This simulator turns the classic ABCDE dermatology screening mnemonic, Asymmetry, Border irregularity, Color variegation, Diameter and Evolution, into five live controls. Each one procedurally reshapes a drawn lesion on canvas and feeds an illustrative composite risk band, the same kind of simple hand-craftable features that early skin-lesion classifiers used before deep learning on datasets like the ISIC archive took over.
Asymmetry blends two half-shapes so they diverge more as the slider rises. Border irregularity adds jagged radial noise to the lesion outline. Color count draws that many distinct pigment blotches inside the shape. Diameter scales the lesion against a 6mm pencil-eraser reference circle, the classic clinical size cue. Evolution is a simple yes/no flag, since real evolution can only be judged by comparing photos over time, not from a single static rendering.
Asymmetry slider, border-irregularity slider, color-count slider, diameter slider, and an evolution yes/no selector.
Did you know that ABCDE was designed as a quick visual screening heuristic for patients and primary-care clinicians, not a diagnostic algorithm, and that some melanomas, particularly nodular melanoma, fail to fit the "big, asymmetric, multicolored" pattern at all? Large annotated dermoscopy datasets like ISIC let deep-learning models learn far richer texture, color-gradient and border features than the five ABCDE letters can capture, which is part of why CNN classifiers have matched dermatologist-level accuracy on some narrow benchmark tasks.
This simulator turns the classic ABCDE dermatology screening mnemonic, Asymmetry, Border irregularity, Color variegation, Diameter and Evolution, into five live controls that reshape a procedurally-drawn lesion on canvas and feed a clearly-labeled illustrative composite risk band, mirroring the kind of simple feature engineering that predates deep-learning classifiers trained on datasets like the ISIC dermoscopy archive.
This simulator turns the classic ABCDE dermatology screening mnemonic, Asymmetry, Border irregularity, Color variegation, Diameter and Evolution, into five live controls that reshape a procedurally-drawn lesion on canvas and feed a clearly-labeled illustrative composite risk band, mirroring the kind of simple feature engineering that predates deep-learning classifiers trained on datasets like the ISIC dermoscopy archive.
Drag the asymmetry, border, color and diameter sliders to watch the lesion shape, blotching and size change in real time against a 6mm pencil-eraser reference circle, then flip the evolution toggle to see how a recent change shifts the illustrative risk band between Low, Moderate and High.
Did you know that ABCDE was designed as a quick visual screening heuristic for patients and primary-care clinicians, not a diagnostic algorithm, and that some melanomas, particularly nodular melanoma, fail to fit the "big, asymmetric, multicolored" pattern at all? Large annotated dermoscopy datasets like ISIC let deep-learning models learn far richer texture, color-gradient and border features than the five ABCDE letters can capture, which is part of why CNN classifiers have matched dermatologist-level accuracy on some narrow benchmark tasks.