Web-Based System for the Diagnosis of Skin Lesions Using Deep Convolutional Neural Networks and Transfer Learning Techniques
Fragment książki (Rozdział monografii pokonferencyjnej)
MNiSW
5
spoza wykazu
| Status: | |
| Autorzy: | Jaén Elmer, Omiotek Zbigniew, Pinzón Cristian |
| Dyscypliny: | |
| Aby zobaczyć szczegóły należy się zalogować. | |
| Wersja dokumentu: | Elektroniczna |
| Język: | angielski |
| Strony: | 1 - 7 |
| Scopus® Cytowania: | 0 |
| Bazy: | Scopus |
| Efekt badań statutowych | NIE |
| Materiał konferencyjny: | TAK |
| Nazwa konferencji: | 22nd International Multi-Conference for Engineering, Education and Technology |
| Skrócona nazwa konferencji: | 22nd LACCEI 2024 |
| URL serii konferencji: | LINK |
| Termin konferencji: | 17 lipca 2024 do 19 lipca 2024 |
| Miasto konferencji: | San Jose |
| Państwo konferencji: | KOSTARYKA |
| Publikacja OA: | TAK |
| Licencja: | |
| Sposób udostępnienia: | Witryna wydawcy |
| Wersja tekstu: | Ostateczna wersja opublikowana |
| Czas opublikowania: | W momencie opublikowania |
| Data opublikowania w OA: | 15 sierpnia 2024 |
| Abstrakty: | angielski |
| The diagnosis of skin lesions plays a crucial role in the early detection and treatment of various dermatological conditions. In this study, we present a web-based system for skin lesions diagnosis that utilizes deep learning models to support the identification of six different types of skin lesions (nevus, pigmented benign keratosis, seborrheic keratosis, melanoma, basal cell carcinoma and squamous cell carcinoma). The web application allows users to upload images, which are then processed by the classifier to determine the most likely skin lesion present. Six pre-trained DCNN architectures (VGG16, VGG19, DenseNet201, InceptionV3, MobileNetV2, and Xception) were used in this research. A dataset containing 2400 images was used to train the models. Data augmentation techniques were employed to increase the number of training samples. After conducting experimentation and a comprehensive evaluation, we concluded that the deep learning models provided satisfactory results in detecting the different skin lesions. Notably, the VGG16 model exhibited superior classification accuracy (86%) and fast response times, making it the most effective model among the six. The web-based system, designed with a user-friendly and easy-to-use interface serves two purposes: empowering patients to perform self-diagnosis and providing dermatologists with support for more accurate diagnoses. Our findings highlight the potential of deep learning models, particularly the VGG16 architecture, in assisting with the diagnosis of skin lesions. Our work proved that it is possible to build an efficient skin lesions diagnosis tool based on existing web technologies and machine learning methods. |
