Are you represented? Subjective vs objective skin color determination for healthcare and research purposes.

This study evaluates the limitations of existing subjective skin color classification methods against objective spectrophotometer measurements and proposes a new, cost-effective 10-category system (Nottingham Skin Categories) to improve accuracy and representativeness in healthcare and research.

The Gist

Imagine you are trying to bake a cake for a huge party with people from every corner of the world. To make sure the cake tastes right for everyone, you need to know exactly how sweet or sour their palates are.

In the world of medicine and technology, skin color is that "palate." It determines how light interacts with our bodies, which is crucial for things like laser treatments, skin cancer detection, and even the smartwatches on our wrists that measure our heart rate.

This paper is like a group of scientists saying, "Hey, our current way of guessing people's skin color is broken, and it's causing some people to get the wrong 'recipe' for their health."

Here is the story of what they found and what they propose, explained simply:

1. The Old Way: The "Guessing Game" (Fitzpatrick Scale)

For decades, doctors and researchers have used a questionnaire called the Fitzpatrick Skin Type Scale (FST). It's like asking someone, "Do you burn easily? Do you tan?" to guess their skin tone.

• The Problem: It's like trying to guess the exact temperature of a room by asking someone if they feel "a little hot" or "very hot." It's too vague.
• The Result: The paper found that this method is terrible for people with darker skin. It often misclassifies them, grouping very different skin tones into the same bucket. This is dangerous because medical devices (like pulse oximeters used during the pandemic) were calibrated mostly on "light" guesses, leading to inaccurate readings for darker-skinned patients.

2. The "Swatch Book" Method (Pantone)

Then there is the Pantone SkinTone Guide. Imagine a giant book of 110+ colored cards. You hold them up to a person's arm and say, "Which one matches?"

• The Good News: The scientists found this is actually much better than the questionnaire. It correlates very well with the "real" science.
• The Bad News: It's overwhelming. Trying to pick the exact match from 110+ cards is like trying to find a specific needle in a haystack of 100 similar needles. People get confused, and it takes too long.

3. The "Digital Filter" Method (Monk Scale)

Recently, a new scale called the Monk Skin Tone Scale (MST) was created, mostly for computers and AI to recognize faces better. It has 10 levels.

• The Problem: The scientists tested this against real human skin and found it was missing the middle ground. It's like a photo filter that only has "Super Pale," "Medium," and "Super Dark," but skips all the beautiful, complex shades in between. It doesn't represent the actual diversity of human skin found in hospitals and labs.

4. The "Gold Standard" (Spectrophotometers)

The only way to get it 100% right is using a Spectrophotometer. Think of this as a high-tech "skin scanner" that measures the exact light bouncing off the skin. It gives a precise number (called an ITA value).

• The Catch: These machines are expensive (costing about $10,000) and bulky. You can't carry one into a doctor's office in a small clinic or use it for a quick check-up.

The Solution: The "Nottingham Skin Categories" (NSC)

The authors realized we need a middle ground: something cheap and easy like the swatch book, but accurate like the high-tech scanner.

They propose a new system called Nottingham Skin Categories (NSC).

• The Analogy: Imagine instead of 110 confusing cards or a vague questionnaire, you have a simple deck of 9 cards.
• How it works: These 9 cards act as "fence posts" that divide human skin into 10 clear, equal-sized groups.
  • If your skin is lighter than Card 1, you are in Group 1.
  • If your skin is between Card 1 and Card 2, you are in Group 2.
  • And so on, all the way to Group 10 (the darkest skin).
• Why it's great:
  1. Simple: No more staring at 100 cards. Just find the two cards your skin fits between.
  2. Accurate: The scientists mapped these cards directly to the expensive "Gold Standard" scanner numbers. So, if a doctor says "You are Group 5," a researcher knows exactly what that means scientifically.
  3. Fair: It ensures that dark skin is represented properly, not just lumped into one big "dark" bucket.

Why Should You Care?

If we don't fix how we measure skin color, medical technology will keep failing certain groups of people.

• Smartwatches might not count your steps or heart rate correctly.
• Lasers used for hair removal or skin treatments could burn darker skin because the doctor guessed the wrong setting.
• AI used to diagnose skin cancer might miss tumors on darker skin because it was only "trained" on light skin.

The Bottom Line:
This paper suggests we stop guessing and stop using confusing charts. Instead, let's use a simple, 9-card system that is cheap, easy to use, and scientifically accurate. It's like upgrading from a blurry, black-and-white map to a clear, color-coded GPS that works for everyone, no matter where they are from.

Technical Summary

1. Problem Statement

Optical medical technologies (e.g., pulse oximeters, lasers, OCT, wearables) rely on the interaction of light with human tissue. However, the development and testing of these devices have historically been biased toward fair skin.

• Consequences: This bias leads to inaccurate readings and poorer health outcomes for individuals with darker skin tones (e.g., pulse oximeters overestimating oxygen saturation in dark-skinned patients).
• Current Limitations:
  • Subjective Methods: The Fitzpatrick Skin Type (FST) scale is the industry standard but relies on self-reported UV reaction, making it highly inaccurate for dark skin and prone to misclassification.
  • Objective Methods: Spectrophotometers provide accurate, quantitative data (Individual Typology Angle - ITA) but are expensive (~$10,000–£10,000), require specialized training, and are impractical for widespread clinical or large-scale research use.
  • Emerging Scales: The Pantone SkinTone Guide (PST) and Monk Skin Tone Scale (MST) have been proposed as alternatives, but their efficacy and representativeness against objective data remain unverified in a clinical context.

