Automatic deep learning-based pleural effusion segmentation in lung ultrasound images

Damjan Vukovic; Andrew Wang; Maria Antico; Marian Steffens; Igor Ruvinov; Ruud J.G. van Sloun; David Canty; Alistair Royse; Colin Royse; Kavi Haji; Jason Dowling; Girija Chetty; Davide Fontanarosa

doi:10.1186/s12911-023-02362-6

Automatic deep learning-based pleural effusion segmentation in lung ultrasound images

Damjan Vukovic, Andrew Wang, Maria Antico, Marian Steffens, Igor Ruvinov, Ruud J.G. van Sloun, David Canty, Alistair Royse, Colin Royse, Kavi Haji, Jason Dowling, Girija Chetty, Davide Fontanarosa

Information Systems

Research output: Contribution to journal › Article › peer-review

Abstract

Background
Point-of-care lung ultrasound (LUS) allows real-time patient scanning to help diagnose pleural effusion (PE) and plan further investigation and treatment. LUS typically requires training and experience from the clinician to accurately interpret the images. To address this limitation, we previously demonstrated a deep-learning model capable of detecting the presence of PE on LUS at an accuracy greater than 90%, when compared to an experienced LUS operator.

Methods
This follow-up study aimed to develop a deep-learning model to provide segmentations for PE in LUS. Three thousand and forty-one LUS images from twenty-four patients diagnosed with PE were selected for this study. Two LUS experts provided the ground truth for training by reviewing and segmenting the images. The algorithm was then trained using ten-fold cross-validation. Once training was completed, the algorithm segmented a separate subset of patients.

Results
Comparing the segmentations, we demonstrated an average Dice Similarity Coefficient (DSC) of 0.70 between the algorithm and experts. In contrast, an average DSC of 0.61 was observed between the experts.

Original language	English
Article number	274
Pages (from-to)	1-14
Number of pages	14
Journal	BMC Medical Informatics and Decision Making
Volume	23
Issue number	2023
DOIs	https://doi.org/10.1186/s12911-023-02362-6
Publication status	Published - 29 Nov 2023

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

Access to Document

10.1186/s12911-023-02362-6Licence: CC BY

Cite this

Vukovic, D., Wang, A., Antico, M., Steffens, M., Ruvinov, I., van Sloun, R. J. G., Canty, D., Royse, A., Royse, C., Haji, K., Dowling, J., Chetty, G., & Fontanarosa, D. (2023). Automatic deep learning-based pleural effusion segmentation in lung ultrasound images. BMC Medical Informatics and Decision Making, 23(2023), 1-14. [274]. https://doi.org/10.1186/s12911-023-02362-6

@article{a462d0a736e74b70bd7ddeb2b24b4dbc,

title = "Automatic deep learning-based pleural effusion segmentation in lung ultrasound images",

abstract = "BackgroundPoint-of-care lung ultrasound (LUS) allows real-time patient scanning to help diagnose pleural effusion (PE) and plan further investigation and treatment. LUS typically requires training and experience from the clinician to accurately interpret the images. To address this limitation, we previously demonstrated a deep-learning model capable of detecting the presence of PE on LUS at an accuracy greater than 90%, when compared to an experienced LUS operator.MethodsThis follow-up study aimed to develop a deep-learning model to provide segmentations for PE in LUS. Three thousand and forty-one LUS images from twenty-four patients diagnosed with PE were selected for this study. Two LUS experts provided the ground truth for training by reviewing and segmenting the images. The algorithm was then trained using ten-fold cross-validation. Once training was completed, the algorithm segmented a separate subset of patients.ResultsComparing the segmentations, we demonstrated an average Dice Similarity Coefficient (DSC) of 0.70 between the algorithm and experts. In contrast, an average DSC of 0.61 was observed between the experts.",

keywords = "Deep Learning, Lung Ultrasound, Pleural Effusion / diagnostic imaging, Point-of-Care Ultrasound, Algorithms",

