Skip to Main content Skip to Navigation
Conference papers

Design of a System for Melanoma Detection Through the Processing of Clinical Images Using Artificial Neural Networks

Abstract : Skin cancer is one of the most important challenges in modern medicine, especially skin melanoma, being the main causer of deaths for this disease. Images analysis is one of the most transcendental techniques for Melanoma early detection as a prevention method. Artificial neural networks are one of the many developed techniques for images digital processing and characteristic similarities detection. In this work a graphic processing unit (GPU) is developed for clinical skin images analysis getting through an artificial neural networks system for similar patterns detection through processing in a collection of modules tasked of silhouette detection of the object to analyze into the image, and tasked to study borders or contour to determinate a final diagnostic, the dataset used for the training of the artificial neural network designed is gotten from the MED-NODE project and project of international skin images collaboration (ISIC) with 730 images of positive and negative cases as full, the proposed system presents finally an accuracy level of 76.67%, with a level of success of 78.79% in melanoma specific cases, and 74.07% in benign lesions cases.
Complete list of metadata

Cited literature [21 references]  Display  Hide  Download

https://hal.inria.fr/hal-02274187
Contributor : Hal Ifip <>
Submitted on : Thursday, August 29, 2019 - 3:39:52 PM
Last modification on : Friday, August 30, 2019 - 10:10:24 AM

File

474698_1_En_53_Chapter.pdf
Files produced by the author(s)

Licence


Distributed under a Creative Commons Attribution 4.0 International License

Identifiers

Citation

Marco Mahecha, Octavio Parra, Julio Velandia. Design of a System for Melanoma Detection Through the Processing of Clinical Images Using Artificial Neural Networks. 17th Conference on e-Business, e-Services and e-Society (I3E), Oct 2018, Kuwait City, Kuwait. pp.605-616, ⟨10.1007/978-3-030-02131-3_53⟩. ⟨hal-02274187⟩

Share

Metrics

Record views

44

Files downloads

11