BODY IMAGING
Imaging characteristics of nephrogenic rests versus small Wilms tumors: a report from the Children’s Oncology Group Study AREN03B2
Sandberg JK, Chi YY, Smith EA et al.
AJR 2020;214:987-994.
https://doi.org/10.2214/AJR.19.22301
Question(s) What imaging characteristics can be used to distinguish nephrogenic rests from Wilms tumors?
Design Retrospective study from July 2006 to August 2016
Setting All cases from the Children’s Oncology Group (COG) AREN03B2 study
Participants 52 patients younger than 5 years of age were enrolled in the AREN03B2 trial met one of the two inclusion criteria: either they had a central pathologic diagnosis of a nephrogenic rest or a small (less than 5 cm in diameter) Wilms tumor.
Intervention Imaging evaluation of lesions on contrast-enhanced CT or MRI were confirmed pathologically.
Outcomes Imaging characteristics, such as tumor diameter, shape (spherical or non-spherical), exophytic appearance, interface of lesion vs. adjacent kidney, density (homogenous or inhomogeneous), and location (e.g. cortex, medulla, or corticomedullary junction).
Main Results The median diameter of nephrogenic rests was 1.3 and of Wilms tumor was 3.2 cm (p<0.001). Wilms tumors were more commonly spherical in shape (p<0.001) and exophytic (p<0.001) compared to rests. The optimal size threshold for distinguishing between nephrogenic rests and Wilms tumors was 1.75 cm (based on ROC analysis).
Commentary While there are several overlapping imaging characteristics between nephrogenic rests and Wilms tumors, the study suggests that larger tumor (>1.75 cm), exophytic appearance and spherical shape may help distinguish Wilms tumors from rests in children younger than age 5 years. Limitations included inability to confirm all multifocal lesions pathologically, particularly in patients with greater than 2 lesions (as surgical excision was at the discretion of the surgeon), relatively small sample size due to strict inclusion criteria, and variability in MRI techniques.
The use of computed tomography versus clinical acumen in diagnosing appendicitis in children: a two-institution international study
El-Gohary Y, Molina M, Chang J et al.
Journal of Pediatric Surgery (2020). Journal Pre-Proof.
https://doi.org/10.1016/j.jpedsurg.2020.09.061.
Question(s) What are the differences in imaging use in children with suspected appendicitis between a children’s hospital in the United States versus in Spain?
Design Retrospective study from 2015 to 2017
Setting Multi-institutional (Le Bonheur Children’s Hospital, Memphis, Tennessee and Hospital Universitario Virgen Del Rocio, Sevilla, Spain)
Participants 1,952 children (1,288 in the U.S. center)
Intervention Computed tomography (CT) compared to clinical acumen and ultrasound (US), with surgical pathology as the reference standard.
Outcomes Primary outcomes included rates of imaging modalities used (US only, CT only, CT+US), rate of no imaging used, accuracy rates of diagnosis. Secondary outcomes included sensitivity, specificity, positive predictive value, and negative predictive value of the overall diagnostic approach at each center.
Main Results The most common imaging modality used in the U.S. was US only (39.9%), followed by CT only (39.3%), and CT+US (19.3%). Clinical acumen in the U.S. was used 1.6% of the time. In Spain, clinical acumen was most commonly used (48.6%), followed by US only (48.5%), US+CT (2.4%), and CT only (0.5%). Sensitivity, specificity, PPV, NPV, and accuracy rates for diagnosis of acute appendicitis in children were 99.8%, 75.2%, 93.8%, 98.9%, and 94.7% for the U.S. hospital and 92.5%, 96.7%, 94.7%, 95.2%, and 95.1% for the Spanish hospital respectively. The negative appendectomy rate in the US center was 6.1% and 4.9% in the Spanish center while the missed appendicitis rate was 0.2% in the US center and 7.5% in the Spanish center.
Commentary The authors conclude that the accuracy rates between the U.S. and Spain were comparable in this study; nevertheless, the center in Spain achieved this accuracy rate with much less reliance on imaging. The trade-off is a lower rate of missed appendicitis at the U.S. center. Pertinently, the article also briefly acknowledges issues of cost-effectiveness, including imaging cost, hospitalization/observation costs, and litigation costs. One relevant limitation was that the study did not compare rates of perforation, which is a major determinant of clinical outcome and costs.
CHEST IMAGING
Value of multi-slice spiral computed tomography for diagnosis of tracheobronchial foreign body aspiration in children: 5-year retrospective study
Shen J, Huang L, and Hao C.
Pediatrics International 2020;62:1184-1188.
Question(s) What is the diagnostic value of CT for tracheobronchial foreign body (TFB) aspiration in children?
Design Retrospective study from July 2008 to June 2013
Setting Single center (Children’s Hospital of Soochow University, Suzhou, China)
Participants 382 children, who underwent chest CT and bronchoscopy. 81 of these patients underwent chest radiographs.
