Deep-Learning Models for Abdominal CT Organ Segmentation in Children: Development and Validation in Internal and Heterogeneous Public Datasets.

Background: Deep-learning abdominal organ segmentation algorithms have shown excellent results in adults; validation in children is sparse. Objective: To develop and validate deep-learning models for liver, spleen, and pancreas segmentation on pediatric CT examinations. Methods: This retrospective study developed and validated deep-learning models for liver, spleen, and pancreas segmentation using 1731 CT examinations (1504 training, 221 testing), derived from three internal institutional pediatric (age ≤18) datasets (n=483) and three public datasets comprising pediatric and adult examinations with various pathologies (n=1248). Three deep-learning model architectures (SegResNet, DynUNet, and SwinUNETR) from the Medical Open Network for AI (MONAI) framework underwent training using native training (NT), relying solely on institutional datasets, and transfer learning (TL), incorporating pre-training on public datasets. For comparison, TotalSegmentator (TS), a publicly available segmentation model, was applied to test data without further training. Segmentation performance was evaluated using mean Dice similarity coefficient (DSC), with manual segmentations as reference. Results: For internal pediatric data, DSC for normal liver was 0.953 (TS), 0.964-0.965 (NT models), and 0.965-0.966 (TL models); normal spleen, 0.914 (TS), 0.942-0.945 (NT models), and 0.937-0.945 (TL models); normal pancreas, 0.733 (TS), 0.774-0.785 (NT models), and 0.775-0.786 (TL models); pancreas with pancreatitis, 0.703 (TS), 0.590-0.640 (NT models), and 0.667-0.711 (TL models). For public pediatric data, DSC for liver was 0.952 (TS), 0.876-0.908 (NT models), and 0.941-0.946 (TL models); spleen, 0.905 (TS), 0.771-0.827 (NT models), and 0.897-0.926 (TL models); pancreas, 0.700 (TS), 0.577-0.648 (NT models), and 0.693-0.736 (TL models). For public primarily adult data, DSC for liver was 0.991 (TS), 0.633-0.750 (NT models), and 0.926-0.952 (TL models); spleen, 0.983 (TS), 0.569-0.604 (NT models), and 0.923-0.947 (TL models); pancreas, 0.909 (TS), 0.148-0.241 (NT models), and 0.699-0.775 (TL models). DynUNet-TL was selected as the best-performing NT or TL model and was made available as an opensource MONAI bundle (https://github.com/cchmc-dll/pediatric_abdominal_segmentation_bundle.git). Conclusion: TL models trained on heterogeneous public datasets and fine-tuned using institutional pediatric data outperformed internal NT models and TotalSegmentator across internal and external pediatric test data. Segmentation performance was better in liver and spleen than in pancreas. Clinical Impact: The selected model may be used for various volumetry applications in pediatric imaging.

Endemic mycoses in children in North America: a review of radiologic findings.

Clinically significant endemic mycoses (fungal infections) in the United States (U.S.) include Blastomyces dermatitidis, Histoplasma capsulatum, and Coccidioides immitis/posadasii. While the majority of infections go clinically unnoticed, symptomatic disease can occur in immunocompromised or hospitalized patients, and occasionally in immune-competent individuals. Clinical manifestations vary widely and their diagnosis may require fungal culture, making the rapid diagnosis a challenge. Imaging can be helpful in making a clinical diagnosis prior to laboratory confirmation, as well as assist in characterizing disease extent and severity. In this review, we discuss the three major endemic fungal infections that occur in the U.S., including mycology, epidemiology, clinical presentations, and typical imaging features with an emphasis on the pediatric population.

Prospective characterization of incident hepatic steatosis in pediatric and adolescent patients after total pancreatectomy with islet autotransplantation.

