Incidence and Risk Factors for Brown Adipose Tissue Uptake in PET Imaging in Pediatric Patients.

BACKGROUND: Positron emission tomography (PET) scans are used in disease diagnosis and evaluation for pediatric oncology patients. Brown adipose tissue (BAT) 18 F-fluorodeoxyglucose-PET uptake is reported in 35% to 47% of pediatric patients. Several risk factors may be associated with BAT uptake. OBJECTIVE: The aim was to determine the incidence and risk factors for BAT in pediatric patients using a consensus-based system and a novel grading scale. METHODS: A total of 285 PET scans in 154 patients were retrospectively reviewed for the presence of BAT from September 2015 through December 2016. A consensus review was done by 2 radiologists, who graded BAT on a 0 to 3 scale and assessed its impact on PET interpretation. RESULTS: The presence of moderate to severe BAT occurred in 11% of PET scans, and 6% of PETs had limited interpretation. Hodgkin lymphoma (n=53) patients had a 3.62-fold increased odds of moderate or severe BAT and a 6.59-fold increased odds of limited interpretation on PET imaging. CONCLUSION: The incidence of BAT was low but impacted radiologic interpretation when present. Further studies with a larger group of Hodgkin lymphoma patients are needed to explore the risk factors associated with moderate or severe BAT.

Unexpected extracardiac findings in cardiac computed tomography from neonates to young adults.

BACKGROUND: Modern CT scanners with lower radiation doses have resulted in large numbers of cardiac CTs being performed in children. As seen in adults, pediatric cardiac CT has the potential to demonstrate extracardiac variants and pathology that can occur in conjunction with congenital heart disease (CHD). Prior publications demonstrated a high incidence of extracardiac findings in various locations but the prevalence of urgent unexpected extracardiac findings in children is unknown. OBJECTIVE: The purpose of this study was to describe the incidence, distribution and clinical significance of the extracardiac findings on pediatric cardiac CT at a tertiary referral center. MATERIALS AND METHODS: We reviewed all reports (n = 648) for 554 children through young adults who received a cardiac CT study between Jan. 2, 2018, and March 10, 2020, at our tertiary referral pediatric hospital. We interrogated CT reports for extracardiac findings and categorized them by system (airway, pulmonary, abdomen, malpositioned lines and musculoskeletal). We then subclassified each of these findings by level of clinical importance based upon the need for intervention or treatment into low, medium or high importance. High-importance findings were confirmed with a focused chart review. If a patient had more than one CT with a persistent extracardiac finding, the finding was only counted once. RESULTS: We identified 562 individual extracardiac findings, with one or more extracardiac findings present in 91% of the study population. Extracardiac findings with high clinical importance, requiring urgent attention or intervention, were present in 10% (57/554) of cases. The most common location of extracardiac findings was pulmonary (50%; 280/562), followed by airway (22%; 125/562) and abdomen (9%; 52/562). CONCLUSION: Unexpected highly important extracardiac findings were found in 10% of patients. Therefore, extracardiac structures should be scrutinized for the timely identification of potentially highly important findings.

Computed Tomographic Angiography Provides Reliable Coronary Artery Evaluation in Infants With Pulmonary Atresia Intact Ventricular Septum.

Evaluate the use of coronary CTA as an initial assessment for determining Right Ventricle Dependent Coronary Circulation (RVDCC) in neonates with Pulmonary Atresia with Intact Ventricular Septum (PA IVS). Retrospective review of cases with coronary CTA and compare with available catheter angiography, pathology, surgical reports, and outcomes from Mar 2015 to May 2022. In our cohort of 16 patients, 3 were positive for RVDCC, confirmed by pathologic evaluation, and there was concordance for presence or absence of RVDCC with catheter angiography in 5 patients (4 negatives for RVDCC, 1 positive). Clinical follow up for the 8 patients that underwent RV decompression had no clinical evidence of myocardial ischemia. Our findings suggest that coronary CTA is reliable as first-line imaging for determination of RVDCC in neonates with PA IVS. These findings, if supported by further prospective study, may reserve invasive coronary angiography for cases with diagnostic uncertainty or at the time of necessary transcatheter interventions.

Chest computed tomography findings of ground-glass nodules with enhancing central vessel/nodule in pediatric patients with BMPR2 mutations and plexogenic arteriopathy.