Intervention Multi-slice, spiral CT imaging of the entire chest (parameters: pitch, 0.9; 120 kV; 300 mA; and section thickness, 2.0 mm) was reviewed by 2 pediatric radiologists
Outcomes Demographics (sex and age), chest CT presentations, location and nature of the TFBs, and sensitivity of CT
Main Results The majority of patients were male (67%) and between the age of 1 and 2 years (69%). Presentation at chest CT was predominantly fewer than 5 days after the aspiration incident (66%). The majority of aspirated foreign bodies were vegetation items (96%), mostly peanut kernels (48%) and sunflower seeds (26%). In descending order, the locations of the foreign bodies were the left main bronchus (51%), right main bronchus (44%), and trachea (5%). Indirect findings that suggested the presence of an aspirated foreign body included emphysema (42%), localized obstruction (50%), mediastinal shift and bilateral mediastinal shadow (31%), subcutaneous emphysema (0.8%), and bronchiectasis (0.5%). The sensitivity of CT for direct detection of TFBs was 94% and up to 96% indirectly (compared to radiographs at 43.2%).
Commentary This study showed that chest CT has high sensitivity for detection of TFBs. However, there are a few major disadvantages of CT include ionizing radiation and its inability to remove the foreign body. Additional diagnostic parameters, such as specificity, would have been helpful to include in the study’s analysis.
The use of lung ultrasound compared to chest X-rays to diagnose pneumothorax following Nuss procedure for pectus excavatum repair in children
Bonfiglio R, Disma N, Stagnaro N et al.
Pediatric Anesthesia 2020;00:1-9.
https://doi.org/10.1111/pan.14007
Question(s) How does ultrasonography compare with radiography for evaluation of pneumothorax status post Nuss procedure?
Design Inter-reader prospective observational study
Setting Single center (Istituto Giannina Gaslini, Genoa, Italy)
Participants 68 pediatric patients
Intervention Chest ultrasound (US) compared to chest radiographs
Outcomes Inter-reader agreement (Cohen’s kappa) among 2 anesthesiologists (A1, A2), a medical student (A3), and 12 total radiologists (a consultant radiologist R1, and the remaining 11 were collectively labeled as R2).
Main Results Overall agreement among the readers (A1 vs. A2 vs. A3 vs. R1 vs. R2) was fair (k=0.32). Stratified analysis showed moderate inter-reader agreement (k=0.58) among the US users (A1 vs. A2 vs. A3). Agreement was fair (k=0.39) between the chest radiograph readers (R1 vs. R2). The authors conclude that US interpretation was superior to that of chest radiograph analysis, based on the inter-reader agreements.
Commentary While the inter-reader agreement among the US users was reportedly better than that of the X-ray users (moderate vs. fair), the study design was not set up to truly compare US to X-ray for detection of pneumothorax after Nuss procedure. That is, the absence of radiology interpretation of US in addition to radiology analysis of X-ray limits comparison of the two modalities, resulting in a major limitation that the study only evaluated inter-reader agreement without a reference standard. Additionally, the study did not assess the clinical importance of detected pneumothoraces.
INTERVENTIONAL RADIOLOGY
Managing systemic venous occlusions in children
Gill AE and Shivaram GM.
CVIR Endovascular 2020;3(59):1-11.
Question(s) What are the etiologies, presentations, diagnostic workup, general technical considerations of recanalization, and procedural complications related to pediatric systemic venous occlusive disease?
Design Review Article
Setting Single institution (Emory University School of Medicine, Atlanta, Georgia)
Commentary This review article includes a brief background of systemic venous occlusive disease in children, and it describes the imaging workup helpful in such clinical situations (including Doppler ultrasound, contrast-enhanced MRI, dynamic MR venography, and contrast-enhanced CT). The primary focus of the article is to provide insight on the endovascular management of both acute and chronic venous occlusions, including procedural equipment, technique, and potential complications encountered.
Spinal ultrasound after failed landmarked-based lumbar puncture: a single institutional experience
Jeyakumar A, Weaver JJ, Chick JFB et al.
Pediatric Radiology 2020. Published online 17 September 2020.
https://doi.org/10.1007/s00247-020-04831-w.
Question(s) Does pre-procedure ultrasound (US) improve technical success and safety of US-guided (USG) lumbar puncture (LP) after failed landmarked-based (bedside) LP?
