Dermal features derived from optoacoustic tomograms via machine learning correlate microangiopathy phenotypes with diabetes stage

Angelos Karlas; Nikoletta Katsouli; Nikolina Alexia Fasoula; Michael Bariotakis; Nikolaos Kosmas Chlis; Murad Omar; Hailong He; Dimitrios Iakovakis; Christoph J. Schaffer; Michael A. Kallmayer; Martin Füchtenbusch; Annette Ziegler; Hans Henning Eckstein; Leontios J. Hadjileontiadis; Vasilis Ntziachristos

doi:10.1038/s41551-023-01151-w

Dermal features derived from optoacoustic tomograms via machine learning correlate microangiopathy phenotypes with diabetes stage

Angelos Karlas, Nikoletta Katsouli, Nikolina Alexia Fasoula, Michael Bariotakis, Nikolaos Kosmas Chlis, Murad Omar, Hailong He, Dimitrios Iakovakis, Christoph J. Schaffer, Michael A. Kallmayer, Martin Füchtenbusch, Annette Ziegler, Hans Henning Eckstein, Leontios J. Hadjileontiadis, Vasilis Ntziachristos

Biomedical Engineering & Biotechnology

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

15 Scopus citations

Abstract

Skin microangiopathy has been associated with diabetes. Here we show that skin-microangiopathy phenotypes in humans can be correlated with diabetes stage via morphophysiological cutaneous features extracted from raster-scan optoacoustic mesoscopy (RSOM) images of skin on the leg. We obtained 199 RSOM images from 115 participants (40 healthy and 75 with diabetes), and used machine learning to segment skin layers and microvasculature to identify clinically explainable features pertaining to different depths and scales of detail that provided the highest predictive power. Features in the dermal layer at the scale of detail of 0.1–1 mm (such as the number of junction-to-junction branches) were highly sensitive to diabetes stage. A ‘microangiopathy score’ compiling the 32 most-relevant features predicted the presence of diabetes with an area under the receiver operating characteristic curve of 0.84. The analysis of morphophysiological cutaneous features via RSOM may allow for the discovery of diabetes biomarkers in the skin and for the monitoring of diabetes status. © 2023, The Author(s).

Original language	American English
Pages (from-to)	1667-1682
Number of pages	16
Journal	Nature Biomedical Engineering
Volume	7
Issue number	12
DOIs	https://doi.org/10.1038/s41551-023-01151-w
State	Published - 2023

Keywords

Machine learning
Dermal layers
Machine-learning
Mesoscopies
Predictive power
Raster scans
Receiver operating characteristic curves
Relevant features
Skin layer
Skin microvasculature
Image segmentation
diabetes mellitus
diagnostic imaging
human
machine learning
phenotype
photoacoustics
procedures
skin
vascularization
Diabetes Mellitus
Humans
Machine Learning
Phenotype
Photoacoustic Techniques
Skin

UN SDGs

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

Access to Document

10.1038/s41551-023-01151-w

https://www.scopus.com/inward/record.uri?eid=2-s2.0-85178420467&doi=10.1038%2fs41551-023-01151-w&partnerID=40&md5=4f95e94bd760b6b92a265c67f15095ec

Cite this

Karlas, A., Katsouli, N., Fasoula, N. A., Bariotakis, M., Chlis, N. K., Omar, M., He, H., Iakovakis, D., Schaffer, C. J., Kallmayer, M. A., Füchtenbusch, M., Ziegler, A., Eckstein, H. H., Hadjileontiadis, L. J., & Ntziachristos, V. (2023). Dermal features derived from optoacoustic tomograms via machine learning correlate microangiopathy phenotypes with diabetes stage. Nature Biomedical Engineering, 7(12), 1667-1682. https://doi.org/10.1038/s41551-023-01151-w

@article{9c3ec4a267a94b9cbab4212ddd99a79e,

title = "Dermal features derived from optoacoustic tomograms via machine learning correlate microangiopathy phenotypes with diabetes stage",

abstract = "Skin microangiopathy has been associated with diabetes. Here we show that skin-microangiopathy phenotypes in humans can be correlated with diabetes stage via morphophysiological cutaneous features extracted from raster-scan optoacoustic mesoscopy (RSOM) images of skin on the leg. We obtained 199 RSOM images from 115 participants (40 healthy and 75 with diabetes), and used machine learning to segment skin layers and microvasculature to identify clinically explainable features pertaining to different depths and scales of detail that provided the highest predictive power. Features in the dermal layer at the scale of detail of 0.1–1 mm (such as the number of junction-to-junction branches) were highly sensitive to diabetes stage. A {\textquoteleft}microangiopathy score{\textquoteright} compiling the 32 most-relevant features predicted the presence of diabetes with an area under the receiver operating characteristic curve of 0.84. The analysis of morphophysiological cutaneous features via RSOM may allow for the discovery of diabetes biomarkers in the skin and for the monitoring of diabetes status. {\textcopyright} 2023, The Author(s).",

keywords = "Machine learning, Dermal layers, Machine-learning, Mesoscopies, Predictive power, Raster scans, Receiver operating characteristic curves, Relevant features, Skin layer, Skin microvasculature, Image segmentation, diabetes mellitus, diagnostic imaging, human, machine learning, phenotype, photoacoustics, procedures, skin, vascularization, Diabetes Mellitus, Humans, Machine Learning, Phenotype, Photoacoustic Techniques, Skin",

