Performance Analysis in Children of Traditional and Deep Learning CT Lung Nodule Computer-Aided Detection Systems Trained on Adults.

BACKGROUND. Although primary lung cancer is rare in children, chest CT is commonly performed to assess for lung metastases in children with cancer. Lung nodule computer-aided detection (CAD) systems have been designed and studied primarily using adult training data, and the efficacy of such systems when applied to pediatric patients is poorly understood. OBJECTIVE. The purpose of this study was to evaluate in children the diagnostic performance of traditional and deep learning CAD systems trained with adult data for the detection of lung nodules on chest CT scans and to compare the ability of such systems to generalize to children versus to other adults. METHODS. This retrospective study included pediatric and adult chest CT test sets. The pediatric test set comprised 59 CT scans in 59 patients (30 boys, 29 girls; mean age, 13.1 years; age range, 4-17 years), which were obtained from November 30, 2018, to August 31, 2020; lung nodules were annotated by fellowship-trained pediatric radiologists as the reference standard. The adult test set was the publicly available adult Lung Nodule Analysis (LUNA) 2016 subset 0, which contained 89 deidentified scans with previously annotated nodules. The test sets were processed through the traditional FlyerScan (github.com/rhardie1/FlyerScanCT) and deep learning Medical Open Network for Artificial Intelligence (MONAI; github.com/Project-MONAI/model-zoo/releases) lung nodule CAD systems, which had been trained on separate sets of CT scans in adults. Sensitivity and false-positive (FP) frequency were calculated for nodules measuring 3-30 mm; nonoverlapping 95% CIs indicated significant differences. RESULTS. Operating at two FPs per scan, on pediatric testing data FlyerScan and MONAI showed significantly lower detection sensitivities of 68.4% (197/288; 95% CI, 65.1-73.0%) and 53.1% (153/288; 95% CI, 46.7-58.4%), respectively, than on adult LUNA 2016 subset 0 testing data (83.9% [94/112; 95% CI, 79.1-88.0%] and 95.5% [107/112; 95% CI, 90.0-98.4%], respectively). Mean nodule size was smaller (p < .001) in the pediatric testing data (5.4 ± 3.1 [SD] mm) than in the adult LUNA 2016 subset 0 testing data (11.0 ± 6.2 mm). CONCLUSION. Adult-trained traditional and deep learning-based lung nodule CAD systems had significantly lower sensitivity for detection on pediatric data than on adult data at a matching FP frequency. The performance difference may relate to the smaller size of pediatric lung nodules. CLINICAL IMPACT. The results indicate a need for pediatric-specific lung nodule CAD systems trained on data specific to pediatric patients.

Performance of Lung-Nodule Computer-Aided Detection Systems on Standard-Dose and Low-Dose Pediatric CT Scans: An Intraindividual Comparison.

Background: When applying lung-nodule computer-aided detection (CAD) systems for pediatric CT, performance may be degraded on low-dose scans due to increased image noise. Objective: To conduct an intraindividual comparison of the performance for lung nodule detection of two CAD systems trained using adult data between low-dose and standard-dose pediatric chest CT scans. Methods: This retrospective study included 73 patients (32 female, 41 male; mean age, 14.7 years; age range, 4-20 years) who underwent both clinical standard-dose and investigational low-dose chest CT examinations within the same encounter from November 30, 2018 to August 31, 2020 as part of an earlier prospective study. Fellowship-trained pediatric radiologists annotated lung nodules to serve as the reference standard. Both CT scans were processed using two publicly available lung-nodule CAD systems previously trained using adult data: FlyerScan and Medical Open Network for Artificial Intelligence (MONAI). The systems' sensitivities for nodules measuring 3-30 mm (n=247) were calculated when operating at a fixed frequency of two false-positives per scan. Results: FlyerScan exhibited detection sensitivities of 76.9% (190/247; 95% CI: 73.3-80.8%) on standard-dose scans and 66.8% (165/247; 95% CI: 62.6-71.5) on low-dose scans. MONAI exhibited detection sensitivities of 67.6% (167/247, 95% CI: 61.5-72.1) on standard-dose scans and 62.3% (154/247, 95% CI: 56.1-66.5%) on low-dose scans. The number of detected nodules for standard-dose versus low-dose scans for 3-mm nodules was 33 versus 24 (FlyerScan) and 16 versus 13 (MONAI), 4-mm nodules was 46 versus 42 (FlyerScan) and 39 versus 30 (MONAI), 5-mm nodules was 38 versus 33 (FlyerScan) and 32 versus 31 (MONAI), and 6-mm nodules was 27 versus 20 (FlyerScan) and 24 versus 24 (MONAI). For nodules measuring ≥7 mm, detection did not show a consistent pattern between standard-dose and low-dose scans for either system. Conclusions: Two lung-nodule CAD systems demonstrated decreased sensitivity on low-dose versus standard-dose pediatric CT scans performed in the same patients. The reduced detection at low dose was overall more pronounced for nodules measuring less than 5 mm. Clinical Impact: Caution is needed when using low-dose CT protocols in combination with CAD systems to help detect small lung nodules in pediatric patients.

Connective tissue disease-associated lung disease in children.

Connective tissue diseases are a heterogeneous group of autoimmune diseases that can affect a variety of organ systems. Lung parenchymal involvement is an important contributor to morbidity and mortality in children with connective tissue disease. Connective tissue disease-associated lung disease in children often manifests as one of several radiologic-pathologic patterns of disease, with certain patterns having a propensity to occur in association with certain connective tissue diseases. In this article, key clinical, histopathologic, and computed tomography (CT) features of typical patterns of connective tissue disease-associated lung disease in children are reviewed, with an emphasis on radiologic-pathologic correlation, to improve recognition of these patterns of lung disease at CT and to empower the pediatric radiologist to more fully contribute to the care of pediatric patients with these conditions.