BACKGROUND: Germline mutation in bone morphogenetic protein type II (BMPR2) is the most common cause of idiopathic/heritable pulmonary hypertension in pediatric patients. Despite the discovery of this gene there are no known descriptions of the CT or CT angiography findings in these children. OBJECTIVE: To correlate the clinical presentation, pathology and chest CT findings in pediatric patients with pulmonary hypertension caused by mutations in the BMPR2 gene. MATERIALS AND METHODS: We performed a search to identify pediatric patients with a BMPR2 mutation and CT or CT angiography with the clinical history of pulmonary hypertension. Three pediatric radiologists reviewed the children's CT imaging findings and ranked the dominant findings in order of prevalence via consensus. RESULTS: We identified three children with pulmonary hypertension and confirmed germline BMPR2 mutations, two of whom had undergone lung biopsy. We then correlated the imaging findings with histopathology and clinical course. CONCLUSION: All of our patients with BMPR2 mutations demonstrated a distinct CT pattern of ground-glass nodules with a prominent central enhancing vessel/nodule. These findings correlated well with the pathological findings of plexogenic arteriopathy.

Pediatric mediastinal mass algorithm: A quality improvement initiative to reduce time from presentation to biopsy.

BACKGROUND: Mediastinal masses in children may present with compression of the great vessels and airway. An interdisciplinary plan for rapid diagnosis, acute management, and treatment prevents devastating outcomes and optimizes care. Emergency pretreatment with steroids or radiation is more likely to be administered when care is variable, which may delay and complicate diagnosis and treatment. Strategies to standardize care and expedite diagnosis may improve acute patient safety and long-term outcomes. AIMS: The aim of this quality improvement project was to decrease time from presentation to diagnostic biopsy for children with an anterior mediastinal mass by 50% over 3 years within a tertiary healthcare system. METHODS: This quality improvement project involved a single center with data collected and analyzed retrospectively and prospectively for 71 patients presenting with anterior mediastinal mass between February 2008 and January 2018. The Model for Improvement was utilized for project design and development of a driver diagram and smart aim. An algorithm was implemented to facilitate communication between teams and standardize initial care of patients with mediastinal masses. The algorithm underwent multiple Plan-Do-Study-Act (PDSA) cycles. Data were collected before and after algorithm implementation and between each PDSA cycle. The primary outcome measure included time from presentation to biopsy, which was monitored with a statistical process control chart. Several process measures were evaluated with Student's t-tests including administration of emergency pretreatment. RESULTS: Nineteen patients preintervention and 52 patients postintervention were included in the analysis. Time from presentation to biopsy significantly decreased from 48 h at baseline to 24 h postimplementation. Although not statistically significant, emergency pretreatment decreased from a baseline of 26.3% to 6.7% postimplementation. CONCLUSION: Implementation of a diagnostic and management algorithm coordinating care among multidisciplinary teams significantly reduced time to biopsy for children presenting with mediastinal mass and may result in decreased use of emergent pretreatment.

Deep learning for chest radiograph diagnosis: A retrospective comparison of the CheXNeXt algorithm to practicing radiologists.