2. Methodology

The study involved a comparative analysis of three subjective skin color assessment tools against objective spectrophotometric measurements.

• Cohort: 87 volunteers (56% female, 44% male; ages 18–71). The study specifically recruited dark-skinned volunteers to ensure diversity, though the "very light" ITA category was not represented in this specific cohort.
• Objective Measurement (Ground Truth):
  • Instrument: Konica Minolta CM700d spectrophotometer.
  • Parameters: Measured spectral reflectance (400–700 nm) under D65 illumination.
  • Metric: Calculated the Individual Typology Angle (ITA) using CIELAB values ($L^*$ and $b^*$) via the formula: $ITA^\circ = \arctan((L^* - 50)/b^*) \times 180/\pi$.
  • Classification: Volunteers were categorized into 6 objective ITA groups: Very Light, Light, Intermediate, Tan, Brown, and Dark Brown.
• Subjective Assessments:
  1. Fitzpatrick Skin Type (FST): Volunteers completed a questionnaire regarding genetic traits and sun reaction.
  2. Pantone SkinTone Guide (PST): Volunteers matched their skin to one of 110+ swatches under D65 lighting. The chosen swatch was then measured by the spectrophotometer to determine its ITA.
  3. Monk Skin Tone Scale (MST): The 10-level MST was analyzed by converting its published CIELAB values to ITA and comparing them against the volunteer data distribution.
• Data Analysis: Correlation coefficients (Pearson and Spearman) were calculated between subjective methods and objective ITA values. Data from this study was also collated with large-scale datasets from Del Bino & Bernerd (n=3,500) and Lipnick et al. (n=789) to assess global representativeness.

3. Key Results

• FST Performance:
  • Showed a poor negative correlation with objective ITA ($\rho = -0.65$).
  • Highly subjective: Volunteers with FST I–III often fell into the "Intermediate" ITA category, while FST IV–V covered a massive range from "Light" to "Dark Brown."
  • Confirmed that FST is biased toward light skin and fails to accurately categorize dark skin tones.
• PST Performance:
  • Showed a strong positive correlation with objective ITA ($\rho = 0.90$).
  • Bias: Volunteers tended to select swatches that were lighter than their objectively measured skin tone (e.g., FST I volunteers selected PST swatches corresponding to "Very Light" ITA, despite having "Intermediate" objective ITA).
  • Usability Issue: The 110+ swatches were described by volunteers as "overwhelming" and excessive, leading to difficulty in selection.
• MST Performance:
  • Poor Representativeness: When mapped to objective ITA data, the MST failed to cover the full spectrum of human skin.
  • MST levels 1–4 all fell into the "Very Light" category (representing <0.03% of the global volunteer pool).
  • MST Level 5 fell outside the range of all measured volunteer data.
  • "Intermediate" skin tones were not represented by any MST level.
  • Levels 9 and 10 did not correspond to measured skin colors in the combined dataset.
  • Conclusion: The MST is optimized for computer vision/digital representation, not clinical dermatology, and suffers from printing variability and lighting sensitivity.

4. Key Contributions: The Nottingham Skin Categories (NSC)

Based on the limitations of existing tools and the strong correlation between PST and objective ITA, the authors propose a new standardized framework:

• Concept: A simplified set of 9 color swatches (Nottingham Skin Categories - NSC) designed to divide the skin color spectrum into 10 evenly distributed groups.
• Structure:
  • Groups are separated by 15° ITA intervals.
  • Group 1: ITA > 55° (Lightest).
  • Group 10: ITA < -65° (Darkest).
  • Swatches: 9 swatches (A–I) act as boundaries between the 10 groups.
• Advantages:
  • Cost-Effective: Drastically cheaper than spectrophotometers (~$150 vs. ~$10,000).
  • Usability: Reduces the "overwhelming" nature of the 110+ Pantone swatches to a manageable 9.
  • Standardization: Directly maps to objective ITA values, ensuring studies are representative of global populations (including dark skin).
  • Application: Suitable for primary care, dermatology, photodynamic therapy dosing, and optical device testing.

5. Significance and Future Impact

• Addressing Health Disparities: The NSC scale offers a practical, low-cost solution to the "fair skin bias" in medical device testing, directly addressing recent mandates from the UK NHS and FDA regarding device suitability for dark skin.
• Research Reproducibility: By providing a standardized, objective-mapped classification, the NSC will improve comparability between different research studies and clinical trials.
• Clinical Utility: Enables accurate skin categorization for dosage assessment in topical medicine and laser treatments without requiring expensive equipment.
• Call to Action: The authors emphasize that while objective spectrophotometry is the gold standard, the NSC provides a necessary bridge for widespread implementation. Future work is needed to validate these categories against global melanin content data and ensure inclusivity across all geographic populations.

Conclusion: The paper demonstrates that while the FST is obsolete for accurate skin classification and the MST is unsuitable for clinical dermatology, a simplified, objective-mapped swatch system (NSC) can effectively standardize skin color determination, ensuring equity in healthcare and research.