Exploring convolutional neural networks with transfer learning for diagnosing Lyme disease from skin lesion images
The study tested twenty-three Convolutional Neural Network (CNN) architectures to diagnose Lyme disease using skin lesion images, specifically erythema migrans (EM). By using transfer learning from ImageNet and pretraining on a skin lesion dataset, the best model achieved 84.42% accuracy with ResNet50. This research confirms that CNNs can be used in mobile apps for initial Lyme disease self-assessment and referral to dermatologists.