Facial Botox Injection Point Detection Using YOLOv8 Enhanced with CBAM and BiFPN: A Multi-Perspective Deep Learning...

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Cosmetic Dermatology

Background:

• Botox injections are common in cosmetic dermatology for wrinkle reduction but are technically challenging.
• High variability exists in identifying precise Botox injection points, impacting procedure consistency and outcomes.

Purpose of the Study:

• To develop an enhanced YOLOv8 object detection framework for accurate and automated identification of Botox injection points.
• To improve the precision and consistency of Botox injection site localization in facial images.

Main Methods:

• An enhanced YOLOv8 object detection framework was developed by integrating the Convolutional Block Attention Module (CBAM) and Bidirectional Feature Pyramid Network (BiFPN).
• The system was evaluated on a novel high-resolution dataset comprising four clinically relevant facial image subsets.

Main Results:

• The proposed framework demonstrated consistent performance improvements over the baseline YOLOv8n architecture.
• Relative mean average precision (mAP@0.5) gains ranged from 1.3% to 4.2%, with notable enhancements in localizing small wrinkles.
• The integration of CBAM and BiFPN modules significantly improved the accuracy of injection point detection.

Conclusions:

• The developed AI system offers a promising solution for automated and precise Botox injection point detection.
• This technology has the potential to serve as AI-assisted clinical decision support in cosmetic dermatology, enhancing procedural accuracy and patient outcomes.