Elanchezhian Somasundaram

Background: Endotracheal tubes (ETTs) are critical life-support devices for mechanically ventilated pediatric patients, yet automated ETT assessment on pediatric chest radiographs (CXRs) remains limited. Objective: To develop and evaluate a two-stage deep learning pipeline for automated detection and localization of ETTs on pediatric CXRs. Materials and methods: This retrospective study included 1,000 pediatric CXRs (476 ETT-positive, 524 ETT-negative) acquired in 2021 at a single institution. ETT segmentation masks and distal tip coordinates were annotated by trained analysts and verified by pediatric radiologists. A two-stage pipeline consisting of a ResNet classification model followed by a U-Net segmentation model was developed for ETT detection and localization. Performance was evaluated on a held-out test set using multiple metrics, including the area under the receiver operating characteristic curve (AUROC) and the mean absolute error (MAE), with 95% confidence intervals (CI). Inter-observer variability was assessed as a reference for localization performance. Results: Inter-observer variability for ETT tip localization was 2.01 mm MAE on the held-out test set. The pipeline achieved an AUROC of 0.994 (95% CI 0.986, 1.000) for ETT detection. For localization, the pipeline achieved a MAE of 6.59 mm (95% CI 5.19, 8.21 mm). Incorporating the classification stage substantially reduced false-positive segmentations from 17 to 3 among ETT-negative CXRs compared with the standalone segmentation model. Conclusion: A two-stage deep learning pipeline demonstrated high performance for automated ETT detection and promising performance for tip localization on pediatric CXRs in this single-center pilot study. Further evaluation in larger and external pediatric cohorts is needed to assess generalizability.

Background: Fetal magnetic resonance imaging (MRI)-derived lung volume measurements are used for prenatal risk stratification in conditions associated with pulmonary hypoplasia, but manual segmentation is time-intensive. Objective: To evaluate whether automated fetal MRI lung segmentation enables rapid and reliable volumetric assessment across diverse pulmonary hypoplasia phenotypes. Materials and methods: In this retrospective study, fetal MRI examinations performed from 2016 to 2025 with expert-adjudicated lung segmentations obtained during clinical care were used for model development and testing. Automated segmentation pipelines were developed for total lung volume (TLV) and percent predicted lung volume (PPLV). Performance was assessed using Dice similarity coefficient (DSC), absolute percentage error (APE), Bland-Altman analysis, intraclass correlation coefficient (ICC), case-level risk-category agreement, and inference time. Lowest-performing TLV outputs were adjudicated by a fetal radiologist to simulate a human-in-the-loop workflow. Results: The final dataset included 906 fetal MRI examinations with complete and recoverable lung segmentations, acquired at a mean gestational age of 27.1 weeks, range 16-38 weeks. Automated TLV estimation showed excellent agreement with manual volumetry (ICC, 0.955; mean DSC, 0.81±0.10; median APE 11.1%). Automated PPLV estimation showed lower performance (ICC, 0.727; mean DSC, 0.76±0.12; median APE 14.5%). Mean inference time was 4 s per examination. Correction of the 25 lowest-performing TLV outputs required 3.1 min per case. Conclusion: Automated fetal MRI lung segmentation enables rapid TLV estimation across diverse pulmonary hypoplasia phenotypes and may improve clinical workflow efficiency while preserving expert oversight for challenging cases.

Background/Objectives: Radiation therapy (RT) is part of standard‐of‐care therapy in high‐risk neuroblastoma (HR‐NBL) but can significantly affect nearby tissue. The objectives of this study were (1) to characterize changes in pancreas volume after RT and (2) to characterize changes in body composition, including body weight percentile, subcutaneous fat area, and total psoas muscle area (tPMA), after RT in children with abdominal HR‐NBL. Design/Methods: This retrospective study included patients aged 0–17 years old with HR‐NBL cared for at a quaternary pediatric hospital (January 2012–December 2022). Pre‐ to post‐RT differences in outcomes were compared using paired t ‐test, then modeled using linear regression to adjust for covariates and additional variables. Results: Fifty patients (median age 3 [IQR 2], 56% male) met the inclusion criteria. Median days from end of radiation to first post‐radiation imaging was 15.5 days (IQR 14). Pre‐ to post‐RT, mean pancreatic volume decreased significantly (from 16,731 mm ³ [SD 7659] to 14,288 mm ³ [SD 7418], p < 0.001) and patient body weight percentile declined significantly (48.4–44.9, p = 0.016), although neither was linearly associated with radiation dose. There were non‐significant decreases in subcutaneous fat area standardized to height (38.6–36.9 cm ² /m ² , p = 0.41) and in tPMA z ‐scores (−1.56 to −2.09, p = 0.26). Conclusions: Patients undergoing RT for abdominal HR‐NBL experienced significant pancreatic volume loss. This may place them at risk for pancreatic insufficiency. Patients also had significant decreases in weight percentile with corresponding non‐significant decreases in subcutaneous fat and psoas muscle area. Further research is needed to characterize the association of these findings with pancreatic function and nutritional status.

