ESFPNet: Efficient deep learning architecture for real-time lesion segmentation in autofluorescence bronchoscopic video

Qi Chang; Danish Ahmad; Jennifer Toth; Rebecca Bascom; William E. Higgins

doi:10.1117/12.2647897

ESFPNet: Efficient deep learning architecture for real-time lesion segmentation in autofluorescence bronchoscopic video

Qi Chang
, Danish Ahmad
, Jennifer Toth
, Rebecca Bascom
, William E. Higgins

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

51 Scopus citations

Abstract

Lung cancer tends to be detected at an advanced stage, resulting in a high patient mortality rate. Thus, recent research has focused on early disease detection. Autofluorescence bronchoscopy (AFB) is an effective noninvasive way of detecting early manifestations of lung cancer. Unfortunately, manual inspection of AFB video is extremely tedious and error-prone, while limited effort has been expended toward potentially more robust automatic AFB lesion analysis. We propose a real-time deep-learning architecture dubbed ESFPNet for accurate segmentation and robust detection of bronchial lesions in an AFB video stream. Our approach gives the best segmentation results (mDice = 0.756, mIoU=0.624) on our AFB dataset among recent architectures. Moreover, our model shows promising potential applicability to other domains, as evidenced by its state-of-the-art (SOTA) performance on the CVC-ClinicDB,ETIS-LaribPolypDB datasets, and superior performance on the Kvasir, CVC-ColonDB datasets.

Original language	English (US)
Title of host publication	Medical Imaging 2023
Subtitle of host publication	Biomedical Applications in Molecular, Structural, and Functional Imaging
Editors	Barjor S. Gimi, Andrzej Krol
Publisher	SPIE
ISBN (Electronic)	9781510660410
DOIs	https://doi.org/10.1117/12.2647897
State	Published - 2023
Event	Medical Imaging 2023: Biomedical Applications in Molecular, Structural, and Functional Imaging - San Diego, United States Duration: Feb 19 2023 → Feb 22 2023

Publication series

Name	Progress in Biomedical Optics and Imaging - Proceedings of SPIE
Volume	12468
ISSN (Print)	1605-7422

Conference

Conference	Medical Imaging 2023: Biomedical Applications in Molecular, Structural, and Functional Imaging
Country/Territory	United States
City	San Diego
Period	2/19/23 → 2/22/23

UN SDGs

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

SDG 3 Good Health and Well-being

All Science Journal Classification (ASJC) codes

Electronic, Optical and Magnetic Materials
Atomic and Molecular Physics, and Optics
Biomaterials
Radiology Nuclear Medicine and imaging

Access to Document

10.1117/12.2647897

Cite this

Chang, Q., Ahmad, D., Toth, J., Bascom, R., & Higgins, W. E. (2023). ESFPNet: Efficient deep learning architecture for real-time lesion segmentation in autofluorescence bronchoscopic video. In B. S. Gimi, & A. Krol (Eds.), Medical Imaging 2023: Biomedical Applications in Molecular, Structural, and Functional Imaging Article 1246803 (Progress in Biomedical Optics and Imaging - Proceedings of SPIE; Vol. 12468). SPIE. https://doi.org/10.1117/12.2647897

Chang, Qi ; Ahmad, Danish ; Toth, Jennifer et al. / ESFPNet : Efficient deep learning architecture for real-time lesion segmentation in autofluorescence bronchoscopic video. Medical Imaging 2023: Biomedical Applications in Molecular, Structural, and Functional Imaging. editor / Barjor S. Gimi ; Andrzej Krol. SPIE, 2023. (Progress in Biomedical Optics and Imaging - Proceedings of SPIE).

@inproceedings{ee6fe71cdca84e659c8a0b5afb539ee2,

title = "ESFPNet: Efficient deep learning architecture for real-time lesion segmentation in autofluorescence bronchoscopic video",

abstract = "Lung cancer tends to be detected at an advanced stage, resulting in a high patient mortality rate. Thus, recent research has focused on early disease detection. Autofluorescence bronchoscopy (AFB) is an effective noninvasive way of detecting early manifestations of lung cancer. Unfortunately, manual inspection of AFB video is extremely tedious and error-prone, while limited effort has been expended toward potentially more robust automatic AFB lesion analysis. We propose a real-time deep-learning architecture dubbed ESFPNet for accurate segmentation and robust detection of bronchial lesions in an AFB video stream. Our approach gives the best segmentation results (mDice = 0.756, mIoU=0.624) on our AFB dataset among recent architectures. Moreover, our model shows promising potential applicability to other domains, as evidenced by its state-of-the-art (SOTA) performance on the CVC-ClinicDB,ETIS-LaribPolypDB datasets, and superior performance on the Kvasir, CVC-ColonDB datasets.",

author = "Qi Chang and Danish Ahmad and Jennifer Toth and Rebecca Bascom and Higgins, \{William E.\}",

note = "Publisher Copyright: {\textcopyright} COPYRIGHT SPIE. Downloading of the abstract is permitted for personal use only.; Medical Imaging 2023: Biomedical Applications in Molecular, Structural, and Functional Imaging ; Conference date: 19-02-2023 Through 22-02-2023",

