Harnessing Deep Learning for Scalp and Hair Disease Classification: A Comparative Study of Convolutional Neural Network Architectures

The study evaluates the effectiveness of various convolutional neural network (CNN) architectures, including VGG16, VGG19, Inception-V3, ResNet50, and ResNet152, for classifying scalp and hair diseases using a dataset of 12,530 images. VGG16 and VGG19 outperform other models in accuracy across all disease categories, with test accuracies of 96.81% and 96.73%, respectively, demonstrating their robustness for this application. Inception-V3 also performs well, particularly in complex cases like psoriasis, with a test accuracy of 95.13%. ResNet models underperform due to their complexity and the limited dataset size. The study suggests that expanding datasets and exploring ensemble models could further enhance diagnostic accuracy and reliability, aiming to improve automated diagnostic methods and clinical decision-making.