BACKGROUND: Chest radiograph interpretation is critical for the detection of thoracic diseases, including tuberculosis and lung cancer, which affect millions of people worldwide each year. This time-consuming task typically requires expert radiologists to read the images, leading to fatigue-based diagnostic error and lack of diagnostic expertise in areas of the world where radiologists are not available. Recently, deep learning approaches have been able to achieve expert-level performance in medical image interpretation tasks, powered by large network architectures and fueled by the emergence of large labeled datasets. The purpose of this study is to investigate the performance of a deep learning algorithm on the detection of pathologies in chest radiographs compared with practicing radiologists. METHODS AND FINDINGS: We developed CheXNeXt, a convolutional neural network to concurrently detect the presence of 14 different pathologies, including pneumonia, pleural effusion, pulmonary masses, and nodules in frontal-view chest radiographs. CheXNeXt was trained and internally validated on the ChestX-ray8 dataset, with a held-out validation set consisting of 420 images, sampled to contain at least 50 cases of each of the original pathology labels. On this validation set, the majority vote of a panel of 3 board-certified cardiothoracic specialist radiologists served as reference standard. We compared CheXNeXt's discriminative performance on the validation set to the performance of 9 radiologists using the area under the receiver operating characteristic curve (AUC). The radiologists included 6 board-certified radiologists (average experience 12 years, range 4-28 years) and 3 senior radiology residents, from 3 academic institutions. We found that CheXNeXt achieved radiologist-level performance on 11 pathologies and did not achieve radiologist-level performance on 3 pathologies. The radiologists achieved statistically significantly higher AUC performance on cardiomegaly, emphysema, and hiatal hernia, with AUCs of 0.888 (95% confidence interval [CI] 0.863-0.910), 0.911 (95% CI 0.866-0.947), and 0.985 (95% CI 0.974-0.991), respectively, whereas CheXNeXt's AUCs were 0.831 (95% CI 0.790-0.870), 0.704 (95% CI 0.567-0.833), and 0.851 (95% CI 0.785-0.909), respectively. CheXNeXt performed better than radiologists in detecting atelectasis, with an AUC of 0.862 (95% CI 0.825-0.895), statistically significantly higher than radiologists' AUC of 0.808 (95% CI 0.777-0.838); there were no statistically significant differences in AUCs for the other 10 pathologies. The average time to interpret the 420 images in the validation set was substantially longer for the radiologists (240 minutes) than for CheXNeXt (1.5 minutes). The main limitations of our study are that neither CheXNeXt nor the radiologists were permitted to use patient history or review prior examinations and that evaluation was limited to a dataset from a single institution. CONCLUSIONS: In this study, we developed and validated a deep learning algorithm that classified clinically important abnormalities in chest radiographs at a performance level comparable to practicing radiologists. Once tested prospectively in clinical settings, the algorithm could have the potential to expand patient access to chest radiograph diagnostics.

Retrospective 4D MR image construction from free-breathing slice Acquisitions: A novel graph-based approach.

PURPOSE: Dynamic or 4D imaging of the thorax has many applications. Both prospective and retrospective respiratory gating and tracking techniques have been developed for 4D imaging via CT and MRI. For pediatric imaging, due to radiation concerns, MRI becomes the de facto modality of choice. In thoracic insufficiency syndrome (TIS), patients often suffer from extreme malformations of the chest wall, diaphragm, and/or spine with inability of the thorax to support normal respiration or lung growth (Campbell et al., 2003, Campbell and Smith, 2007), as such patient cooperation needed by some of the gating and tracking techniques are difficult to realize without causing patient discomfort and interference with the breathing mechanism itself. Therefore (ventilator-supported) free-breathing MRI acquisition is currently the best choice for imaging these patients. This, however, raises a question of how to create a consistent 4D image from such acquisitions. This paper presents a novel graph-based technique for compiling the best 4D image volume representing the thorax over one respiratory cycle from slice images acquired during unencumbered natural tidal-breathing of pediatric TIS patients. METHODS: In our approach, for each coronal (or sagittal) slice position, images are acquired at a rate of about 200-300ms/slice over several natural breathing cycles which yields over 2000 slices. A weighted graph is formed where each acquired slice constitutes a node and the weight of the arc between two nodes defines the degree of contiguity in space and time of the two slices. For each respiratory phase, an optimal 3D spatial image is constructed by finding the best path in the graph in the spatial direction. The set of all such 3D images for a given respiratory cycle constitutes a 4D image. Subsequently, the best 4D image among all such constructed images is found over all imaged respiratory cycles. Two types of evaluation studies are carried out to understand the behavior of this algorithm and in comparison to a method called Random Stacking - a 4D phantom study and 10 4D MRI acquisitions from TIS patients and normal subjects. The 4D phantom was constructed by 3D printing the pleural spaces of an adult thorax, which were segmented in a breath-held MRI acquisition. RESULTS: Qualitative visual inspection via cine display of the slices in space and time and in 3D rendered form showed smooth variation for all data sets constructed by the proposed method. Quantitative evaluation was carried out to measure spatial and temporal contiguity of the slices via segmented pleural spaces. The optimal method showed smooth variation of the pleural space as compared to Random Stacking whose behavior was erratic. The volumes of the pleural spaces at the respiratory phase corresponding to end inspiration and end expiration were compared to volumes obtained from breath-hold acquisitions at roughly the same phase. The mean difference was found to be roughly 3%. CONCLUSIONS: The proposed method is purely image-based and post-hoc and does not need breath holding or external surrogates or instruments to record respiratory motion or tidal volume. This is important and practically warranted for pediatric patients. The constructed 4D images portray spatial and temporal smoothness that should be expected in a consistent 4D volume. We believe that the method can be routinely used for thoracic 4D imaging.