Artificial Intelligence Algorithm Improves Radiologist Performance in Skeletal Age Assessment: A Prospective Multicenter Randomized Controlled Trial
Background Previous studies suggest that use of artificial intelligence (AI) algorithms as diagnostic aids may improve the quality of skeletal age assessment, though these studies lack evidence from clinical practice. Purpose To compare the accuracy and interpretation time of skeletal age assessment on hand radiograph examinations with and without the use of an AI algorithm as a diagnostic aid. Materials and Methods In this prospective randomized controlled trial, the accuracy of skeletal age assessment on hand radiograph examinations was performed with (n = 792) and without (n = 739) the AI algorithm as a diagnostic aid. For examinations with the AI algorithm, the radiologist was shown the AI interpretation as part of their routine clinical work and was permitted to accept or modify it. Hand radiographs were interpreted by 93 radiologists from six centers. The primary efficacy outcome was the mean absolute difference between the skeletal age dictated into the radiologists' signed report and the average interpretation of a panel of four radiologists not using a diagnostic aid. The secondary outcome was the interpretation time. A linear mixed-effects regression model with random center- and radiologist-level effects was used to compare the two experimental groups. Results Overall mean absolute difference was lower when radiologists used the AI algorithm compared with when they did not (5.36 months vs 5.95 months; P = .04). The proportions at which the absolute difference exceeded 12 months (9.3% vs 13.0%, P = .02) and 24 months (0.5% vs 1.8%, P = .02) were lower with the AI algorithm than without it. Median radiologist interpretation time was lower with the AI algorithm than without it (102 seconds vs 142 seconds, P = .001). Conclusion Use of an artificial intelligence algorithm improved skeletal age assessment accuracy and reduced interpretation times for radiologists, although differences were observed between centers. Clinical trial registration no. NCT03530098 Â© RSNA, 2021 Online supplemental material is available for this article. See also the editorial by Rubin in this issue.
Congenital lung lesions: a radiographic pattern approach
Congenital lung malformations represent a spectrum of abnormalities that can overlap in imaging appearance and frequently coexist in the same child. Imaging diagnosis in the neonatal period can be challenging; however, the recognition of several archetypal radiographic patterns can aid in narrowing the differential diagnosis. Major radiographic archetypes include (1) hyperlucent lung, (2) pulmonary cysts, (3) focal opacity and (4) normal radiograph. Here we review the multimodality imaging appearances of the most commonly seen congenital lung malformations, categorized by their primary imaging archetypes. Along with the congenital lung malformations, we present several important imaging mimickers.
Sporadic Burkitt Lymphoma Presenting with Middle Cranial Fossa Masses with Sphenoid Bony Invasion and Acute Pancreatitis in a Child [Case Report]
Acute pancreatitis in children is usually due to infection, trauma, or anatomical abnormalities and is rarely due to obstruction from malignancy. Sporadic Burkitt lymphoma (BL) is an aggressive non-Hodgkin B-cell lymphoma that usually involves the bowel or pelvis, with isolated cases presenting as acute pancreatitis. We report a case of BL in a 12-year-old male presenting as acute pancreatitis with obstructive jaundice and a right middle cranial fossa mass invading the sphenoid bone. The common bile duct in this case was dilated to 21â€‰mm in diameter on abdominal ultrasound and to 26â€‰mm on magnetic resonance cholangiopancreatography (MRCP), significantly greater than any value reported in the literature for BL. Given the rapidly progressing nature of BL, we emphasize the importance of recognizing heterogeneous presentations of this disease to improve patient survival. We also conclude that it is important to consider malignancy in a child with acute pancreatitis, particularly in the presence of obstructive jaundice or multisystem involvement. Other Presentations. This case report has no prior publications apart from the abstract being accepted to the 2020 SIOP (International Society of Pediatric Oncology) meeting and 2020 ASPHO conference (canceled due to the COVID-19 pandemic) and subsequently published as an abstract only in Pediatric Blood and Cancer. We have also presented the abstract as a poster presentation at our institution's (NYU Langone Hospital-Long Island, previously known as NYU Winthrop) annual research day conference in 2020.
Sporadic Burkitt Lymphoma Presenting With Sphenoid Bone Invasion and Acute Pancreatitis in a Child [Meeting Abstract]
Ovarian neoplasms of childhood
Ovarian neoplasms are rare in children. Although usually asymptomatic, they sometimes present with abdominal pain, abdominal distension or palpable mass. The distribution of neoplasms in the pediatric population is different from in adults; benign mature cystic teratoma is the most common ovarian tumor in children. Radiologists should be familiar with the variable sonographic, CT and MRI findings of ovarian neoplasms. Although the less frequently encountered ovarian malignancies cannot be reliably distinguished by imaging alone, it does play an important role in workup. This review discusses the imaging and relevant clinical manifestations of the more commonly encountered pediatric ovarian neoplasms.
