Towards to deep neural network application with limited training data: Synthesis of melanoma's diffuse reflectance spectral images

Katrina Bolochko; Dmitrijs Bliznuks; Dilshat Uteshev; Ilze Lihacova; Alexey Lihachev; Yuriy Chizhov; Andrey Bondarenko

doi:10.1117/12.2527173

Towards to deep neural network application with limited training data: Synthesis of melanoma's diffuse reflectance spectral images

Katrina Bolochko^*
, Dmitrijs Bliznuks
, Dilshat Uteshev
, Ilze Lihacova
, Alexey Lihachev
, Yuriy Chizhov
, Andrey Bondarenko

^*Corresponding author for this work

Institute of Atomic Physics and Spectroscopy

Riga Technical University
C.T.Co. LTD

Research output: Chapter in Book/Report/Conference proceeding › Conference paper › Research › peer-review

1 Citation (Scopus)

Abstract

The goal of our study is to train artificial neural networks (ANN) using multispectral images of melanoma. Since the number of multispectral images of melanomas is limited, we offer to synthesize them from multispectral images of benign skin lesions. We used the previously created melanoma diagnostic criterion p'. This criterion is calculated from multispectral images of skin lesions captured under 526nm, 663nm, and 964nm LED illumination. We synthesize these three images from multispectral images of nevus so that the p' map matches the melanoma criteria (the values in the lesion area is >1, respectively). Demonstrated results show that by transforming multispectral images of benign nevus is possible to get a reliable multispectral images of melanoma usable for ANN training.

Original language	English
Title of host publication	Diffuse Optical Spectroscopy and Imaging VII
Editors	Hamid Dehghani, Heidrun Wabnitz
Publisher	SPIE
ISBN (Electronic)	9781510628410
DOIs	https://doi.org/10.1117/12.2527173
Publication status	Published - 2019
Event	Diffuse Optical Spectroscopy and Imaging VII 2019 - Munich, Germany Duration: 23 Jun 2019 → 25 Jun 2019

Publication series

Name	Progress in Biomedical Optics and Imaging - Proceedings of SPIE
Volume	11074
ISSN (Print)	1605-7422

Conference

Conference	Diffuse Optical Spectroscopy and Imaging VII 2019
Country/Territory	Germany
City	Munich
Period	23/06/19 → 25/06/19

UN SDGs

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

SDG 3 Good Health and Well-being

Keywords

artificial neural networks
diffuse reflectance
early diagnostics
skin cancer

Access to Document

10.1117/12.2527173

Cite this

Bolochko, K., Bliznuks, D., Uteshev, D., Lihacova, I., Lihachev, A., Chizhov, Y., & Bondarenko, A. (2019). Towards to deep neural network application with limited training data: Synthesis of melanoma's diffuse reflectance spectral images. In H. Dehghani, & H. Wabnitz (Eds.), Diffuse Optical Spectroscopy and Imaging VII Article 110741O (Progress in Biomedical Optics and Imaging - Proceedings of SPIE; Vol. 11074). SPIE. https://doi.org/10.1117/12.2527173

@inproceedings{732e64880513430cb0ef49a0b1408922,

title = "Towards to deep neural network application with limited training data: Synthesis of melanoma's diffuse reflectance spectral images",

abstract = "The goal of our study is to train artificial neural networks (ANN) using multispectral images of melanoma. Since the number of multispectral images of melanomas is limited, we offer to synthesize them from multispectral images of benign skin lesions. We used the previously created melanoma diagnostic criterion p'. This criterion is calculated from multispectral images of skin lesions captured under 526nm, 663nm, and 964nm LED illumination. We synthesize these three images from multispectral images of nevus so that the p' map matches the melanoma criteria (the values in the lesion area is >1, respectively). Demonstrated results show that by transforming multispectral images of benign nevus is possible to get a reliable multispectral images of melanoma usable for ANN training.",

keywords = "artificial neural networks, diffuse reflectance, early diagnostics, skin cancer",

author = "Katrina Bolochko and Dmitrijs Bliznuks and Dilshat Uteshev and Ilze Lihacova and Alexey Lihachev and Yuriy Chizhov and Andrey Bondarenko",

note = "Publisher Copyright: {\textcopyright} 2019 SPIE.; Diffuse Optical Spectroscopy and Imaging VII 2019 ; Conference date: 23-06-2019 Through 25-06-2019",

