The site of the Media Sphera Publishers contains materials intended solely for healthcare professionals.
By closing this message, you confirm that you are a certified medical professional or a student of a medical educational institution.

Login
EN
+7 (495) 482-4118
+7 (495) 482-4329
+8 800 250-18-12
  • (toll-free number for subscriptions)
    Mon-Fri from 10 a.m. to 6 p.m.

  • © 1998-2026
+7 (495) 482-43-29
EN
All journals
Journal
Year
Year
Search result: 0

Koshechkin K.A.

The First Sechenov Moscow State Medical University

Ignat’ev A.A.

Laboratory of Advanced Technologies

Potekaev N.N.

Moscow Research and Practical Center for Dermatovenereology and Cosmetology, Department of Healthcare;
Pirogov Russian National Research Medical University (RNRMU)

Dolya O.V.

Moscow Research and Practical Center for Dermatovenereology and Cosmetology

Frigo N.V.

Moscow Scientic and Practical Center of Dermatvenerology and Cosmetology;
Central state medical academy of department of presidential affairs

Kochetkov M.A.

Moscow Research and Practical Center for Dermatovenereology and Cosmetology of the Department of Healthcare

Selection of a neural network model for early detection of skin melanoma

Authors:

Koshechkin K.A., Ignat’ev A.A., Potekaev N.N., Dolya O.V., Frigo N.V., Kochetkov M.A.

More about the authors

Journal: Russian Journal of Clinical Dermatology and Venereology. 2023;22(3): 287‑295

DOI: 10.17116/klinderma202322031287

Read: 2828 times

Contact the author
Table of contents

SELECTION OF A NEURAL NETWORK MODEL FOR EARLY DETECTION OF SKIN MELANOMA
SELECTION OF A NEURAL NETWORK MODEL FOR EARLY DETECTION OF SKIN MELANOMA

To cite this article:

Koshechkin KA, Ignat’ev AA, Potekaev NN, Dolya OV, Frigo NV, Kochetkov MA. Selection of a neural network model for early detection of skin melanoma. Russian Journal of Clinical Dermatology and Venereology. 2023;22(3):287‑295. (In Russ.)
https://doi.org/10.17116/klinderma202322031287

Подписка 2026 Клиническая дерматология и венерология (Russian Journal of Clinical Dermatology and Venereology)
МСФ на ЯМ
Использование инфографики авторами научных работ
Close metadata
Abstract:

The approach to diagnosis and treatment in dermatovenerology has changed significantly with the introduction of new technologies and inventions. Computer algorithms have been used to assist dermatovenerologists in diagnosing diseases, including malignant melanoma of the skin. Instrumental methods for diagnosing skin melanoma, based on the processing of dermatoscopic images, can be improved by using artificial intelligence. Machine learning is a subset of artificial intelligence in which computer programs automatically learn from experience without explicit operation instructions. The most promising direction is deep learning of artificial neural networks. In this paper, several models of artificial neural networks were studied to assess the prospects for their use for the early diagnosis of skin melanoma. The efficacy of the various models was analyzed using the HAM10000 dataset image classification. Models of convolutional neural networks, pre-trained neural networks with transfer learning, and the “soft attention” mechanism were evaluated: Xception, ResNet50, RAN50, SEnet50, ARL-CNN50, Inception Resnet V2 — IRv2 12x12+SA, IRv2 5x5+SA. The results showed that the most promising model was Inception Resnet V2 with the addition of a «soft attention» mechanism.

Keywords:

melanoma
HAM10000
convolutional neural network
Xception
«soft attention»

Authors:

Koshechkin K.A.

The First Sechenov Moscow State Medical University

Ignat’ev A.A.

Laboratory of Advanced Technologies

Potekaev N.N.

Moscow Research and Practical Center for Dermatovenereology and Cosmetology, Department of Healthcare;
Pirogov Russian National Research Medical University (RNRMU)

Dolya O.V.

Moscow Research and Practical Center for Dermatovenereology and Cosmetology

Frigo N.V.

Moscow Scientic and Practical Center of Dermatvenerology and Cosmetology;
Central state medical academy of department of presidential affairs

Kochetkov M.A.