Visualization of the normal appendix in children: feasibility of a single contrast-enhanced radial gradient recalled echo MRI sequence
BACKGROUND:Magnetic resonance imaging (MRI) assessment for appendicitis is limited by exam time and patient cooperation. The radially sampled 3-dimensional (3-D) T1-weighted, gradient recalled echo sequence (radial GRE) is a free-breathing, motion robust sequence that may be useful in evaluating appendicitis in children. OBJECTIVE:To compare the rate of detection of the normal appendix with contrast-enhanced radial GRE versus contrast-enhanced 3-D GRE and a multi-sequence study including contrast-enhanced radial GRE. MATERIALS AND METHODS/METHODS:This was a retrospective study of patients ages 7-18Â years undergoing abdominal-pelvic contrast-enhanced MRI between Jan. 1, 2012, and April 1, 2016. Visualization of the appendix was assessed by consensus between two pediatric radiologists. The rate of detection of the appendix for each sequence and combination of sequences was compared using a McNemar test. RESULTS:The rate of detection of the normal appendix on contrast-enhanced radial GRE was significantly higher than on contrast-enhanced 3-D GRE (76% vs. 57.3%, P=0.003). The rate of detection of the normal appendix with multi-sequence MRI including contrast-enhanced radial GRE was significantly higher than on contrast-enhanced 3-D GRE (81.3% vs. 57%, P<0.001). There was no significant difference between the rate of detection of the normal appendix on contrast-enhanced radial GRE alone and multi-sequence MRI including contrast-enhanced radial GRE (76% vs. 81.3%, P=0.267). CONCLUSION/CONCLUSIONS:Contrast-enhanced radial GRE allows superior detection of the normal appendix compared to contrast-enhanced 3-D GRE. The rate of detection of the normal appendix on contrast-enhanced radial GRE alone is nearly as good as when the contrast-enhanced radial GRE is interpreted with additional sequences.
Multi-institutional implementation of an automated tool to predict pediatric skeletal bone age: How we did it [Meeting Abstract]
Purpose or Case Report: Skeletal bone age assessment is a common clinical practice to investigate endocrinology, genetic and growth disorders of children. Clinical interpretation and bone age analyses are time-consuming, labor intensive and often subject to inter-observer variability. Bone age prediction models developed with deep learning methodologies can be leveraged to automate bone age interpretation and reporting. The bone age model developed at our institution was offered to interested health systems and institutions to implement and validate the model. This study discusses the logistical, technical, and clinical issues encountered with this model implementation. Methods & Materials: After IRB approval, multiple U.S. based radiology departments were solicited to adopt and validate the Stanford University bone age model. A total of 8 institutions (4 standalone pediatric hospitals and 4 academic radiology departments) agreed to partner with the primary investigators. IRBs at each institution were required in addition to registration with ClinicalTrials.gov registry. Standardization of the data use agreements was performed. Patient data and protected health information data was retained at each institution. Technical requirements included model hosting at each institution and integration to send images to the model server and results to the interpreting radiologists.
Result(s): Multiple logistical, technical, and clinical issues were encountered. IRBs at the various institutions had different requirements including waiving patient consent. Technical differences between institutions included model hosting, PACS integrations, interfaces with the reporting system, and image preprocessing. Clinical differences included report templates, calculation of bone age standard deviation, use of Brush foundation, and ability to directly send bone predictions to the reporting system (versus displaying the results as a separate interface). The bone age model was successfully implemented at 7 institutions and approximately 190 studies have been evaluated.
Conclusion(s): There are myriad challenges to implementing and validating models developed with deep learning methodologies. As models are developed for various clinical use cases including bone age assessment, it will be incumbent on radiology practices and health information systems to integrate these models into clinical practice
Visualization of the normal appendix in children on MRI using radial vibe - A contrast enhanced, free-breathing, radially sampled, 3D T1-weighted, gradient-echo sequence [Meeting Abstract]
Disclosures: All authors have disclosed no financial interests, arrangements or affiliations in the context of this activity. Purpose or Case Report: Current MRI evaluation of appendicitis is limited by duration of examination and patient cooperation. The radially sampled 3D T1 weighted, gradient recalled echo sequence (radial VIBE) is a free-breathing, motion robust sequence that may prove useful in the evaluation of appendicitis in children. The purpose of this investigation is to determine the detection rate of the normal appendix with contrast enhanced (CE) radial VIBE alone compared with CE conventional 3D gradient recalled echo volumetric interpolated breath-hold examination (conventional VIBE) alone and multi-sequence abdominal pelvic MRI including CE radial VIBE. Methods& Materials:We conducted a retrospective, HIPAA compliant and IRB approved study of patients between 7 and 18 years of age who underwent an abdominal and pelvic contrast enhanced MRI between January 1, 2012 and April 1, 2016. Patients with active right lower quadrant inflammation, pelvic masses, or history of appendectomy were