Design Retrospective cohort study from Oct 2017 to Apr 2020
Setting Single center (University of Washington/Seattle Children’s Hospital, Seattle, Washington)
Participants 47 children
Intervention Urgent USG LP (need to be performed within 24h)
Outcomes Technical successes, adverse events, relevant abnormalities identified on pre-procedural US
Main Results All 36 patients who underwent an image-guided LP had a successful tap. The remaining 11 children (23%) referred to interventional radiology did not undergo image-guided LP due to paucity of cerebral spinal fluid (CSF) and/or improvement of clinical condition, which obviated the need for LP. 15 LPs (42%) resulted in a traumatic tap, which is a “mild” adverse event (as classified by the Society of Interventional Radiology/SIR). No moderate or severe adverse events were documented. Abnormalities identified on pre-procedural US included paucity of CSF volume and epidural hematoma.
Commentary The study concludes that focused spinal US after failed bedside LP attempt can aid in technical success and augment patient selection for image-guided LP by identifying procedure-related complications, such as paucity of CSF volume and/or presence of an epidural hematoma. Limitations of the paper include small sample size/focused patient inclusion criteria, difficulty in including complications based on patient reporting (e.g. post-LP headache), and difficulty in distinction of traumatic tap truly related to IR-performed LP versus pre-existing blood from bedside LP attempt.
MUSCULOSKELETAL RADIOLOGY
Clinical application of artificial intelligence-assisted diagnosis using anteroposterior pelvic radiographs in children with developmental dysplasia of the hip
Zhang SC, Sun J, Liu CB et al.
The Bone & Joint Journal 2020;102-B(11)1574-1581.
Question(s) How can we develop a deep learning system for diagnosing developmental dysplasia of the hip (DDH) on anteroposterior (AP) pelvic radiographs, and what is the feasibility of its clinical application?
Design Retrospective study from April 2014 to December 2018
Setting Single center (Anhui Provincial Children’s Hospital and University of Science and Technology of China, Hefei, Anhui, China)
Participants 10,219 AP pelvic radiographs were used to train and optimize the deep learning system. 1,138 radiographs were then used to compare the diagnoses made by the deep learning system and clinicians
Intervention Deep learning system compared to clinicians (which included 8 pediatric Orthopedic attending surgeons, 2 pediatric Orthopedic associate chief physicians and 2 pediatric associate chief radiologists)
Outcomes Accuracy, sensitivity, specificity, and consistency of acetabular index measurements as evaluated by Bland-Altman plots
Main Results The area under the receiver operating characteristic (ROC) curve (AUC) for the deep learning system for diagnosing hip dislocation was 0.975. The sensitivity and specificity for the deep learning system were 95.5% and 99.5%, respectively. Compared to clinician-led diagnoses, the deep learning system was more consistent in measuring acetabular indices on radiographs of non-dislocated and dislocated hips (-3.27O-2.94O vs. -7.36O-5.36O, p<0.001).
Commentary The study describes the experience with development of a deep learning system for evaluation of DDH on radiographs and demonstrates its performance compared to clinicians, which could improve their current screening referral process. Limitations of the study include its single-centered nature, differences in radiographic annotations, and limited evaluation of hip radiographs in older children and children with severe dislocations.
NEURORADIOLOGY
Image-guided percutaneous sclerotherapy of venous malformations of the head and neck: clinical and MR-based volumetric mid-term outcome
Vollherbst DF, Gebhart P, Kargus S et al.
PLoS ONE 2020;15(10): e0241347.
https://doi.org/10.1371/journal.pone.0241347.
Question(s) What are the clinical and MRI-based volumetric mid-term outcomes after image-guided percutaneous sclerotherapy (PS) with 95% ethanol or polidocanol of venous malformations (VM) of the head and neck.
Design Retrospective analysis from Sept 2010 to May 2015 of a prospectively maintained database
Setting Single center (Kepler University Hospital: Kepler Universitatsklinikum GmbH, Austria)
Participants 27 patients (mean age 29.2 years) with a total of 51 treatment sessions
Intervention Image-guided PS
Outcomes Clinical outcome as categorized: 0 = worse, 1 = unchanged, 2 = minor improvement, 3 = major improvement (performed by both a neuroradiologist and a surgeon and included visual inspection, subjective patient symptoms, and/or impression of parents and/or physicians). MRI volumetric analysis performed after 4 weeks of treatment using a 3T system with volume assessment based on pre- and post-treatment imaging (response to therapy was categorized into 0 = worse (>10% increase), 1 = unchanged (<10% decrease or up to 10% increase), 2 = minor improvement (>10% to 25% decrease), 3 = intermediate improvement (>25% to 50% decrease), and 4 = major improvement (>50% decrease). The VMs were then classified as “large” (larger than median volume) or “small” (smaller than median volume) VMs. The rate of permanent complications was also evaluated.
Main Results Clinical outcome was worse for 7.4% of patients and unchanged for 3.7% patients. There was minor clinical improvement for 7.4% of patients and major improvement of 81.5%. The volumes of the treated VMs were worse in 7.4% patients and unchanged in 14.8%. Minor improvement in VM volume was seen in 22.2%, intermediate improvement in 44.4%, and major improvement in 11.1%. Permanent complications occurred in 3.7%.