author = "Angelos Karlas and Nikoletta Katsouli and Fasoula, {Nikolina Alexia} and Michael Bariotakis and Chlis, {Nikolaos Kosmas} and Murad Omar and Hailong He and Dimitrios Iakovakis and Schaffer, {Christoph J.} and Kallmayer, {Michael A.} and Martin F{\"u}chtenbusch and Annette Ziegler and Eckstein, {Hans Henning} and Hadjileontiadis, {Leontios J.} and Vasilis Ntziachristos",

note = "cited By 0",

year = "2023",

doi = "10.1038/s41551-023-01151-w",

language = "American English",

volume = "7",

pages = "1667--1682",

journal = "Nature Biomedical Engineering",

number = "12",

}

Karlas, A, Katsouli, N, Fasoula, NA, Bariotakis, M, Chlis, NK, Omar, M, He, H, Iakovakis, D, Schaffer, CJ, Kallmayer, MA, Füchtenbusch, M, Ziegler, A, Eckstein, HH, Hadjileontiadis, LJ & Ntziachristos, V 2023, 'Dermal features derived from optoacoustic tomograms via machine learning correlate microangiopathy phenotypes with diabetes stage', Nature Biomedical Engineering, vol. 7, no. 12, pp. 1667-1682. https://doi.org/10.1038/s41551-023-01151-w

TY - JOUR

T1 - Dermal features derived from optoacoustic tomograms via machine learning correlate microangiopathy phenotypes with diabetes stage

AU - Karlas, Angelos

AU - Katsouli, Nikoletta

AU - Fasoula, Nikolina Alexia

AU - Bariotakis, Michael

AU - Chlis, Nikolaos Kosmas

AU - Omar, Murad

AU - He, Hailong

AU - Iakovakis, Dimitrios

AU - Schaffer, Christoph J.

AU - Kallmayer, Michael A.

AU - Füchtenbusch, Martin

AU - Ziegler, Annette

AU - Eckstein, Hans Henning

AU - Hadjileontiadis, Leontios J.

AU - Ntziachristos, Vasilis

N1 - cited By 0

PY - 2023

Y1 - 2023

N2 - Skin microangiopathy has been associated with diabetes. Here we show that skin-microangiopathy phenotypes in humans can be correlated with diabetes stage via morphophysiological cutaneous features extracted from raster-scan optoacoustic mesoscopy (RSOM) images of skin on the leg. We obtained 199 RSOM images from 115 participants (40 healthy and 75 with diabetes), and used machine learning to segment skin layers and microvasculature to identify clinically explainable features pertaining to different depths and scales of detail that provided the highest predictive power. Features in the dermal layer at the scale of detail of 0.1–1 mm (such as the number of junction-to-junction branches) were highly sensitive to diabetes stage. A ‘microangiopathy score’ compiling the 32 most-relevant features predicted the presence of diabetes with an area under the receiver operating characteristic curve of 0.84. The analysis of morphophysiological cutaneous features via RSOM may allow for the discovery of diabetes biomarkers in the skin and for the monitoring of diabetes status. © 2023, The Author(s).

AB - Skin microangiopathy has been associated with diabetes. Here we show that skin-microangiopathy phenotypes in humans can be correlated with diabetes stage via morphophysiological cutaneous features extracted from raster-scan optoacoustic mesoscopy (RSOM) images of skin on the leg. We obtained 199 RSOM images from 115 participants (40 healthy and 75 with diabetes), and used machine learning to segment skin layers and microvasculature to identify clinically explainable features pertaining to different depths and scales of detail that provided the highest predictive power. Features in the dermal layer at the scale of detail of 0.1–1 mm (such as the number of junction-to-junction branches) were highly sensitive to diabetes stage. A ‘microangiopathy score’ compiling the 32 most-relevant features predicted the presence of diabetes with an area under the receiver operating characteristic curve of 0.84. The analysis of morphophysiological cutaneous features via RSOM may allow for the discovery of diabetes biomarkers in the skin and for the monitoring of diabetes status. © 2023, The Author(s).

KW - Machine learning

KW - Dermal layers

KW - Machine-learning

KW - Mesoscopies

KW - Predictive power

KW - Raster scans

KW - Receiver operating characteristic curves

KW - Relevant features

KW - Skin layer

KW - Skin microvasculature

KW - Image segmentation

KW - diabetes mellitus

KW - diagnostic imaging

KW - human

KW - machine learning

KW - phenotype

KW - photoacoustics

KW - procedures

KW - skin

KW - vascularization

KW - Diabetes Mellitus

KW - Humans

KW - Machine Learning

KW - Phenotype

KW - Photoacoustic Techniques

KW - Skin

U2 - 10.1038/s41551-023-01151-w

DO - 10.1038/s41551-023-01151-w

M3 - Article

C2 - 38049470

VL - 7

SP - 1667

EP - 1682

JO - Nature Biomedical Engineering

JF - Nature Biomedical Engineering

IS - 12

ER -

Dermal features derived from optoacoustic tomograms via machine learning correlate microangiopathy phenotypes with diabetes stage

Abstract

Keywords

UN SDGs

Access to Document

Fingerprint

Cite this