Classification of Botulinum Toxin Dosage for Upper Facial Wrinkles Using Inception-V3 Based on Grey Level Co-Occurrence Matrix Feature

• Published in: 2024 8th International Conference on Information Technology, Information Systems and Electrical Engineering (ICITISEE) pp. 99 - 104
• Main Authors: Wahyuni, Ayutri, Zainuddin, Zahir, Nurtanio, Ingrid
• Format: Conference Proceeding
• Language: English
• Published: IEEE 29-08-2024
• Subjects: Accuracy; Analytical models; Botulinum Toxin; Feature extraction; GLCM; Inception-V3; Information systems; Information technology; Measurement; Medical treatment; Skin; Training; Visualization; Wrinkle
• DOI: 10.1109/ICITISEE63424.2024.10729993

The human skin consists of several ectodermal layers, with the outermost layer acting as a protective barrier against environmental factors. While BOTOX injections effectively reduce wrinkles and improve skin texture, there are challenges in visually detecting wrinkles. Wrinkles often exhibit variations in depth and pattern that are difficult to assess through visual observation alone. Therefore, this study focuses on addressing variations in skin imagery, such as rough texture and folds, which can facilitate more accurate BOTOX dosage determination by medical practitioners. This study integration a multi-input approach by combining the Gray Level Co-occurrence Matrix (GLCM) for skin texture analysis and facial image processing using the Inception-V3 model. The integration of GLCM with Inception-V3 achieved a classification accuracy of 94.57%, showing a significant improvement over the Inception-V3 model without GLCM, which only achieved an accuracy of 83.33%. This approach shows great potential in improving precision in clinical applications and optimizing dosage adjustment in aesthetic therapy, marking an important advance in this field.