Related Experiment Video
Updated: Jun 18, 2026
11:07
Visualizing and Quantifying Pharmaceutical Compounds within Skin using Coherent Raman Scattering Imaging
Published on: November 24, 2021
2.9K
Automated Facial Acne Lesion Detecting and Counting Algorithm for Acne Severity Evaluation and Its Utility in
Dong Hyo Kim1,2, Sukkyu Sun3, Soo Ick Cho1
1Department of Dermatology, Seoul National University College of Medicine, Seoul, South Korea.
American Journal of Clinical Dermatology
|May 9, 2023
Summary
An automated algorithm accurately detects and counts facial acne lesions by type, improving clinical evaluation of acne severity. This tool assists readers in more precise lesion assessment.
Area of Science:
- Dermatology
- Medical Imaging
- Artificial Intelligence
Background:
- Facial acne severity assessment traditionally relies on lesion counting, which is challenging to implement.
- Accurate lesion quantification is crucial for effective acne treatment monitoring.
Purpose of the Study:
- To develop and validate an automated algorithm for detecting and counting facial acne lesions by type.
- To assess the algorithm's clinical utility as an assistive tool in a reader study.
Main Methods:
- A convolutional neural network (CNN) was trained and validated on 20,699 manually labeled acne lesions from 1213 facial images.
- Algorithms were developed for both five-class lesion classification and binary classification (noninflammatory vs. inflammatory).
- Reader tests were conducted to compare lesion counting accuracy with and without the algorithm.
Main Results:
- The binary classification algorithm achieved a mean average precision of 28.48.
- High Pearson's correlation coefficients were observed between algorithm and ground-truth lesion counts (0.72 for noninflammatory, 0.90 for inflammatory).
- Readers using the algorithm demonstrated significantly improved accuracy in detecting and counting lesions.
Conclusions:
- The developed algorithm shows clinically applicable performance in detecting and classifying facial acne lesions.
- The algorithm serves as a valuable assistance tool for enhancing the accuracy of acne severity evaluations.
