Anthony Yang

Purpose: To develop a deep learning model for multi-anatomy segmentation of diverse anatomic structures on MRI. Materials and Methods: In this retrospective study, 44 structures were annotated using a model-assisted workflow with manual human finalization in 2 curated datasets: an internal dataset of 1518 MRI sequences (843 patients) from various clinical sites within a health system, and an external dataset of 397 MRI sequences (263 patients) from an independent imaging center for benchmarking. The internal dataset was used to train an nnU-Net model (MRAnnotator), while the external dataset evaluated MRAnnotator’s generalizability across significant image acquisition distribution shifts. MRAnnotator was further benchmarked against an nnU-Net model trained on the AMOS dataset and 2 current multi-anatomy MRI segmentation models, TotalSegmentator MRI (TSM) and MRSegmentator (MRS). Performance throughout was quantified using the Dice score. Results: MRAnnotator achieved an overall average Dice score of 0.878 (95% CI: 0.873, 0.884) on the internal dataset test set and 0.875 (95% CI: 0.869, 0.880) on the external dataset benchmark, demonstrating strong generalization (P = .899). On the AMOS test set, MRAnnotator achieved comparable performance for relevant classes (0.889 [0.866, 0.909]) to an AMOS-trained nnU-Net (0.895 [0.871, 0.915]) (P = .361) and outperformed TSM (0.822 [0.800, 0.842], P < .001) and MRS (0.867 [0.844, 0.887], P < .001). TSM and MRS were also evaluated on the relevant classes from the internal and external datasets and were unable to achieve comparable performance to MRAnnotator. Conclusion: MRAnnotator achieves robust and generalizable MRI segmentation across 44 anatomic structures. Future direction will incorporate additional anatomic structures into the datasets and model. Model weights are publicly available on GitHub. The external test set with annotations is available upon request.

“Just Accepted” papers have undergone full peer review and have been accepted for publication in Radiology: Artificial Intelligence. This article will undergo copyediting, layout, and proof review ...

Haley M. Zylberberg, Sheila D. Rustgi, Anthony Yang, Anne Aronson, Elizabeth Kessel, Sunil Amin, and Aimee L. Lucas. Gut and Liver 2021;15:782-90. https://doi.org/10.5009/gnl20297

Objectives: The ideal timing and technique of tracheostomy vary among patients and may impact outcomes. We aim to examine the association between tracheostomy timing, placement technique, and patient demographics on survival.Study Design: Retrospective cohort study.Methods: A retrospective review was performed for all patients who underwent tracheostomy in 2016 and 2017 at one urban academic tertiary‐care hospital. Kaplan–Meier curves were created based on combinations of tracheostomy timing and technique (early percutaneous, early non‐percutaneous, late percutaneous, and late non‐percutaneous). Cox proportional hazard models were used to determine multivariable effects of timing, technique, and other demographic factors. Primary outcome measures were tracheostomy‐related mortality and overall survival. Secondary outcomes were in‐hospital, 30‐day, and 90‐day mortality.Results: Our study included 523 patients. There were six tracheostomy‐related deaths, with hemorrhage and tracheoesophageal fistula being the most common causes. Tracheostomy timing and technique combinations were not associated with differences in all‐cause mortality or survival following discharge. Cox proportional hazard models showed that Charlson Comorbidity Index (CCI) and unknown partner status were associated with a decrease in survival (P < .01 and P = .05, respectively). Additionally, patient age, gender, race, CCI, and body mass index were not independently associated with changes in survival.Conclusion: Late and non‐percutaneous tracheostomies were associated with more tracheostomy‐related deaths, but timing and technique were not associated with differences in patient survival. Multiple regression analysis showed that increased patient comorbidities, measured via CCI, and unknown partner status were independently associated with decreased survival. Proceduralists should discuss timing, technique, and patient social factors together with the medical care team when constructing plans for postdischarge management.Level of Evidence: 4 Laryngoscope, 131:1468–1473, 2021

