Deep learning for tubes and lines detection in critical illness: Generalizability and comparison with residents

Background: Artificial intelligence (AI) has been proven useful for the assessment of tubes and lines on chest radiographs of general patients. However, validation on intensive care unit (ICU) patients remains imperative. Methods: This retrospective case-control study evaluated the performance of de...

Full description

Saved in:
Bibliographic Details
Main Authors: Pootipong Wongveerasin, Trongtum Tongdee, Pairash Saiviroonporn
Format: Article
Language:English
Published: Elsevier 2024-12-01
Series:European Journal of Radiology Open
Subjects:
Online Access:http://www.sciencedirect.com/science/article/pii/S2352047724000480
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1846122146014691328
author Pootipong Wongveerasin
Trongtum Tongdee
Pairash Saiviroonporn
author_facet Pootipong Wongveerasin
Trongtum Tongdee
Pairash Saiviroonporn
author_sort Pootipong Wongveerasin
collection DOAJ
description Background: Artificial intelligence (AI) has been proven useful for the assessment of tubes and lines on chest radiographs of general patients. However, validation on intensive care unit (ICU) patients remains imperative. Methods: This retrospective case-control study evaluated the performance of deep learning (DL) models for tubes and lines classification on both an external public dataset and a local dataset comprising 303 films randomly sampled from the ICU database. The endotracheal tubes (ETTs), central venous catheters (CVCs), and nasogastric tubes (NGTs) were classified into “Normal,” “Abnormal,” or “Borderline” positions by DL models with and without rule-based modification. Their performance was evaluated using an experienced radiologist as the standard reference. Results: The algorithm showed decreased performance on the local ICU dataset, compared to that of the external dataset, decreasing from the Area Under the Curve of Receiver (AUC) of 0.967 (95 % CI 0.965–0.973) to the AUC of 0.70 (95 % CI 0.68–0.77). Significant improvement in the ETT classification task was observed after modifications were made to the model to allow the use of the spatial relationship between line tips and reference anatomy with the improvement of the AUC, increasing from 0.71 (95 % CI 0.70 – 0.75) to 0.86 (95 % CI 0.83 – 0.94) Conclusions: The externally trained model exhibited limited generalizability on the local ICU dataset. Therefore, evaluating the performance of externally trained AI before integrating it into critical care routine is crucial. Rule-based algorithm may be used in combination with DL to improve results.
format Article
id doaj-art-eadeca859e2a46668506a5afc21d4034
institution Kabale University
issn 2352-0477
language English
publishDate 2024-12-01
publisher Elsevier
record_format Article
series European Journal of Radiology Open
spelling doaj-art-eadeca859e2a46668506a5afc21d40342024-12-15T06:15:46ZengElsevierEuropean Journal of Radiology Open2352-04772024-12-0113100593Deep learning for tubes and lines detection in critical illness: Generalizability and comparison with residentsPootipong Wongveerasin0Trongtum Tongdee1Pairash Saiviroonporn2Corresponding author.; Department of Radiology, Faculty of Medicine Siriraj Hospital, Mahidol University, Bangkok, ThailandDepartment of Radiology, Faculty of Medicine Siriraj Hospital, Mahidol University, Bangkok, ThailandDepartment of Radiology, Faculty of Medicine Siriraj Hospital, Mahidol University, Bangkok, ThailandBackground: Artificial intelligence (AI) has been proven useful for the assessment of tubes and lines on chest radiographs of general patients. However, validation on intensive care unit (ICU) patients remains imperative. Methods: This retrospective case-control study evaluated the performance of deep learning (DL) models for tubes and lines classification on both an external public dataset and a local dataset comprising 303 films randomly sampled from the ICU database. The endotracheal tubes (ETTs), central venous catheters (CVCs), and nasogastric tubes (NGTs) were classified into “Normal,” “Abnormal,” or “Borderline” positions by DL models with and without rule-based modification. Their performance was evaluated using an experienced radiologist as the standard reference. Results: The algorithm showed decreased performance on the local ICU dataset, compared to that of the external dataset, decreasing from the Area Under the Curve of Receiver (AUC) of 0.967 (95 % CI 0.965–0.973) to the AUC of 0.70 (95 % CI 0.68–0.77). Significant improvement in the ETT classification task was observed after modifications were made to the model to allow the use of the spatial relationship between line tips and reference anatomy with the improvement of the AUC, increasing from 0.71 (95 % CI 0.70 – 0.75) to 0.86 (95 % CI 0.83 – 0.94) Conclusions: The externally trained model exhibited limited generalizability on the local ICU dataset. Therefore, evaluating the performance of externally trained AI before integrating it into critical care routine is crucial. Rule-based algorithm may be used in combination with DL to improve results.http://www.sciencedirect.com/science/article/pii/S2352047724000480Tubes and lineschest radiographartificial intelligencedeep learninggeneralizabilityresidents
spellingShingle Pootipong Wongveerasin
Trongtum Tongdee
Pairash Saiviroonporn
Deep learning for tubes and lines detection in critical illness: Generalizability and comparison with residents
European Journal of Radiology Open
Tubes and lines
chest radiograph
artificial intelligence
deep learning
generalizability
residents
title Deep learning for tubes and lines detection in critical illness: Generalizability and comparison with residents
title_full Deep learning for tubes and lines detection in critical illness: Generalizability and comparison with residents
title_fullStr Deep learning for tubes and lines detection in critical illness: Generalizability and comparison with residents
title_full_unstemmed Deep learning for tubes and lines detection in critical illness: Generalizability and comparison with residents
title_short Deep learning for tubes and lines detection in critical illness: Generalizability and comparison with residents
title_sort deep learning for tubes and lines detection in critical illness generalizability and comparison with residents
topic Tubes and lines
chest radiograph
artificial intelligence
deep learning
generalizability
residents
url http://www.sciencedirect.com/science/article/pii/S2352047724000480
work_keys_str_mv AT pootipongwongveerasin deeplearningfortubesandlinesdetectionincriticalillnessgeneralizabilityandcomparisonwithresidents
AT trongtumtongdee deeplearningfortubesandlinesdetectionincriticalillnessgeneralizabilityandcomparisonwithresidents
AT pairashsaiviroonporn deeplearningfortubesandlinesdetectionincriticalillnessgeneralizabilityandcomparisonwithresidents