Development and Validation of a Total Body Photo‐Numeric Freckling Density Scale

Dear Editor,

Melanoma incidence and mortality continues to increase posing substantial burdens on individuals and health systems. While current evidence does not support population level screening, targeting screening to those at highest risk may prove advantageous [1]. Risk stratification often relies on assessment of phenotypic characteristics, including freckling density [2]. While freckling density has a strong association with other pigmentation characteristics, it is largely independent of both naevus count and skin colour, indicating the importance of maintaining freckling density in risk models [2]. Classic freckles, known as ephelides, are small, sharply defined macular lesions, measuring 1−2 mm, and commonly found on the face, neck, chest and arms. They vary in colour from red to light brown and can have round, oval or irregular shapes. They first appear during childhood after sun exposure, and mainly in fair‐skinned and/or red‐haired individuals. Ephelides usually increase during adolescence and often fade with age and in winter. Freckling density is largely genetically determined and associated with MC1R polymorphisms. Freckling density is generally recorded as either present or absent, or by level of severity (none/few/some/many). Some studies employ a standardised pictorial chart [3] or photo‐numeric scales [4], with the majority of these scales focused on facial freckling [4]. Inter‐observer agreement ranges from fair to substantial, depending on the number of classes evaluated and the experience of the rater [5, 6].

The objective of this study was to develop a photo‐numeric scale to reliably assess freckling from 3D total body photography. Our study focused on three common body sites of freckling: the face, dorsal aspects of the arms and upper trunk.

Study participants were selected from two completed studies previously conducted at the Princess Alexandra Hospital, Brisbane, Australia. The first study, ‘Mind your Moles,’ recruited adults from a general‐risk population living in Southeast Queensland, who had at least one naevus [7]. The second study, ‘Health Outcomes Program’ targeted recruitment of high‐risk adults, who had either ≥1 melanomas diagnosed before age 40, or ≥2 melanomas diagnosed before age 65, or a strong family history of melanoma, and/or known pathogenic variations, and/or dysplastic naevus syndrome [8]. In both studies, participants had undergone 3D total body photography using the VECTRA WB360 whole‐body scanner (Canfield Scientific Inc., Parsippany, NJ, USA). Participants were excluded for the current validation study if they had a condition which prohibited assessment of freckling (e.g. albinism), or were over 40 years of age, as age‐related sun damage would prohibit accurate assessment of freckling. Mind Your Moles was approved by the following Human Research Ethics Committees (HRECs):

Metro South Health (HREC/16/QPAH/125), The University of Queensland (2016000554), Queensland University of Technology (1600000515), and QIMR Berghofer (P2271). The Health Outcomes Program was approved by Metro South Health (HREC/17/QPAH/816) and The University of Queensland (2018000074).

To develop the freckling density scale, an iterative process was used to select image tiles best representing the three levels of freckling severity: none/mild, moderate and severe (Figure 1). The assessment of severity considered density (number of freckles), uniformity (size variability and asymmetry), pigmentation (variability in pigmentation) and background (presence of other lesions, e.g., naevi and solar lentigines). Written instructions were provided alongside the scale (Supplementary 1). To assess the reliability of the scale, four individuals (one junior doctor [SK], two medical students [MS, CF], and one researcher [BBS]) independently rated the faces, shoulders and arms for 20% of the images included in the study. Inter‐rater reliability using the photo‐numeric scale was calculated using weighted Cohen's Kappa. Following the same methods, a medical student (MS) evaluated 100% of the images and categorised the freckling density of the face, shoulders, and arms as none/mild, moderate or severe. If a participant was wearing makeup, or had a pigmentation disorder, the data were recorded as missing. Total body images were evaluated in full colour and using the UV filter (Figure 1).

FIGURE 1.

Photo‐numeric scale for rating freckling using total body photography. Images are presented in full colour and using the UV filter.

A total of 92 participants ≤40 years of age were selected for this study, including n = 29 (32%) from the general‐risk population, and n = 63 (68%) from the high‐risk population. The median age was 33 years (20–40 years), 24% were male (n = 22) and majority were of European ancestry (n = 83, 90%). A subset of 17 (20%) participant image sets was included in the reliability study of the freckling density scale. For all body sites, across all raters, average weighted kappa showed good agreement (0.73, range 0.62–0.80). The results per body‐site were moderate for face (0.60, range 0.36–0.77), good for both the shoulders (0.74, range 0.62–0.78) and arms (0.80, range 0.68–0.84).

The majority of participants had no/mild freckling on the face (n = 54, 59%), shoulders (n = 60, 65%) and arms (n = 68, 74%). Full results are reported in Table 1. While no statistically significant differences were observed between populations, a higher proportion of high‐risk participants had severe freckling on their shoulders (13%) and arms (13%), compared with general risk population (3% shoulders, and 7% arms).

TABLE 1.

Freckling density by body site and population.

	High risk (n = 63)	General population (n = 29)	Total (n = 92)	p value
Face				0.792
None/Mild	36 (57.1%)	18 (62.1%)	54 (58.7%)
Moderate	17 (27.0%)	6 (20.7%)	23 (25.0%)
Severe	8 (12.7%)	3 (10.3%)	11 (11.9%)
Shoulders				0.248
None/Mild	38 (60.3%)	22 (75.9%)	60 (65.2%)
Moderate	17 (27.0%)	6 (20.7%)	23 (25.0%)
Severe	8 (12.7%)	1 (3.4%)	9 (9.8%)
Arms				0.687
None/Mild	46 (73.0%)	22 (75.9%)	68 (73.9%)
Moderate	9 (14.3%)	5 (17.2%)	14 (15.2%)
Severe	8 (12.7%)	2 (6.9%)	10 (10.9%)

We developed and tested the inter‐rater reliability of a photo‐numeric scale for evaluating freckling density from total body photography. Compared to previous scales, the inter‐rated reliability of this scale was good [4]. Using this scale on the face, shoulders and arms, we found most participants had none/mild freckling density (59%, 65% and 74%, respectively). Similar results were found in a previous study from Queensland, in which none/mild freckling density on the face was self‐reported by 78% of participants [9]. The lack of significant differences between the two risk cohorts was likely related to the small sample size. A previous study compared manual naevi measurements with automated techniques from total body photography and reported a higher accuracy for manual methods. However, the observed automated 3D measurements served as a reliable substitute, offering a reproducible and time‐saving option for a clinical setting [10].

Study limitations include the restriction of age (≤40 years), limiting the applicability to older individuals. Furthermore, the cohort consisted of mostly individuals with European ancestry, limiting generalizability to other populations [11]. Last, the small sample size limited the evaluation of freckling differences between average‐ and high‐risk populations.

We foresee the utility of this scale extending to image labelling for future machine learning methods in dermatology. In such scenarios, algorithms could use freckling assessment on total body images in combination with other phenotype characteristics for automated melanoma risk stratification [12].

Conflicts of Interest

HPS is a shareholder of MoleMap NZ Limited and e‐derm consult GmbH and undertakes regular teledermatological reporting for both companies. HPS is a Medical Consultant for Canfield Scientific Inc., MoleMap Australia Pty Ltd., Blaze Bioscience Inc., Revenio Research Oy and a Medical Advisor for First Derm.

Supporting information

Supporting Information

Data Availability Statement

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.