Skin Lesion ABCDE Risk-Score Simulator

Not medical advice. This simulator is an educational illustration of how the ABCDE screening mnemonic combines five visual features into a simple composite score. It does not diagnose skin cancer, does not analyze real photographs, and cannot tell you whether any specific mole is dangerous. Any real lesion that concerns you, regardless of how it scores on any tool, needs an in-person examination by a dermatologist.

A: Asymmetry — Why One Half Not Matching the Other Matters

Most common benign moles are reasonably round or oval, and if you draw an imaginary line through the middle in any direction, the two halves look roughly like mirror images of each other. Melanocytic growths that have started dividing and spreading irregularly tend to lose that symmetry, producing a shape where one half genuinely does not resemble the other no matter which axis you draw the line along. That is the clinical basis for the "A" in ABCDE: not a precise measurement, but a visual gestalt that most people can learn to spot with a little practice. In this simulator, the asymmetry slider runs from 0, a nearly circular lesion whose two halves are close to identical, up to 2, where one half of the drawn shape visibly bulges and skews away from the other. The lesion-drawing code biases the radius of each point on the outline differently depending on which half of the shape it falls in, then scales that bias by the slider value, so what you see updating on canvas is a direct visual analogue of the "fold it in half mentally" test a clinician performs.

B: Border Irregularity — From Smooth Edges to Notched, Scalloped Outlines

Benign moles typically have a smooth, well-defined edge, as though drawn with a single confident pencil stroke. Lesions of concern more often show a ragged, notched or scalloped border, sometimes described clinically as having poorly defined margins that blur into the surrounding skin in places and jut out sharply in others. Dermoscopy scoring systems, including elements of the widely used 7-point checklist, formalize border assessment on a numeric scale rather than a simple present-or-absent judgment, which is the convention this simulator borrows for its 0-8 border-irregularity slider. Turning the border slider up layers increasing amounts of procedural noise onto the lesion's outline calculation, so low values produce a nearly circular edge and high values produce a visibly jagged, uneven silhouette with multiple inward and outward notches. It is worth being clear that this is an illustrative visual analogue built for teaching purposes, not a validated dermoscopic border-scoring algorithm; real dermoscopic border assessment involves examining the lesion under magnification and polarized light, well beyond what a slider can capture.

C: Color Variegation — More Distinct Colors, More Reason to Look Closer

A single, fairly uniform shade of brown is the hallmark of most common moles. Melanoma and other lesions of concern more frequently show multiple distinct colors within the same lesion, commonly described as combinations of tan, brown, black, red, white and blue-gray, reflecting uneven melanin production, inflammation, regression and, in some cases, bleeding within different parts of the growth. Counting how many genuinely distinct colors are visible, rather than subtle shading gradients within one color, is the practical version of the "C" assessment. The color-count slider here runs from 1 to 6 and directly controls how many distinct pigment blotches the canvas renderer scatters inside the lesion outline, each drawn from a small palette spanning tan through near-black and a reddish tone, clipped to stay inside the lesion shape. At the low end you see a single, fairly uniform fill; at the high end the lesion visibly mottles into several overlapping patches, giving an intuitive sense of what "color variegation" looks like rather than just a number.

D: Diameter — The Pencil-Eraser Reference, and Why It Is a Weighted Cue, Not a Cutoff

The most commonly cited version of ABCDE flags lesions larger than about 6 millimeters across, roughly the size of a standard pencil eraser, as more worth investigating, since melanomas have historically tended to be diagnosed at larger average sizes than typical benign moles. This simulator draws a dashed 6mm reference circle behind the lesion so you can see the diameter slider's 0-15mm range in genuine visual proportion, rather than as an abstract number. It is important to understand what this threshold does and does not mean clinically. Diameter is one of five weighted contributors in the ABCDE framework, not an independent pass-fail test. A meaningful fraction of melanomas, particularly ones caught early through vigilant self-examination, are diagnosed well under 6mm, and the overwhelming majority of moles larger than 6mm are entirely benign. Treating 6mm as a hard line in either direction, "it's small so it's fine" or "it's over 6mm so it's dangerous," misrepresents how the guideline is meant to be used alongside the other four letters.

E: Evolution — The Letter a Single Photograph Can Never Capture

The fifth letter, added to the original ABCD mnemonic specifically because clinicians recognized its diagnostic value, asks a fundamentally different kind of question than the first four: has this lesion changed in size, shape, color, elevation or symptoms compared with how it looked previously? Evolution cannot be assessed from a single image or a single moment in time, only from a genuine before-and-after comparison, which is exactly why serial photography, sometimes called total body photography or mole mapping, has become a standard surveillance tool in dermatology clinics for patients with many moles or a personal or family history of melanoma. Because a static rendering fundamentally cannot show change over time without misrepresenting how evolution is actually assessed, this simulator represents it honestly as a simple yes/no toggle rather than attempting a fake before-and-after animation. Flipping it to "yes" shifts the illustrative composite score upward and outlines the lesion in a warning color on canvas, a deliberately simple visual cue standing in for what, in real practice, would be a genuine comparison between two photographs taken months apart.