BACKGROUND: Hepatic steatosis has been described as a common finding in adults following total pancreatectomy with islet autotransplantation (TPIAT) but it is unknown if this occurs in children and adolescents. OBJECTIVES: To define the frequency of post-TPIAT hepatic steatosis in a sample of children and adolescents and to identify clinical predictors of incident steatosis post-TPIAT. METHODS: In this prospective study, consecutive participants at least 1-month post-TPIAT underwent a liver MRI with proton density fat fraction (PDFF) and blood draw at our pediatric academic medical center between April 2021 and January 2022. Comparison clinical pre-TPIAT liver MRI or ultrasound and insulin use and graft function data were extracted from the medical record. T-tests were used for the comparison of means across continuous variables between participants with and without post-TPIAT steatosis. RESULTS: A total of 20 participants (mean: 13 ± 4 years; 12 female) were evaluated. Mean liver PDFF at research MRI was 7.4 ± 6.2% (range: 2-25%). Seven participants (35%) had categorical hepatic steatosis (PDFF>5%) post-TPIAT, five of whom had pre-TPIAT steatosis, reflecting a 13% (2/15; 95% CI: 2-40%) incidence of post-TPIAT steatosis. Participant characteristics were not significantly different between subgroups with and without post-TPIAT steatosis. Mean PDFF at research MRI was not different between graft function subgroups (7.5% optimal/good vs. 7.3% marginal/failure; p = .96). CONCLUSION: Our study shows a moderate prevalence but low incidence of hepatic steatosis in a small sample of children and adolescents post-TPIAT. This study raises questions about a causal relationship between TPIAT and hepatic steatosis.

Pancreas volumes and predictive factors in healthy children.

BACKGROUND: Pancreas volume might be a quantitative metric of pancreas health and function in children. OBJECTIVE: To establish normative pancreas volumes and determine factors associated with pancreas volume. MATERIALS AND METHODS: We conducted a retrospective study of 140 healthy children (balanced from 0 to 18 years, stratified by age and gender) who underwent contrast-enhanced CT of the abdomen. Pancreas volume was manually segmented by a single reviewer using 3D Slicer and corrected by a pediatric radiologist. We used Bland-Altman difference analysis to quantify differences in initial and refined segmented pancreas volume, and the Mann-Whitney U test to compare continuous variables. We used Pearson correlation for univariate associations. To determine predictors, we used multivariable regression. Finally, we generated quantile regression equations to determine pancreas volume based on age or body surface area (BSA). RESULTS: Pancreas volume for the study sample ranged from 2 mL to 99 mL. Age (r=0.90, P<0.0001), body mass index (BMI) (r=0.66, P<0.0001), BSA (r=0.94, P<0.0001), height (r=0.91, P<0.0001) and weight (r=0.90, P<0.0001) were all positively correlated with pancreas volume on univariate analysis. On multivariable analysis, BSA (+36 mL/m2, P<0.0001) and female gender (-2.8 mL, P=0.062) were significant independent predictors of pancreas volume. The mean difference between initial and refined segmentation was 0.80 mL (95% limits of agreement: -7.9 mL to 9.5 mL). CONCLUSION: We report pancreas volumes for healthy children. We found that age, BMI, BSA, height and weight were each significantly, positively correlated with pancreas volume in univariate analyses, while BSA and female gender were significant independent predictive factors on multivariable analysis.

Pancreatic Masses in Children and Young Adults: Multimodality Review with Pathologic Correlation.

Masses and masslike lesions of the pancreas are uncommon in the pediatric population. However, owing to overlapping clinical and imaging features, it can be challenging to differentiate the various causes of pediatric pancreatic masses at initial patient presentation. Clinical data such as patient age, signs and symptoms at presentation, laboratory test results, and potential underlying cancer predisposition syndrome can be helpful when formulating a differential diagnosis. US may be the first imaging study to depict a pancreatic mass in a child, as this examination is frequently performed in children with nonspecific abdominal signs and symptoms because of its wide availability and relatively low cost and the lack of a need for sedation or anesthesia. CT or MRI is typically required for more thorough characterization of the mass and surgical planning. Complete characterization of pancreatic masses includes assessment of vascular involvement, local invasion, and extrapancreatic spread of tumor. The authors provide an up-to-date comprehensive review of the clinical manifestations, histopathologic features, and imaging findings of primary and secondary tumors of the pancreas in children and young adults. Advances in imaging, current prognostic information, and treatment paradigms also are highlighted. Finally, nontumorous masslike lesions of the pediatric pancreas, including vascular malformations, cystic disorders (eg, von Hippel-Lindau syndrome, cystic fibrosis), intrapancreatic accessory spleen, and autoimmune pancreatitis, are discussed. Online supplemental material is available for this article. ©RSNA, 2021.

