Unsupervised Segmentation of Wounds in Optical Coherence Tomography Images Using Invariant Information Clustering

Julia Andresen; Timo Kepp; Michael Wang-Evers; Jan Ehrhardt; Dieter Manstein; Heinz Handels

Unsupervised Segmentation of Wounds in Optical Coherence Tomography Images Using Invariant Information Clustering

Julia Andresen^*, Timo Kepp, Michael Wang-Evers, Jan Ehrhardt, Dieter Manstein, Heinz Handels

^*Corresponding author for this work

Harvard University

1 Citation (Scopus)

Abstract

Monitoring wound healing with optical coherence tomography (OCT) imaging is a promising research field. So far, however, few data and even less manual annotations of OCT wound images are available. To address this problem, a fully unsupervised clustering method based on convolutional neural networks (CNNs) is presented. The CNN takes image patches as input and assigns them to either wound or healthy skin clusters. Network training is based on a new combination of loss functions that require information invariance and locality preservation. No expensive expert annotations are needed. Locality preservation is applied to different levels of the network and shown to improve the segmentation. Promising results are achieved with an average Dice score of 0.809 and an average rand index of 0.871 for the best performing network version.

Original language	English
Title of host publication	Bildverarbeitung für die Medizin 2022 : Proceedings of the German Workshop on Medical Image Computing
Number of pages	6
Publication date	05.04.2022
Pages	1-6
Publication status	Published - 05.04.2022
Event	Bildverarbeitung für die Medizin 2022: Proceedings, German Workshop on Medical Image Computing - Heidelberg, Heidelberg, Germany Duration: 26.06.2022 → 28.06.2022

UN SDGs

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

SDG 3 Good Health and Well-being
SDG 4 Quality Education
SDG 9 Industry, Innovation, and Infrastructure
SDG 11 Sustainable Cities and Communities
SDG 12 Responsible Consumption and Production
SDG 14 Life Below Water
SDG 15 Life on Land

Research Areas and Centers

Centers: Center for Artificial Intelligence Luebeck (ZKIL)

Cite this

@inproceedings{365479a337eb46c585650dc43090c48a,

title = "Unsupervised Segmentation of Wounds in Optical Coherence Tomography Images Using Invariant Information Clustering",

abstract = "Monitoring wound healing with optical coherence tomography (OCT) imaging is a promising research field. So far, however, few data and even less manual annotations of OCT wound images are available. To address this problem, a fully unsupervised clustering method based on convolutional neural networks (CNNs) is presented. The CNN takes image patches as input and assigns them to either wound or healthy skin clusters. Network training is based on a new combination of loss functions that require information invariance and locality preservation. No expensive expert annotations are needed. Locality preservation is applied to different levels of the network and shown to improve the segmentation. Promising results are achieved with an average Dice score of 0.809 and an average rand index of 0.871 for the best performing network version.",

author = "Julia Andresen and Timo Kepp and Michael Wang-Evers and Jan Ehrhardt and Dieter Manstein and Heinz Handels",

year = "2022",

month = apr,

day = "5",

language = "English",

series = "Informatik Aktuell",

publisher = "Springer",

pages = "1--6",

booktitle = "Bildverarbeitung f{\"u}r die Medizin 2022",

note = "Bildverarbeitung f{\"u}r die Medizin 2022 ; Conference date: 26-06-2022 Through 28-06-2022",

}

Andresen, J, Kepp, T, Wang-Evers, M, Ehrhardt, J, Manstein, D & Handels, H 2022, Unsupervised Segmentation of Wounds in Optical Coherence Tomography Images Using Invariant Information Clustering. in Bildverarbeitung für die Medizin 2022: Proceedings of the German Workshop on Medical Image Computing. Informatik Aktuell, pp. 1-6, Bildverarbeitung für die Medizin 2022, Heidelberg, Germany, 26.06.22.

Unsupervised Segmentation of Wounds in Optical Coherence Tomography Images Using Invariant Information Clustering. / Andresen, Julia; Kepp, Timo; Wang-Evers, Michael et al.
Bildverarbeitung für die Medizin 2022: Proceedings of the German Workshop on Medical Image Computing. 2022. p. 1-6 (Informatik Aktuell).

Research output: Chapters in Books/Reports/Conference Proceedings › Conference contribution › peer-review

TY - GEN

T1 - Unsupervised Segmentation of Wounds in Optical Coherence Tomography Images Using Invariant Information Clustering

AU - Andresen, Julia

AU - Kepp, Timo

AU - Wang-Evers, Michael

AU - Ehrhardt, Jan

AU - Manstein, Dieter

AU - Handels, Heinz

PY - 2022/4/5

Y1 - 2022/4/5

N2 - Monitoring wound healing with optical coherence tomography (OCT) imaging is a promising research field. So far, however, few data and even less manual annotations of OCT wound images are available. To address this problem, a fully unsupervised clustering method based on convolutional neural networks (CNNs) is presented. The CNN takes image patches as input and assigns them to either wound or healthy skin clusters. Network training is based on a new combination of loss functions that require information invariance and locality preservation. No expensive expert annotations are needed. Locality preservation is applied to different levels of the network and shown to improve the segmentation. Promising results are achieved with an average Dice score of 0.809 and an average rand index of 0.871 for the best performing network version.

AB - Monitoring wound healing with optical coherence tomography (OCT) imaging is a promising research field. So far, however, few data and even less manual annotations of OCT wound images are available. To address this problem, a fully unsupervised clustering method based on convolutional neural networks (CNNs) is presented. The CNN takes image patches as input and assigns them to either wound or healthy skin clusters. Network training is based on a new combination of loss functions that require information invariance and locality preservation. No expensive expert annotations are needed. Locality preservation is applied to different levels of the network and shown to improve the segmentation. Promising results are achieved with an average Dice score of 0.809 and an average rand index of 0.871 for the best performing network version.

M3 - Conference contribution

T3 - Informatik Aktuell

SP - 1

EP - 6

BT - Bildverarbeitung für die Medizin 2022

T2 - Bildverarbeitung für die Medizin 2022

Y2 - 26 June 2022 through 28 June 2022

ER -

Unsupervised Segmentation of Wounds in Optical Coherence Tomography Images Using Invariant Information Clustering

Abstract

UN SDGs

Research Areas and Centers

Fingerprint

Cite this