With AI Skin and Scalp Analysis
Make In-Store Consultations More Accurate
How Is AI Skin Analysis Accuracy Validated?
CIn skin analysis, it is important to evaluate not only a single result but also how consistently and comparably results are maintained across different skin tones and repeated-measurement conditions.
The ChoiceDx AI skin analysis solution draws on the CTK (ChoiceTech Korea) Derma AI study to present results covering Fitzpatrick I–VI skin types, standardized capture and calibration, pore, pigmentation-spot, wrinkle and skin-tone analysis, expert-assessment comparisons, and repeated measurements.
Study Design Including Diverse Skin Tones
The study quantified key visible facial skin characteristics and compared each analysis result with expert visual assessments.
265
Study Participants
133
Training and Validation Data
I–VI
Fitzpatrick Phototypes
3×
Repeated Captures
Analysis Process from Capture to Validation
The study quantified key visible facial skin characteristics and compared each analysis result with expert visual assessments.
Step 1
Standardized Skin Capture
Lighting and camera conditions are kept consistent to reduce variation in input images.
Step 2
Capture Calibration
Image differences that may arise from skin tone and capture conditions are calibrated. The paper describes this process as Capture Calibration.
Step 3
AI Quantitative Analysis
Pores, spots, wrinkles, and skin tone are analyzed within defined regions of interest.
Step 4
Result Validation
Results are compared with expert assessments, and repeated captures are used to evaluate stability.
“With skin-tone calibration, CTK Derma AI showed stronger correlations with expert assessments than the uncalibrated Control across pores, spots, wrinkles, and skin tone.”
Shu Li, Labina Shrestha, Mirjalol Tuychiev, Tae Yong Jung, Ryan Wonsuk Choi, Kukizo Miyamoto
From Research to Real-World AI Skin Analysis
ChoiceDx connects the capture and analysis principles examined in the research with the user experience of real-world skin analyzers and AI skin analysis solutions.
The results can support non-medical beauty analysis applications such as professional consultations, repeated measurements and change comparisons, and personalized product information.
Professional Skin Analysis
Data-driven consultations using professional skin analyzers and analysis software
Repeated Measurements and Change Comparison
Comparing changes in skin condition using repeated measurements under consistent conditions
Personalized Product Recommendations
An experience that connects skin analysis results with consultation and personalized product information
Frequently Asked Questions
Q. How can AI skin analysis accuracy be evaluated?
It is important to evaluate expert-assessment comparisons, performance across diverse skin tones, standardized capture conditions, and repeatability together.
Q. What skin parameters can an AI skin analyzer evaluate?
The available parameters vary by product and configuration. In this study, pores, pigmented spots, wrinkles, and skin tone were evaluated as key parameters.
Q. Can skin tone affect AI skin analysis results?
In image-based analysis, skin tone and capture conditions can affect input images, making standardized capture and appropriate calibration important validation factors.
Q. Can skin analysis results be used for consultations and product recommendations?
ChoiceDx is designed to connect skin analysis results with customer consultations, repeated-measurement history, and personalized product information in non-medical beauty analysis settings.
Q. Are ChoiceDx analysis systems used in real retail locations?
As of August 7, 2026, 115 systems had been installed. Including installations scheduled for the same month, the total reaches 130 systems. Skin, scalp, and hair analysis systems are operating in retail and beauty environments.