Introduction: Sarcopenia, the loss of muscle mass and function, is a measure of frailty that is correlated with perioperative morbidity in patients with chronic illness and transplant. We sought to understand the prevalence of sarcopenia in our pediatric population undergoing total pancreatectomy with islet cell autotransplantation (TPIAT) and how sarcopenia impacts short-term postsurgical outcomes. Methods: This was a single-center, IRB-approved study of children (1–18.99 y) who underwent TPIAT at our institution. Patients were identified from a prospective registry where clinical data (pain assessments, 30-d postoperative outcomes) were available near the time of surgery. Sarcopenia was assessed through retrospective analysis of cross-sectional imaging using segmentations of the psoas muscle and total abdominal skeletal muscle. Z-scores (<−1.6) calculated using established normal values defined sarcopenia. Change in muscle mass over time was also determined. Results: In our cohort, 32.4% (22/68) of patients were sarcopenic at the time of TPIAT by psoas muscle but only 7.4% (6/81) were sarcopenic by total abdominal skeletal muscle. There was no significant difference in primary outcomes (30-d readmissions, reoperations, and complications; and the number of days spent inpatient after surgery) based on sarcopenia or decreasing muscle mass. There was a statistically significant association between psoas-determined sarcopenia and higher frequency of pain before TPIAT ( P =0.016). Conclusions: Sarcopenia was 32.4% prevalent by psoas muscle measurement but was less prevalent by total abdominal skeletal muscle measurement in children undergoing TPIAT. There was no relationship between the presence of sarcopenia or decreasing muscle mass and 30-day postoperative outcomes but there was a positive association between higher frequency of preoperative pain and sarcopenia.

A regression model that incorporated sex, height, and weight accurately predicted normative liver and spleen volumes in a pediatric population.

Background Some children with inflammatory bowel disease (IBD) have hepatobiliary and/or pancreatic manifestations of the disease. The frequency of related imaging abnormalities at IBD diagnosis is unknown. Objective To identify and quantify hepatobiliary and pancreatic findings on baseline magnetic resonance enterography (MRE) examinations in children with newly diagnosed IBD. Materials and methods Children n=86) with Crohn disease; 22% (n=24) with ulcerative colitis; 1% (n=1) with IBD-unspecified. The mean age at diagnosis was 13.7 years (range, 5.3-17.9 years); 61% (n=43) were male. Subjective intrahepatic biliary dilation by reviewer consensus was present in 11% (n=12). Extrahepatic biliary dilation was present in 6% (n=7). Pancreatic duct dilation was present in 2% (n=2). Low pancreas:spleen T1 SIR (n=16). Pancreatic volume was 95th percentile in 13% (n=14). Conclusions Imaging findings potentially associated with autoimmune-related hepatobiliary or pancreatic disease are present in 2-14% on baseline MR enterography examinations in children with newly diagnosed inflammatory bowel disease. While infrequent, radiologists should be attentive to these findings of multisystem disease.

Background: Increased positron emission tomography (PET) scanner z-axis coverage provides an opportunity in pediatrics to reduce dose, anesthesia, or repeat scans due to motion. Objective: Recently, our digital PET scanner was upgraded from a 25-cm to a 30-cm z-axis coverage. We compare the two systems through National Electrical Manufacturing Association (NEMA) testing and evaluation of paired images from patients scanned on both systems. Materials and methods: NEMA testing and a retrospective review of pediatric patients who underwent clinically indicated 18F-fluorodeoxyglucose (FDG) PET computed tomography (PET/CT) on both systems with unchanged acquisition parameters were performed. Image quality was assessed with liver signal to noise ratio (SNR-liver) and contrast to noise ratio (CNR) in the thigh muscle and liver with results compared with an unpaired t-test. Three readers independently reviewed paired (25 cm and 30 cm) images from the same patient, blinded to scanner configuration. Results: Expansion to 30 cm increased system sensitivity to 29.8% (23.4 cps/kBq to 30.4 cps/kBq). Seventeen patients (6 male/11 female, median age 12.5 (IQR 8.3–15.0) years, median weight 53.7 (IQR 34.2–68.7) kg) were included. SNR-liver and CNR increased by 35.1% (IQR 19.0–48.4%) and 43.1% (IQR 6.2–50.2%) (P-value z-axis coverage from 25 to 30 cm on a current-generation digital PET scanner significantly improved PET system performance and patient image quality, and reduced scan duration.