Commentary The authors conclude that PS can be an effective therapy for patients with VMs of the head and neck, based on clinical improvement and decreased size of the VMs. They note that relief of symptoms does not necessarily require substantial volume reduction. A potential limitation was underestimation of PS success with regard to volumetric analysis as patients who became asymptomatic did not receive follow up MRI.
MRI findings of central nervous system involvement in children with haemophagocytic lymphohistiocytosis: correlation with clinical biochemical tests
Ma W, Li XJ, Li W et al.
Clinical Radiology. Article in Press.
https://doi.org/10.1016/j.crad.2020.09.009.
Question(s) What are the MRI features of hemophagocytic lymphohistiocytosis (HLH) with central nervous system (CNS) involvement? How do these features correlate with clinical biochemical tests?
Design Retrospective study from Jan 2012 to June 2019
Setting Single center (Children’s Hospital of Chongqing Medical University, Chongqing, China)
Participants 118 children with HLH-CNS, grouped according to MRI findings
Intervention MRI of the brain was performed and analyzed by 2 pediatric radiologists (both with >10 years of neuroradiology experience) in conjunction with 1 pediatric neurologist with 11 years of experience. 7 biochemical tests were obtained including: cerebrospinal fluid (CSF) microprotein content, CSF cell count, serum aspartate aminotransferase (AST), serum lactate dehydrogenase (LDH), plasma fibrinogen (FIB), platelet count (PLT), and activated partial thromboplastin time (APTT).
Outcomes Brain MRI findings of “normal” (Group 1), “diffuse volume loss” (Group 2), and “brain parenchymal lesions” (Group 3). Pairwise comparison was made and Spearman’s rank correlation coefficient (r) was calculated for the clinical biochemical tests and the three groups.
Main Results There was statistical differences when comparing the biochemical values among the three groups, except for CSF cell count. There was positive correlation with the MRI groups for the following biochemical tests: CSF cell count (r=0.193, p=0.036), CSF microprotein content (r=0.379, p<0.001), AST (r=0.521, p<0.001), LDH (r=0.514, p<0.001), and APTT (r=0.326, p<0.001). There was negative correlation with the MRI groups for PLT (r=-0.633, p<0.001) and FIB (r=-0.258, p=0.005).
Commentary The authors conclude that brain MRI findings of HLH with CNS involvement correlate well with key biochemical tests, which suggests that brain MRI may help to evaluate the severity and clinical outcomes of HLH; however, more studies are necessary to examine its specific role and how it can complement biochemical tests. However, there are a few important limitations, including the use of two different field-intensity systems of 1.5T vs. 3T (which may lead to unknown errors of lesion classification), and the small proportion of contrast-enhanced MRI for 29 children (which could lead to missing subtle lesions).
Neuroimaging in paediatric mild traumatic brain injury: a systematic review
Rausa VC, Shapiro J, Seal ML et al.
Neuroscience and Biobehavioral Reviews 2020;118:643-653.
https://doi.org/10.1016/j.neubiorev.2020.08.017.
Question(s) What is the relationship between advanced neuroimaging and post-concussive symptoms (PCS) in children? What is the quality of the evidence?
Design Systematic review
Setting Single center (The Royal Children’s Hospital, Melbourne). A literature search of MEDLINE, PsycINFO, EMBASE, and PubMed was performed.
Participants 58 imaging studies met inclusion criteria (totaling 2047 mild traumatic brain injury patients and 1243 control patients)
Outcomes Imaging findings of various MR techniques in neuroimaging, such as T1-/T2-weighted imaging, susceptibility weighted imaging (SWI), diffusion weighted imaging (DWI), functional MRI (fMRI), cerebral blood flow/volume (including perfusion weighted imaging, arterial spin labelling, and phase contrast angiography), and spectroscopy. Additionally, the study assessed the quality of the imaging studies using the Newcastle Ottawa Scale (NOS) and the Committee on Best Practices in Data Analysis and Sharing (COBIDAS) checklists.
Main Results T1-/T2-weighted images have mixed findings with regards to grey matter changes and how they relate to PCS. There may be a positive trend between SWI findings and injury severity. Some studies showed that various metrics of DWI have potential to add prognostic value in combination with clinical indicators of injury severity. Some studies demonstrated an association between metabolic activity (changes in blood oxygen level dependent/BOLD signal) and time of return to play in athletes, in addition to analysis of task-related cognitive activation. Cerebral blood flow and volume analyses were variable in patients with mild traumatic brain injury. Few studies showed increased GABA/Creatinine ratio in the frontal lobe after 1st time-concussions and decreased NAA/Creatinine and NAA/Choline ratios months after injury.
Commentary This systematic review states that the current literature provides insufficient evidence to draw meaningful conclusions due to differences in methodologies and relatively small sample sizes.
References
