Classification Framework for Healthy Hairs and Alopecia Areata: A Machine Learning (ML) Approach

Area of Science:

• Dermatology
• Medical Imaging
• Machine Learning

Background:

• Alopecia areata is an autoimmune disorder causing hair loss, affecting 1 in 1000 individuals globally.
• Machine learning (ML) shows promise in dermatology for disease classification, prediction, and diagnosis.

Purpose of the Study:

• To propose and evaluate a machine learning framework for classifying healthy hair versus alopecia areata.
• To compare the performance of Support Vector Machine (SVM) and K-Nearest Neighbors (KNN) algorithms for this classification task.

Main Methods:

• Image preprocessing (enhancement, segmentation) was applied to 200 healthy hair images (Figaro1k) and 68 alopecia areata images (Dermnet).
• Features including texture, shape, and color were extracted from the images.
• SVM and KNN classifiers were trained on 70% of the data and tested on the remaining 30%, using 10-fold cross-validation.

Main Results:

• Support Vector Machine (SVM) achieved a classification accuracy of 91.4%.
• K-Nearest Neighbors (KNN) achieved a classification accuracy of 88.9%.
• A paired sample T-test indicated a statistically significant difference (p < 0.001) between SVM and KNN accuracies, with SVM performing superiorly.

Conclusions:

• The developed machine learning framework demonstrates significant potential for the accurate classification and prediction of alopecia areata.
• SVM is a highly effective method for classifying alopecia areata from hair images, offering improved diagnostic capabilities in dermatology.