Pancreas ultrasound two-dimensional shear wave elastography in healthy children.

BACKGROUND: Pancreas shear wave speed might be a biomarker of pancreatic disease in children. OBJECTIVE: This study aimed to measure pancreas shear wave speed by two-dimensional (2-D) ultrasound shear wave elastography (SWE) in a balanced cohort of presumed healthy children. MATERIALS AND METHODS: This was a prospective study of 120 children (<18 years of age) without a known history of pancreatic disease, who underwent ultrasound 2-D SWE of the pancreas. Five shear wave speed measurements in the pancreas body and/or tail were obtained for each participant using a Canon Aplio i800 system, i8CX1 transducer. The Mann-Whitney U test or Kruskal-Wallis test were used to compare continuous distributions. Spearman's correlation was used to assess univariate relationships between continuous variables. Multivariable regression with stepwise selection was used to evaluate independent predictors of pancreas shear wave speed. RESULTS: The median age for the study population was 5.0 years (range: 7 days to 17.8 years) and 61 (50.8%) of the participants were female. The median depth of shear wave speed measurement was 4.7 cm (interquartile range [IQR]: 4.2-5.3). The median pancreas shear wave speed was 1.31 m/s (IQR: 1.21-1.40). On multivariable analysis, female biological sex (P=0.051), the number of hours nil per os (P=0.097), the median depth of measurement (P=0.001) and the median liver shear wave speed (P=0.020) were positively associated with pancreas shear wave speed. CONCLUSION: We report pancreas shear wave speed in a large, balanced cohort of children without a known history of pancreatic disease, providing reference values for normal.

Chalcogen extrusion from heteroallenes and carbon monoxide by a three-coordinate rh(i) disilylamide.

We report the reactions of several heteroallenes (carbon disulfide, carbonyl sulfide, and phenyl isocyanate) and carbon monoxide with a three-coordinate, bis(phosphine)-supported Rh(I) disilylamide (1). Carbon disulfide reacts with 1 to afford a silyltrithiocarbonate complex similar to an intermediate previously invoked in the deoxygenation of CO2 by 1, and prolonged heating affords a structurally unusual µ-κ(2)(S,S'):κ(2)(S,S')-trithiocarbonate dimer. Carbonyl sulfide reacts with 1 to afford a structurally unique Rh(SCNCS) metallacycle derived from two insertions of OCS and N-to-O silyl-group migrations. Phenyl isocyanate reacts with 1 to afford a dimeric bis(phenylcyanamido)-bridged complex resulting from multiple silyl-group migrations and nitrogen-for-oxygen metathesis, akin to reactivity previously observed with carbon dioxide. The ability of 1 to activate carbon-chalcogen multiple bonds via silyl-group migration is further supported by its reactivity with carbon monoxide, where a nitrogen-for-oxygen metathesis is also observed with expulsion of hexamethyldisiloxane. For all reported reactions, intermediates are observable under appropriate conditions, allowing the formulation of mechanisms where insertion of the unsaturated substrate is followed by one or more silyl-group migrations to afford the observed products. This rich variety of reactivity confirms the ability of metal silylamides to activate exceptionally strong carbon-element multiple bonds and suggests that silylamides may be useful intermediates in nitrogen-atom and nitrene-group-transfer schemes.