Accelerated Cine Cardiac MRI Using Deep Learning-Based Reconstruction: A Systematic Evaluation.

BACKGROUND: Breath-holding (BH) for cine balanced steady state free precession (bSSFP) imaging is challenging for patients with impaired BH capacity. Deep learning-based reconstruction (DLR) of undersampled k-space promises to shorten BHs while preserving image quality and accuracy of ventricular assessment. PURPOSE: To perform a systematic evaluation of DLR of cine bSSFP images from undersampled k-space over a range of acceleration factors. STUDY TYPE: Retrospective. SUBJECTS: Fifteen pectus excavatum patients (mean age 16.8 ± 5.4 years, 20% female) with normal cardiac anatomy and function and 12-second BH capability. FIELD STRENGTH/SEQUENCE: 1.5-T, cine bSSFP. ASSESSMENT: Retrospective DLR was conducted by applying compressed sensitivity encoding (C-SENSE) acceleration to systematically undersample fully sampled k-space cine bSSFP acquisition data over an acceleration/undersampling factor (R) considering a range of 2 to 8. Quality imperceptibility (QI) measures, including structural similarity index measure, were calculated using images reconstructed from fully sampled k-space as a reference. Image quality, including contrast and edge definition, was evaluated for diagnostic adequacy by three readers with varying levels of experience in cardiac MRI (>4 years, >18 years, and 1 year). Automated DL-based biventricular segmentation was performed commercially available software by cardiac radiologists with more than 4 years of experience. STATISTICAL TESTS: Tukey box plots, linear mixed effects model, analysis of variance (ANOVA), weighted kappa, Kruskal-Wallis test, and Wilcoxon signed-rank test were employed as appropriate. A P-value <0.05 was considered statistically significant. RESULTS: There was a significant decrease in the QI values and edge definition scores as R increased. Diagnostically adequate image quality was observed up to R = 5. The effect of R on all biventricular volumetric indices was non-significant (P = 0.447). DATA CONCLUSION: The biventricular volumetric indices obtained from the reconstruction of fully sampled cine bSSFP acquisitions and DLR of the same k-space data undersampled by C-SENSE up to R = 5 may be comparable. EVIDENCE LEVEL: 3 TECHNICAL EFFICACY: Stage 1.

The undulating line sign and other more common pediatric central catheter malpositions.

Central venous and arterial catheters are among the most commonly assessed support devices by radiologists. The position of these catheters must be carefully assessed to ensure proper placement, as malpositioning may lead to life-threatening consequences. Therefore, it is important for radiologists to understand the anatomy of the central vessels and the expected location of catheters. While this can be difficult in small children and especially in neonates, knowledge of the expected course and ideal termination of catheters allows for recognition of a malpositioned line, which may be unsuspected clinically. The purpose of this article is to discuss appropriate positioning of central catheters in pediatric patients, focusing primarily on venous catheters. We also propose a new radiographic sign to recognize, the undulating line sign, as an indication of an inappropriate course of a newly placed venous catheter.

Characteristics of wake-up stroke.

BACKGROUND: Wake-up stroke (WUS) accounts for up to 29.6% of ischemic strokes, but its mechanisms are poorly understood. The purpose of this study is to identify risk factors and characteristics of WUS. METHODS: Seven-two ischemic strokes were classified as WUS or non-WUS. Collected were demographic information, medical history, cholesterol profile, and stroke characteristics including severity (National Institutes of Health Stroke Scale [NIHSS]) and mechanism (Trial of Org 10172 in Acute Stroke Treatment criteria). Subjects completed questionnaires screening for sleep apnea (Berlin questionnaire) and assessing sleep characteristics. RESULTS: There were 72 ischemic strokes, of which 28 WUS (38.9%). WUS and non-WUS patients were similar in regard to stroke risk factors. WUS patients tended to be African American and were significantly younger. WUS was significantly more likely to result from small-vessel disease mechanism (42.9% versus 14.0%; P=.006) and tended to be less severe WUS (NIHSS score 3 [1, 4] versus 4 [2, 11]; P=.13) than non-WUS. Groups did not differ in regard to scoring positively on the Berlin questionnaire, but WUS sufferers were more likely to snore frequently (90.5% versus 70.0%, P=.08). The lipid profile was significantly worse in WUS compared with non-WUS (low-density lipoprotein 124.6±38.4 versus 103.7±36.8; P=.03; cholesterol to high-density lipoprotein ratio 5.2±1.6 versus 4.3±1.6; P=.02). CONCLUSIONS: WUS is more likely to result from small-vessel disease mechanism. Poorer cholesterol profile and frequent snoring may contribute to WUS.

Imaging in Pediatric Lung Disease: Current Practice and Future Directions.

Pediatric diseases present differently from adult diseases and imaging forms a cornerstone of modern pediatric care through differential diagnosis, disease monitoring, and measuring response to therapy. Imaging is especially well suited to providing novel insights into the underlying mechanisms driving disease through structural and functional imaging. In this review, we describe key imaging findings in standard-of-care and state-of-the-art techniques in pediatric and adult diseases with origins in childhood. We examine applications in small airways disease, large airway disease, diseases of maturity, interstitial lung disease, neuromuscular disease, congenital disease, and pulmonary infection.