year = "2023",

doi = "10.1117/12.2647897",

language = "English (US)",

series = "Progress in Biomedical Optics and Imaging - Proceedings of SPIE",

publisher = "SPIE",

editor = "Gimi, \{Barjor S.\} and Andrzej Krol",

booktitle = "Medical Imaging 2023",

address = "United States",

}

Chang, Q, Ahmad, D, Toth, J , Bascom, R & Higgins, WE 2023, ESFPNet: Efficient deep learning architecture for real-time lesion segmentation in autofluorescence bronchoscopic video. in BS Gimi & A Krol (eds), Medical Imaging 2023: Biomedical Applications in Molecular, Structural, and Functional Imaging., 1246803, Progress in Biomedical Optics and Imaging - Proceedings of SPIE, vol. 12468, SPIE, Medical Imaging 2023: Biomedical Applications in Molecular, Structural, and Functional Imaging, San Diego, United States, 2/19/23. https://doi.org/10.1117/12.2647897

ESFPNet: Efficient deep learning architecture for real-time lesion segmentation in autofluorescence bronchoscopic video. / Chang, Qi; Ahmad, Danish; Toth, Jennifer et al.
Medical Imaging 2023: Biomedical Applications in Molecular, Structural, and Functional Imaging. ed. / Barjor S. Gimi; Andrzej Krol. SPIE, 2023. 1246803 (Progress in Biomedical Optics and Imaging - Proceedings of SPIE; Vol. 12468).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

TY - GEN

T1 - ESFPNet

T2 - Medical Imaging 2023: Biomedical Applications in Molecular, Structural, and Functional Imaging

AU - Chang, Qi

AU - Ahmad, Danish

AU - Toth, Jennifer

AU - Bascom, Rebecca

AU - Higgins, William E.

PY - 2023

Y1 - 2023

N2 - Lung cancer tends to be detected at an advanced stage, resulting in a high patient mortality rate. Thus, recent research has focused on early disease detection. Autofluorescence bronchoscopy (AFB) is an effective noninvasive way of detecting early manifestations of lung cancer. Unfortunately, manual inspection of AFB video is extremely tedious and error-prone, while limited effort has been expended toward potentially more robust automatic AFB lesion analysis. We propose a real-time deep-learning architecture dubbed ESFPNet for accurate segmentation and robust detection of bronchial lesions in an AFB video stream. Our approach gives the best segmentation results (mDice = 0.756, mIoU=0.624) on our AFB dataset among recent architectures. Moreover, our model shows promising potential applicability to other domains, as evidenced by its state-of-the-art (SOTA) performance on the CVC-ClinicDB,ETIS-LaribPolypDB datasets, and superior performance on the Kvasir, CVC-ColonDB datasets.

AB - Lung cancer tends to be detected at an advanced stage, resulting in a high patient mortality rate. Thus, recent research has focused on early disease detection. Autofluorescence bronchoscopy (AFB) is an effective noninvasive way of detecting early manifestations of lung cancer. Unfortunately, manual inspection of AFB video is extremely tedious and error-prone, while limited effort has been expended toward potentially more robust automatic AFB lesion analysis. We propose a real-time deep-learning architecture dubbed ESFPNet for accurate segmentation and robust detection of bronchial lesions in an AFB video stream. Our approach gives the best segmentation results (mDice = 0.756, mIoU=0.624) on our AFB dataset among recent architectures. Moreover, our model shows promising potential applicability to other domains, as evidenced by its state-of-the-art (SOTA) performance on the CVC-ClinicDB,ETIS-LaribPolypDB datasets, and superior performance on the Kvasir, CVC-ColonDB datasets.

UR - https://www.scopus.com/pages/publications/85160856121

UR - https://www.scopus.com/pages/publications/85160856121#tab=citedBy

U2 - 10.1117/12.2647897

DO - 10.1117/12.2647897

M3 - Conference contribution

AN - SCOPUS:85160856121

T3 - Progress in Biomedical Optics and Imaging - Proceedings of SPIE

BT - Medical Imaging 2023

A2 - Gimi, Barjor S.

A2 - Krol, Andrzej

PB - SPIE

Y2 - 19 February 2023 through 22 February 2023

ER -

Chang Q, Ahmad D, Toth J , Bascom R , Higgins WE. ESFPNet: Efficient deep learning architecture for real-time lesion segmentation in autofluorescence bronchoscopic video. In Gimi BS, Krol A, editors, Medical Imaging 2023: Biomedical Applications in Molecular, Structural, and Functional Imaging. SPIE. 2023. 1246803. (Progress in Biomedical Optics and Imaging - Proceedings of SPIE). doi: 10.1117/12.2647897

ESFPNet: Efficient deep learning architecture for real-time lesion segmentation in autofluorescence bronchoscopic video

Abstract

Publication series

Conference

UN SDGs

All Science Journal Classification (ASJC) codes

Access to Document

Other files and links

Fingerprint

Cite this