Objectives/Hypothesis: Little data exists regarding the relationship between socioeconomic and demographic factors and tracheostomy outcomes. The goal of this study was to determine associations between socioeconomic status (SES), demographic factors, and insurance status with hospital length of stay (LOS), intensive care unit (ICU) LOS, and mortality following tracheostomy.Study Design: Retrospective cohort study.Methods: A retrospective analysis of all patients who underwent tracheostomy at an urban tertiary‐care academic hospital from 2016 to 2017 was performed. Patients were aggregated into low‐, middle‐, and high‐income brackets. Other variables included age, sex, race, ethnicity, body mass index, and Charlson Comorbidity Index (CCI). Outcomes included hospital and ICU LOS, in‐hospital mortality, and 30‐day mortality following tracheostomy. Outcomes were compared using Kruskal‐Wallis tests for continuous variables and χ² or Fisher exact tests for categorical variables. The α level was set to .05.Results: In total, 523 patients were included in the study. Patients from high‐income areas were more likely to be male (P < .01), white (P < .01), and had lower body mass index (P = .04). On multiple regression analysis, Hispanic or Latino ethnicity was associated with an increased odds of 30‐day mortality (odds ratio [OR]: 4.43, P = .020). CCI was also associated with increased odds of 30‐day mortality (OR: 1.12, P = .039).Conclusions: Lower SES was not associated with increased morbidity or mortality after tracheostomy. Although Hispanic patients tended to have a lower CCI score, they had increased 30‐day mortality, suggesting there are factors specific to this population that may influence outcomes, and future targeted studies are warranted to study these relationships.Level of Evidence: 4 Laryngoscope, 131:1463–1467, 2021

Objectives/Hypothesis The purpose of this study was to compare timing of procedure, patient characteristics, outcomes, and charges for patients who underwent percutaneous versus surgical tracheostomy. Study Design Retrospective cohort study. Methods A retrospective analysis was performed for all patients who underwent tracheostomy in 2015 to 2016 in New York State. Patients were identified using International Classification of Diseases, 10th Revision, Clinical Modification codes and stratified to the type of tracheostomy performed. The primary outcome of interest was mortality at index stay. Secondary outcomes of interest included length of stay and total hospitalization charges. Results Of the 8,682 patients, 2,488 (28.7%) underwent percutaneous and 6,194 (71.3%) underwent surgical tracheostomy. At hospitals where both procedures were performed, percutaneous tracheostomy patients were older, had more comorbidities, and had lower income (P < .05). Timing of the tracheostomy relative to admission did not affect the type of tracheostomy performed. While controlling for patient characteristics and complications during the visit, percutaneous tracheostomy was associated with increased mortality (odds ratio [OR]: 1.17, 95% confidence interval [CI]: 1.03‐1.33, P = .0153) and increased hospital charges (OR: + 7.76%, 95% CI: 5.4‐10.11, P < .0001). Length of stay was not affected by procedure type. Conclusions Surgical tracheostomies are more commonly performed than percutaneous tracheostomies across New York State. Older, lower‐income, and sicker patients have a higher chance of receiving percutaneous tracheostomies. Percutaneous approaches were associated with statistically significant increased mortality and higher charges despite no difference in length of stay. Further studies are needed to determine if these differences in outcomes are clinically significant. Level of Evidence NA Laryngoscope, 2018

Robust and predictably performing synthetic circuits rely on the use of well-characterized regulatory parts across different genetic backgrounds and environmental contexts. Here we report the large-scale metagenomic mining of thousands of natural 5′ regulatory sequences from diverse bacteria, and their multiplexed gene expression characterization in industrially relevant microbes. We identified sequences with broad and host-specific expression properties that are robust in various growth conditions. We also observed substantial differences between species in terms of their capacity to utilize exogenous regulatory sequences. Finally, we demonstrate programmable species-selective gene expression that produces distinct and diverse output patterns in different microbes. Together, these findings provide a rich resource of characterized natural regulatory sequences and a framework that can be used to engineer synthetic gene circuits with unique and tunable cross-species functionality and properties, and also suggest the prospect of ultimately engineering complex behaviors at the community level.