A Hybrid Deep Learning Approach for Identifying Lyme Disease from Skin Rash Images

Paper ID: 21

Abstract

This study aims to improve the early detection of Lyme disease by classifying rashes using a novel hybrid deep learning architecture. The growing incidence of Lyme disease highlights the importance of rapid diagnosis, as traditional methods are slow and may allow bacterial growth. The proposed method utilizes three pre-trained models—ResNet50 V2, VGG19, and DenseNet201—for initial classification, followed by a hybrid model combining DenseNet201 and VGG19 to enhance detection accuracy. This approach addresses a gap in current research, which often overlooks rash analysis in favor of focusing on tick identification.

Keywords