Trichoscopy of Alopecia Areata: Hair Loss Feature Extraction and Computation Using Grid Line Selection and Eigenvalue

In 2020, researchers Sunyong Seo and Jinho Park developed an algorithm to assist in the self-diagnosis of scalp conditions, specifically alopecia areata, by analyzing hair loss features (HLFs) from microscopic images taken with a smart device-attached microscope. The algorithm preprocessed images to enhance contrast and reduce light reflection, then extracted HLFs such as hair count, follicle presence, and hair thickness. The study tested the algorithm on a set of 100 scalp images from individuals aged 20 to 40, achieving an average accuracy of 96.51% for hair count and 84.07% for hair follicle detection. The researchers proposed that their method could enable consistent evaluation of hair loss and help in monitoring scalp health over time. They also suggested the potential for using the collected data to analyze the causes of alopecia areata. However, the document did not provide the number of participants involved in the study, limiting the assessment of the study's strength. Additionally, the research discussed the challenges of integrating AI into portable devices for trichoscopy due to the high number of parameters required by convolutional neural network-based AI. The data from the study is available upon request, and the research was supported by the MSIT of Korea.