excluded. Visualization of the appendix was assessed by two pediatric radiologists with Certificates of Added Qualification by consensus on the following sequences: CE radial VIBE only, CE conventional VIBE only, and multi-sequence MRI which included CE radial VIBE and at least an axial or coronal single shot fast spin echo (SSFSE) or axial T2 weighted spin echo with fat suppression. The detection rates of the appendix for each sequence or combination of sequences were compared with a McNemar test. Results: Ninety-six patients met inclusion criteria. The detection rate of the normal appendix on CE radial VIBE was significantly higher than on CE conventional VIBE (76% vs 57.3%, p=0.003). The detection rate of the normal appendix with multi-sequence MRI was significantly higher than on CE conventional VIBE (81.3% vs 57%, p<0.001). There was no significant difference between the detection rate of the normal appendix on CE radial VIBE and multi-sequence MRI (76% vs 81.3%, p=0.267). When the appendix was not visualized on the CE radial VIBE (n=23) but detected on the multi-sequence MRI (n=9), it was most often visualized on SSFSE (n=8). Conclusions: CE radial VIBE allows superior detection of the normal appendix compared to CE conventional VIBE. The detection rate of the normal appendix on CE radial VIBE alone is nearly as good as when the CE radial VIBE is interpreted with additional sequences
Virtual radiology rounds: adding value in the digital era
BACKGROUND: To preserve radiology rounds in the changing health care environment, we have introduced virtual radiology rounds, an initiative enabling clinicians to remotely review imaging studies with the radiologist. OBJECTIVE: We describe our initial experience with virtual radiology rounds and referring provider impressions. MATERIALS AND METHODS: Virtual radiology rounds, a web-based conference, use remote sharing of radiology workstations. Participants discuss imaging studies by speakerphone. Virtual radiology rounds were piloted with the Neonatal Intensive Care Unit (NICU) and the Congenital Cardiovascular Care Unit (CCVCU). Providers completed a survey assessing the perceived impact and overall value of virtual radiology rounds on patient care using a 10-point scale. Pediatric radiologists participating in virtual radiology rounds completed a survey assessing technical, educational and clinical aspects of this methodology. RESULTS: Sixteen providers responded to the survey; 9 NICU and 7 CCVCU staff (physicians, nurse practitioners and fellows). Virtual radiology rounds occurred 4-5 sessions/week with an average of 6.4 studies. Clinicians rated confidence in their own image interpretation with a 7.4 average rating for NICU and 7.5 average rating for CCVCU. Clinicians unanimously rated virtual radiology rounds as adding value. NICU staff preferred virtual radiology rounds to traditional rounds and CCVCU staff supported their new participation in virtual radiology rounds. Four of the five pediatric radiologists participating in virtual radiology rounds responded to the survey reporting virtual radiology rounds to be easy to facilitate (average rating: 9.3), to moderately impact interpretation of imaging studies (average rating: 6), and to provide substantial educational value for radiologists (average rating: 8.3). All pediatric radiologists felt strongly that virtual radiology rounds enable increased integration of the radiologist into the clinical care team (average rating: 8.8). CONCLUSION: Virtual radiology rounds are a viable alternative to radiology rounds enabling improved patient care and education of providers.
Reliability of the new urinary tract dilation (UTD) Classification system for the evaluation of postnatal urinary tract dilation [Meeting Abstract]
Purpose or Case Report: To evaluate the reliability of the UTD classification system Table A. Cross-tabulation of results summarizing inter-reader agreement. There are three distinct reader pairs: score 1 is the score from the arbitrarily designated first reader in each pair and score 2 is from the remaining reader in each pair. Numbers in red denote instances of disagreement. Methods &Materials: This IRB approved, retrospective study included 129 renal ultrasound examinations performed from May 2010 - May 2015 in patients less than 6 months of age for the clinical indication of prenatal hydronephrosis identified by key word search in PACS. Three pediatric radiologists independently reviewed each study for the following: anterior posterior renal pelvic diameter (APRPD), central calyceal dilation (CCD), peripheral calyceal dilation (PPD), renal parenchymal appearance (PA), renal parenchymal thickness (PT), ureteral abnormality, and bladder abnormality. Readers assigned each study a UTD category (normal, UTD P1, UTD P2, UTD P3). Inter-rater percent agreement for individual criteria and overall UTD categorization was assessed. Results: There was overall good inter-reader agreement in assessment of individual criteria (APRPKD, PA, PT, ureter, and bladder) ranging from 85.3 to 96.1% for 3 reader pairs. Inter-reader agreement for CCD and PCD was slightly lower, ranging from 69.0 to 97.7%. Inter-reader agreement for overall risk assesment ranged from 50.4 to 67.4%. Agreement across 3 readers was 48.8% for CCD, 64.3% for PCD, and 37.2% for overall risk stratification. Conclusions: The new UTD classification system is intended to guide clinical management of postnatal urinary tract dilation. For it to be widely accepted and useful, users need to apply it with precision and accuracy. Poor agreement for categorization of risk assessment among our experienced readers suggests that further clarification of the system or training for users is necessary for its optimal use in clinical practice. (Table presented)