Moscow Research and Practical Center for Dermatovenereology and Cosmetology of the Department of Healthcare

Received:

16.11.2022

Accepted:

10.04.2023

Close metadata

References:

  1. De A, et al. Use of Artificial Intelligence in Dermatology. Indian J. Dermatol. Wolters Kluwer. Medknow Publications. 2020;65(5):352. 
  2. Dinnes J, et al. Dermoscopy, with and without visual inspection, for diagnosing melanoma in adults. Cochrane Database Syst. Rev. John Wiley and Sons, Inc. and the Cochrane Library. 2018;2018:12. 
  3. Charalambides M, Singh S. Artificial intelligence and melanoma detection: friend or foe of dermatologists? MA Healthcare London. 2020;81:1.  https://doi.org/10.12968/hmed.2019.0322
  4. Pouly M, et al. Artificial intelligence in image analysis—fundamentals and new developments. Hautarzt. Springer Medizin. 2020;71(9):660-668. 
  5. Elder A, et al. The role of artificial intelligence in cosmetic dermatology — Current, upcoming, and future trends. J Cosmet Dermatol. John Wiley & Sons, Ltd. 2021;20(1):48-52. 
  6. Gasanovich MA, Mustafaev AG. The use of artificial neural networks for early diagnosis of diabetes mellitus. Kibernetika i programmirovanie. Aurora Group, s.r.o. 2016;2(2):1-7. (In Russ.)
  7. Chollet F. Xception: Deep Learning With Depthwise Separable Convolutions. 2017;1251-1258.
  8. The HAM10000 dataset, a large collection of multi-source dermatoscopic images of common pigmented skin lesions — ViDIR Dataverse [Electronic resource]. Accessed: 06.11.22.  https://dataverse.harvard.edu/dataset.xhtml?persistentId=doi:10.7910/DVN/DBW86T
  9. Goodfellow I, Bengio Y, Courville A. Deep Learning [Electronic resource]. MIT Press. 2016. Accessed: 06.11.22.  https://www.deeplearningbook.org/
  10. He K, et al. Deep Residual Learning for Image Recognition. Proc IEEE Comput Soc Conf Comput Vis Pattern Recognit. IEEE Computer Society. 2015;2016-December:770-778. 
  11. Szegedy C, et al. Going Deeper with Convolutions. Proc IEEE Comput Soc Conf Comput Vis Pattern Recognit. IEEE Computer Society. 2014; 07-12- June-2015:1-9. 
  12. Xu K, et al. Show, Attend and Tell: Neural Image Caption Generation with Visual Attention. PMLR. 2015;2048-2057.
  13. Shaikh MA, et al. Attention based Writer Independent Verification. Proc Int Conf Front Handwrit Recognition, ICFHR. Institute of Electrical and Electronics Engineers Inc. 2020;2020-September:373-379. 
  14. Tomita N, et al. Attention-Based Deep Neural Networks for Detection of Cancerous and Precancerous Esophagus Tissue on Histopathological Slides. JAMA Netw Open American Medical Association. 2018;2:11. 
  15. Srivastava N, et al. Dropout: A Simple Way to Prevent Neural Networks from Overfitting. J Mach Learn Res. 2014;15(56):1929-1958.
  16. Szegedy C, et al. Inception-v4, Inception-ResNet and the Impact of Residual Connections on Learning. 31st AAAI Conf. Artif. Intell. AAAI 2017. AAAI press. 2016;4278-4284.
  17. Simonyan K, Zisserman A. Very deep convolutional networks for large-scale image recognition. University of Oxford. 2015.
  18. Zhang J, et al. Attention Residual Learning for Skin Lesion Classification. IEEE Trans. Med. Imaging. NLM (Medline). 2019;38(9):2092-2103.
  19. Selvaraju RR, et al. Grad-CAM: Visual Explanations from Deep Networks via Gradient-based Localization. International Journal of Computer Vision. 2020;128:336-359. 
Contact the author
Email
Message
The publishing house "Media Sphere" does not provide treatment services, does not give recommendations on contacting medical institutions and specific specialists, on the prescription of medicines and dietary supplements.

Email Confirmation

An email was sent to test@gmail.com with a confirmation link. Follow the link from the letter to complete the registration on the site.

Email Confirmation

Contact us
If you have any questions, complaints or suggestions, please use this feedback form.
Email
Message
The publishing house "Media Sphere" does not provide treatment services, does not give recommendations on contacting medical institutions and specific specialists, on the prescription of medicines and dietary supplements.
Submit Application

You can become an author of one of our magazines, send a request and we will consider it in during the day.

Name
Phone
E-mail
Message
Login
Registration
E-mail
Password
Forgot Password?

Log in to the site using your account in one of the services

Password recovery
E-mail
Login
Register

We use cооkies to improve the performance of the site. By staying on our site, you agree to the terms of use of cооkies. To view our Privacy and Cookie Policy, please. click here.