, five radiologists independently interpreted CXRs initially without the CAD results (radiologist-alone interpretation), followed by the interpretation with CAD. The sensitivities and specificities for detection of pneumonia were compared between radiologist-alone interpretation and interpretation with CAD. The standalone performance of the CAD was also evaluated, using area under the receiver operating characteristic curve (AUC), sensitivity, and specificity. Moreover, sensitivity and specificity of standalone CAD were compared with those of radiologist-alone interpretation. RESULTS: Among 525 CXRs from 413 patients (52.3% men; median age 59 years), pneumonia was diagnosed in 128 (24.4%) CXRs. In the interpretation with CAD, average sensitivity of radiologists was significantly improved (75.4% to 79.4%, P = 0.003) while their specificity remained similar (75.4% to 76.8%, P = 0.101), compared to radiologist-alone interpretation. The CAD exhibited AUC, sensitivity, and specificity of 0.895, 88.3%, and 68.3%, respectively. The standalone CAD exhibited higher sensitivity (86.6% vs. 75.2%, P < 0.001) and lower specificity (64.8% vs. 75.4%, P < 0.001) compared to radiologist-alone interpretation. CONCLUSIONS: In patients with FN, the deep learning-based CAD system exhibited radiologist-level performance in detecting pneumonia on CXRs and enhanced radiologists' performance.

Performance of a Deep Learning Algorithm Compared with Radiologic Interpretation for Lung Cancer Detection on Chest Radiographs in a Health Screening Population.

Background The performance of a deep learning algorithm for lung cancer detection on chest radiographs in a health screening population is unknown. Purpose To validate a commercially available deep learning algorithm for lung cancer detection on chest radiographs in a health screening population. Materials and Methods Out-of-sample testing of a deep learning algorithm was retrospectively performed using chest radiographs from individuals undergoing a comprehensive medical check-up between July 2008 and December 2008 (validation test). To evaluate the algorithm performance for visible lung cancer detection, the area under the receiver operating characteristic curve (AUC) and diagnostic measures, including sensitivity and false-positive rate (FPR), were calculated. The algorithm performance was compared with that of radiologists using the McNemar test and the Moskowitz method. Additionally, the deep learning algorithm was applied to a screening cohort undergoing chest radiography between January 2008 and December 2012, and its performances were calculated. Results In a validation test comprising 10 285 radiographs from 10 202 individuals (mean age, 54 years ± 11 [standard deviation]; 5857 men) with 10 radiographs of visible lung cancers, the algorithm's AUC was 0.99 (95% confidence interval: 0.97, 1), and it showed comparable sensitivity (90% [nine of 10 radiographs]) to that of the radiologists (60% [six of 10 radiographs]; P = .25) with a higher FPR (3.1% [319 of 10 275 radiographs] vs 0.3% [26 of 10 275 radiographs]; P < .001). In the screening cohort of 100 525 chest radiographs from 50 070 individuals (mean age, 53 years ± 11; 28 090 men) with 47 radiographs of visible lung cancers, the algorithm's AUC was 0.97 (95% confidence interval: 0.95, 0.99), and its sensitivity and FPR were 83% (39 of 47 radiographs) and 3% (2999 of 100 478 radiographs), respectively. Conclusion A deep learning algorithm detected lung cancers on chest radiographs with a performance comparable to that of radiologists, which will be helpful for radiologists in healthy populations with a low prevalence of lung cancer. © RSNA, 2020 Online supplemental material is available for this article. See also the editorial by Armato in this issue.