Overview

Schema Version

https://atlas.rsna.org/schemas/2025-11/dataset.json

Name

Nasogastric Tube Positioning on Chest Radiographs (NHS Greater Glasgow and Clyde)

Link

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10077078/

Indexing

Keywords: nasogastric tube, chest radiograph, malposition, bronchial insertion, feed/do not feed, decision support, adult, retrospective

Content: CH, ER, IN

RadLex: RID35976, RID5566, RID4753, RID46070, RID12734, RID5649, RID34919, RID10534

Author(s)

Ignat Drozdov

Rachael Dixon

Benjamin Szubert

Jessica Dunn

Darren Green

Nicola Hall

Arman Shirandami

Sofia Rosas

Ryan Grech

Srikanth Puttagunta

Mark Hall

David J. Lowe

Organization(s)

Bering Limited

NHS Greater Glasgow and Clyde

Queen Elizabeth University Hospital, Glasgow

Department of Radiology, NHS Greater Glasgow and Clyde

Institute of Health and Wellbeing, University of Glasgow

Industrial Centre for AI Research in Digital Diagnostics (iCAIRD)

Canon Medical Research Europe Limited

License

Text: CC BY 4.0

URL: https://creativecommons.org/licenses/by/4.0/

Contact

Ignat Drozdov (corresponding author)

Funding

Supported by Bering Limited and the Industrial Centre for AI Research in Digital Diagnostics (iCAIRD), funded by the UKRI Industrial Strategy Challenge Fund (project no. 104690).

Ethical review

Delegated research ethics approval granted by the Local Privacy and Advisory Committee at NHS Greater Glasgow and Clyde (reference: 104690/WP6/S1). In Scotland, consent not required for routinely collected data used via an approved Safe Haven.

Comments

Retrospective study of adult frontal chest radiographs used to pretrain and fine-tune a deep learning model for nasogastric tube (NGT) malposition detection, and to evaluate AI impact on junior physicians’ feed/do-not-feed decisions.

Date

Published: 2023-02-01

References

[1] Drozdov I, Dixon R, Szubert B, et al.. "An Artificial Neural Network for Nasogastric Tube Position Decision Support". Radiology: Artificial Intelligence. 2023-01-01. doi:10.1148/ryai.220165. PMID: 37035435. PMCID: PMC10077078.

Dataset

Motivation

Develop and validate a deep learning model to detect NGT malposition on chest radiographs and assess its impact as a decision support tool for junior physicians.

Sampling

Imaging studies requesting NGT position review identified via report keywords; additional manual review for bronchial placements; adults only (≥16 years).

Partitioning scheme

Stratified randomization into training (90%), validation (5%), and testing (5%) with preserved label frequencies, sex, and view positions; separate clinical evaluation set.

Missing information

Patient-level counts and exact per-split patient numbers not reported; detailed demographic breakdown beyond age/sex summary not provided.

Relationships between instances

Multiple radiographs originate from 14 acute sites and various departments; patient-level linkage used to prevent leakage across splits.

Noise

Heterogeneity in acquisition across 11 radiography systems (including portable), multiple departments, and wide resolution range; potential class imbalance.

External data

Neural networks initialized with ImageNet weights; radiograph pretraining and fine-tuning used data from NHS Greater Glasgow and Clyde.

Confidentiality

Data de-identified within the NHS Greater Glasgow and Clyde Safe Haven (Canon SHAIP platform). Identifiable patient data removed from DICOM files and linked radiology reports.

Re-identification

Processing performed within a trusted research environment (Safe Haven). Patient identifiers removed; splits ensured no patient identifier overlap.

Sensitive data

Clinical imaging data from NHS patients; adults only; de-identified for research use within Safe Haven.