Resumen

Objectives: To assess the efficacy and safety of RFMN combined with exosome therapy for atrophic acne scars in skin of color (Pakistani/ South Asian), using AI-based 3D imaging alongside validated clinical scales. Introduction: Acne scar management in South Asian patients (Fitzpatrick IV–V) is limited and remains challenging by the risk of post-inflammatory hyperpigmentation (PIH). Radiofrequency microneedling (RFMN) achieves controlled dermal remodeling, while exosomes enhance regenerative repair. Artificial intelligence (AI)-driven 3D imaging now enables objective, reproducible evaluation of scar outcomes beyond conventional subjective grading. Materials / method: Twenty patients with moderate atrophic acne scars (Goodman & Baron grades II–III, Fitzpatrick IV–V) underwent three sessions of RFMN (depth 1.5–3.5 mm; power 15–25 W) followed by topical exosome application. Outcomes were assessed at baseline and 12 weeks post-treatment using AI-based 3D skin analysis (scar depth, volume, texture, pigmentation, composite skin quality index), Goodman & Baron score, Global Aesthetic Improvement Scale (GAIS), and patient satisfaction. Safety endpoints included PIH, erythema, and downtime. Results: AI analysis showed a 35% reduction in mean scar depth, 32% reduction in scar volume, 28% improvement in texture, and 22% decrease in pigmentation variance. Composite skin quality improved by 40% over baseline. Goodman & Baron scores improved by ≥1 grade in 80% of patients, and 72% were rated “much improved” or higher on GAIS. PIH incidence was <10%, with downtime averaging 2–3 days Conclusion: AI-driven imaging confirms that combining RFMN with exosomes provides safe, quantifiable, and reproducible improvement in acne scars in South Asian skin, introducing a novel, data-driven paradigm for ethnic dermatology. Innovation Statement: First integration of AI-based 3D imaging with RFMN and exosome therapy, providing objective evidence of scar remodeling in South Asian skin for better safety and efficacy of the treatment protocol.