author = "Damjan Vukovic and Andrew Wang and Maria Antico and Marian Steffens and Igor Ruvinov and {van Sloun}, {Ruud J.G.} and David Canty and Alistair Royse and Colin Royse and Kavi Haji and Jason Dowling and Girija Chetty and Davide Fontanarosa",

note = "Publisher Copyright: {\textcopyright} 2023, The Author(s).",

year = "2023",

month = nov,

day = "29",

doi = "10.1186/s12911-023-02362-6",

language = "English",

volume = "23",

pages = "1--14",

journal = "BMC Medical Informatics and Decision Making",

issn = "1472-6947",

publisher = "BioMed Central",

number = "2023",

}

TY - JOUR

T1 - Automatic deep learning-based pleural effusion segmentation in lung ultrasound images

AU - Vukovic, Damjan

AU - Wang, Andrew

AU - Antico, Maria

AU - Steffens, Marian

AU - Ruvinov, Igor

AU - van Sloun, Ruud J.G.

AU - Canty, David

AU - Royse, Alistair

AU - Royse, Colin

AU - Haji, Kavi

AU - Dowling, Jason

AU - Chetty, Girija

AU - Fontanarosa, Davide

PY - 2023/11/29

Y1 - 2023/11/29

N2 - BackgroundPoint-of-care lung ultrasound (LUS) allows real-time patient scanning to help diagnose pleural effusion (PE) and plan further investigation and treatment. LUS typically requires training and experience from the clinician to accurately interpret the images. To address this limitation, we previously demonstrated a deep-learning model capable of detecting the presence of PE on LUS at an accuracy greater than 90%, when compared to an experienced LUS operator.MethodsThis follow-up study aimed to develop a deep-learning model to provide segmentations for PE in LUS. Three thousand and forty-one LUS images from twenty-four patients diagnosed with PE were selected for this study. Two LUS experts provided the ground truth for training by reviewing and segmenting the images. The algorithm was then trained using ten-fold cross-validation. Once training was completed, the algorithm segmented a separate subset of patients.ResultsComparing the segmentations, we demonstrated an average Dice Similarity Coefficient (DSC) of 0.70 between the algorithm and experts. In contrast, an average DSC of 0.61 was observed between the experts.

AB - BackgroundPoint-of-care lung ultrasound (LUS) allows real-time patient scanning to help diagnose pleural effusion (PE) and plan further investigation and treatment. LUS typically requires training and experience from the clinician to accurately interpret the images. To address this limitation, we previously demonstrated a deep-learning model capable of detecting the presence of PE on LUS at an accuracy greater than 90%, when compared to an experienced LUS operator.MethodsThis follow-up study aimed to develop a deep-learning model to provide segmentations for PE in LUS. Three thousand and forty-one LUS images from twenty-four patients diagnosed with PE were selected for this study. Two LUS experts provided the ground truth for training by reviewing and segmenting the images. The algorithm was then trained using ten-fold cross-validation. Once training was completed, the algorithm segmented a separate subset of patients.ResultsComparing the segmentations, we demonstrated an average Dice Similarity Coefficient (DSC) of 0.70 between the algorithm and experts. In contrast, an average DSC of 0.61 was observed between the experts.

KW - Deep Learning

KW - Lung Ultrasound

KW - Pleural Effusion / diagnostic imaging

KW - Point-of-Care Ultrasound

KW - Algorithms

UR - https://www.scopus.com/record/display.uri?eid=2-s2.0-85178181794&origin=resultslist

U2 - 10.1186/s12911-023-02362-6

DO - 10.1186/s12911-023-02362-6

M3 - Article

C2 - 38031040

AN - SCOPUS:85178181794

SN - 1472-6947

VL - 23

SP - 1

EP - 14

JO - BMC Medical Informatics and Decision Making

JF - BMC Medical Informatics and Decision Making

IS - 2023

M1 - 274

ER -

Automatic deep learning-based pleural effusion segmentation in lung ultrasound images

Abstract

UN SDGs

Access to Document

Other files and links

Fingerprint

Cite this