From Five Letters to Millions of Pixels: How ISIC-Trained Classifiers Go Further

ABCDE works because it distills genuine visual patterns into features a human can evaluate in seconds without special equipment. That same intuition, that a lesion's shape, edge, color and size carry diagnostic signal, is also the starting point for computational dermoscopy research, but modern deep-learning classifiers are not limited to five hand-picked letters. Convolutional neural networks trained on datasets like the International Skin Imaging Collaboration (ISIC) archive, a large public repository of dermoscopic images paired with clinician- or biopsy-confirmed diagnoses, learn to detect far subtler patterns: texture variations at the pixel level, gradual color-gradient transitions, fine structural details visible under dermoscopic magnification, and combinations of features that do not correspond to any single human-nameable trait. This is genuinely powerful, and on certain narrowly defined benchmark classification tasks, published research has reported some ISIC-trained models achieving accuracy statistically comparable to panels of board-certified dermatologists. That is a real, notable research result, worth stating plainly rather than either dismissing or overselling. It is equally important to state its limits plainly: benchmark accuracy on a curated, labeled image set is not the same as diagnostic reliability across the full diversity of real-world skin tones, imaging conditions, lesion types, camera equipment and clinical contexts that a dermatology clinic actually encounters day to day. Model performance can also degrade meaningfully on populations or skin tones underrepresented in the training data, a well-documented challenge across medical AI generally, and one that active ISIC-related research continues working to address through more diverse data collection. The honest summary is that these tools are best understood as promising decision-support research, not a replacement for clinical judgment.

Why ABCDE Has Real, Well-Documented Limits

No screening heuristic is perfect, and ABCDE's limitations are well studied in the dermatology literature rather than hidden. Nodular melanoma, a faster-growing and less common subtype, frequently presents as symmetric, evenly colored, and small, precisely the pattern ABCDE was designed to wave through as low-concern, which is why some clinicians advocate supplementing ABCDE with an "EFG" checklist for nodular lesions specifically, Elevated, Firm and Growing. Amelanotic melanoma, which lacks the dark pigmentation the "C" criterion depends on almost entirely, can likewise slip past a color-based assessment. And because ABCDE was validated primarily on populations and lesion types most commonly studied in the dermatology literature, its sensitivity and specificity can vary across different skin tones and lesion presentations, an active and important area of ongoing clinical research. None of this means ABCDE is not useful, sensitivity and specificity studies generally find it a genuinely helpful screening aid when used as intended, as a prompt to seek a professional opinion rather than a self-contained verdict. It does mean that treating a low ABCDE score, or a low score from any illustrative tool like this one, as reassurance to skip a dermatologist visit would be a misuse of what the heuristic was ever designed to do.

Frequently asked questions

Is the ABCDE rule a diagnosis for melanoma?

No. ABCDE is a screening heuristic designed to help patients and primary-care clinicians decide which pigmented lesions deserve a closer look from a dermatologist, not a diagnostic test. A lesion that scores high on several ABCDE letters is more worth having examined, and a lesion that scores low is not automatically safe to ignore, since a proper diagnosis requires dermoscopic or histopathological examination by a trained professional, not a checklist score.

Why does diameter use a 6mm threshold specifically?

Six millimeters is roughly the diameter of a standard pencil eraser, which made it a practical, memorable visual reference when the ABCDE mnemonic was first popularized for patient self-examination decades ago, well before consumer measuring tools were commonly at hand. It was never meant as a hard biological cutoff. Many melanomas are diagnosed under 6mm, and plenty of benign moles exceed it, so diameter functions as one weighted contributor among five, not a standalone pass-fail line.

What does the ISIC dataset actually contain, and why does it matter for AI?

The International Skin Imaging Collaboration (ISIC) archive is a large, growing public repository of dermoscopic images contributed by multiple clinical institutions, each labeled with a clinician-confirmed or biopsy-confirmed diagnosis. Having tens of thousands of consistently labeled images in one place is what let researchers train convolutional neural networks to learn subtle texture, color-gradient and border patterns statistically associated with malignancy, patterns far more detailed than the five ABCDE letters a human can practically eyeball, which is why ISIC-scale datasets became a turning point for automated dermoscopy classification research.

Have AI models really matched dermatologists at diagnosing skin cancer?

On specific, narrowly defined benchmark tasks, such as classifying a fixed set of dermoscopic images into a small number of diagnostic categories, some published studies have reported convolutional neural network classifiers achieving accuracy statistically comparable to panels of board-certified dermatologists. That is a real and important finding, but it describes performance on curated benchmark datasets under controlled conditions, not routine clinical practice across the full diversity of skin tones, imaging equipment, lesion types and real-world edge cases a clinic actually sees, so it should be read as a promising research result rather than a claim that AI has replaced dermatologist judgment.

Why is the 'E' for Evolution just a yes/no toggle in this simulator instead of an animation?

Evolution, the fifth ABCDE letter, is fundamentally a comparison over time. Clinically it means noticing that a mole has changed in size, shape, color or symptoms compared with how it looked weeks or months earlier, something only a before-and-after comparison can capture, not a single photograph or a single rendered frame. This simulator represents that honestly with a simple yes/no toggle that shifts the illustrative risk band, rather than trying to fake a visual animation of change that would misrepresent how evolution is actually assessed in practice.

Try it live

Everything above runs in your browser — open Skin Lesion ABCDE Risk-Score Simulator and change the parameters while it is running. Nothing is installed, nothing is uploaded, the whole model lives in one tab. Remember: this is an educational illustration, not a diagnostic tool, and any real lesion of concern needs an in-person dermatologist exam.

▶ Open Skin Lesion ABCDE Risk-Score Simulator simulation