year = "2019",

doi = "10.1117/12.2527173",

language = "English",

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

publisher = "SPIE",

editor = "Hamid Dehghani and Heidrun Wabnitz",

booktitle = "Diffuse Optical Spectroscopy and Imaging VII",

address = "United States",

}

Bolochko, K, Bliznuks, D, Uteshev, D, Lihacova, I , Lihachev, A, Chizhov, Y & Bondarenko, A 2019, Towards to deep neural network application with limited training data: Synthesis of melanoma's diffuse reflectance spectral images. in H Dehghani & H Wabnitz (eds), Diffuse Optical Spectroscopy and Imaging VII., 110741O, Progress in Biomedical Optics and Imaging - Proceedings of SPIE, vol. 11074, SPIE, Diffuse Optical Spectroscopy and Imaging VII 2019, Munich, Germany, 23/06/19. https://doi.org/10.1117/12.2527173

Towards to deep neural network application with limited training data: Synthesis of melanoma's diffuse reflectance spectral images. / Bolochko, Katrina; Bliznuks, Dmitrijs; Uteshev, Dilshat et al.
Diffuse Optical Spectroscopy and Imaging VII. ed. / Hamid Dehghani; Heidrun Wabnitz. SPIE, 2019. 110741O (Progress in Biomedical Optics and Imaging - Proceedings of SPIE; Vol. 11074).

Research output: Chapter in Book/Report/Conference proceeding › Conference paper › Research › peer-review

TY - GEN

T1 - Towards to deep neural network application with limited training data

T2 - Diffuse Optical Spectroscopy and Imaging VII 2019

AU - Bolochko, Katrina

AU - Bliznuks, Dmitrijs

AU - Uteshev, Dilshat

AU - Lihacova, Ilze

AU - Lihachev, Alexey

AU - Chizhov, Yuriy

AU - Bondarenko, Andrey

PY - 2019

Y1 - 2019

N2 - The goal of our study is to train artificial neural networks (ANN) using multispectral images of melanoma. Since the number of multispectral images of melanomas is limited, we offer to synthesize them from multispectral images of benign skin lesions. We used the previously created melanoma diagnostic criterion p'. This criterion is calculated from multispectral images of skin lesions captured under 526nm, 663nm, and 964nm LED illumination. We synthesize these three images from multispectral images of nevus so that the p' map matches the melanoma criteria (the values in the lesion area is >1, respectively). Demonstrated results show that by transforming multispectral images of benign nevus is possible to get a reliable multispectral images of melanoma usable for ANN training.

AB - The goal of our study is to train artificial neural networks (ANN) using multispectral images of melanoma. Since the number of multispectral images of melanomas is limited, we offer to synthesize them from multispectral images of benign skin lesions. We used the previously created melanoma diagnostic criterion p'. This criterion is calculated from multispectral images of skin lesions captured under 526nm, 663nm, and 964nm LED illumination. We synthesize these three images from multispectral images of nevus so that the p' map matches the melanoma criteria (the values in the lesion area is >1, respectively). Demonstrated results show that by transforming multispectral images of benign nevus is possible to get a reliable multispectral images of melanoma usable for ANN training.

KW - artificial neural networks

KW - diffuse reflectance

KW - early diagnostics

KW - skin cancer

UR - https://www.scopus.com/pages/publications/85076892143

U2 - 10.1117/12.2527173

DO - 10.1117/12.2527173

M3 - Conference paper

AN - SCOPUS:85076892143

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

BT - Diffuse Optical Spectroscopy and Imaging VII

A2 - Dehghani, Hamid

A2 - Wabnitz, Heidrun

PB - SPIE

Y2 - 23 June 2019 through 25 June 2019

ER -

Bolochko K, Bliznuks D, Uteshev D, Lihacova I , Lihachev A, Chizhov Y et al. Towards to deep neural network application with limited training data: Synthesis of melanoma's diffuse reflectance spectral images. In Dehghani H, Wabnitz H, editors, Diffuse Optical Spectroscopy and Imaging VII. SPIE. 2019. 110741O. (Progress in Biomedical Optics and Imaging - Proceedings of SPIE). doi: 10.1117/12.2527173

Towards to deep neural network application with limited training data: Synthesis of melanoma's diffuse reflectance spectral images

Abstract

Publication series

Conference

UN SDGs

Keywords

Access to Document

Other files and links

Fingerprint

Cite this