TY - JOUR
T1 - Using transfer learning method to develop an artificial intelligence assisted triaging for endotracheal tube position on chest X-ray
AU - Yuan, Kuo Ching
AU - Tsai, Lung Wen
AU - Lai, Kevin S.
AU - Teng, Sing Teck
AU - Lo, Yu Sheng
AU - Peng, Syu Jyun
N1 - Funding Information:
Funding: This study was supported by grants from the Taipei Medical University/Taipei Medical University Hospital research grants (107TMU-TMUH-19), supported by the Ministry of Science and Technology, Taiwan, under the project MOST 110-2221-E-038-008, and partly supported by the Higher Education Sprout Project by the Ministry of Education, Taiwan (DP2-110-21121-01-A-12).
Publisher Copyright:
© 2021 by the authors. Licensee MDPI, Basel, Switzerland.
PY - 2021/10
Y1 - 2021/10
N2 - Endotracheal tubes (ETTs) provide a vital connection between the ventilator and patient; how-ever, improper placement can hinder ventilation efficiency or injure the patient. Chest X-ray (CXR) is the most common approach to confirming ETT placement; however, technicians require considerable exper-tise in the interpretation of CXRs, and formal reports are often delayed. In this study, we developed an artificial intelligence-based triage system to enable the automated assessment of ETT placement in CXRs. Three intensivists performed a review of 4293 CXRs obtained from 2568 ICU patients. The CXRs were labeled “CORRECT” or “INCORRECT” in accordance with ETT placement. A region of interest (ROI) was also cropped out, including the bilateral head of the clavicle, the carina, and the tip of the ETT. Transfer learning was used to train four pre-trained models (VGG16, INCEPTION_V3, RESNET, and DENSE-NET169) and two models developed in the current study (VGG16_Tensor Projection Layer and CNN_Tensor Projection Layer) with the aim of differentiating the placement of ETTs. Only VGG16 based on ROI images presented acceptable performance (AUROC = 92%, F1 score = 0.87). The results obtained in this study demonstrate the feasibility of using the transfer learning method in the development of AI models by which to assess the placement of ETTs in CXRs.
AB - Endotracheal tubes (ETTs) provide a vital connection between the ventilator and patient; how-ever, improper placement can hinder ventilation efficiency or injure the patient. Chest X-ray (CXR) is the most common approach to confirming ETT placement; however, technicians require considerable exper-tise in the interpretation of CXRs, and formal reports are often delayed. In this study, we developed an artificial intelligence-based triage system to enable the automated assessment of ETT placement in CXRs. Three intensivists performed a review of 4293 CXRs obtained from 2568 ICU patients. The CXRs were labeled “CORRECT” or “INCORRECT” in accordance with ETT placement. A region of interest (ROI) was also cropped out, including the bilateral head of the clavicle, the carina, and the tip of the ETT. Transfer learning was used to train four pre-trained models (VGG16, INCEPTION_V3, RESNET, and DENSE-NET169) and two models developed in the current study (VGG16_Tensor Projection Layer and CNN_Tensor Projection Layer) with the aim of differentiating the placement of ETTs. Only VGG16 based on ROI images presented acceptable performance (AUROC = 92%, F1 score = 0.87). The results obtained in this study demonstrate the feasibility of using the transfer learning method in the development of AI models by which to assess the placement of ETTs in CXRs.
KW - Artificial intelligence
KW - Chest X-ray
KW - Endotracheal tube
KW - Transfer learning
UR - http://www.scopus.com/inward/record.url?scp=85117166589&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85117166589&partnerID=8YFLogxK
U2 - 10.3390/diagnostics11101844
DO - 10.3390/diagnostics11101844
M3 - Article
AN - SCOPUS:85117166589
SN - 2075-4418
VL - 11
JO - Diagnostics
JF - Diagnostics
IS - 10
M1 - 1844
ER -