Acne Classification and Detection using ML and Quantum Model

One of the most common dermatological disorders is acne, and early and precise subtype identification is essential to optimal treatment planning. A Quantum- Assisted Acne Subtype Classification Framework that combines classical preprocessing and quantum machine learning is presented in this research. Acne lesions are localized using YOLO-labeled bounding boxes, which guarantee accurate region-of-interest extraction. For every cropped patch, dermatologically significant features such as redness intensity, lesion density, color variation, and relative lesion size are calculated to create a compact feature vector. Lesions are first categorized into Rosacea, Folliculitis, Hormonal Acne, and PCOS-related Acne using a rule-based pseudo-labeling approach. For subtype categorization, three quantum models are created and contrasted: Quantum Neural Network (QNN), Variational Quantum Circuit (VQC), and Quantum Convolutional Neural Network (QCNN). According to experimental results, the QCNN demonstrated the best accuracy and capacity for generalization, confirming the potential of quantum-enhanced dermatological analysis. The conceptual results show that scalable, interpretable, and clinically aligned hybrid quantum– classical acne classification is feasible, despite feature sparsity limiting the convergence of a hybrid HDBSCAN–Quantum pipeline.