The Algorithm Era

The interpretation problem that followed the adoption of dermoscopy made one thing clear: the technique was harder than it looked. Simply giving clinicians a magnified view of skin lesions was not enough — they needed structured, reproducible methods for deciding what the patterns meant.

The 1990s became the decade of diagnostic algorithms, each designed to impose discipline on a technique that could easily mislead the untrained eye.

The ABCD rule

Stolz and colleagues at the University of Munich published the ABCD rule of dermatoscopy in 1994, a scoring system based on Asymmetry, Border, Colour, and Differential structures. It was deliberately designed to be learnable by non-experts. In a prospective study of 172 melanocytic lesions, the ABCD rule achieved a diagnostic accuracy for melanoma of 80%, compared with 64.4% by naked-eye examination alone.

A growing toolkit

Other algorithms followed rapidly: the Menzies method from Australia in 1996, the seven-point checklist from Argenziano and colleagues in Italy in 1998, and Soyer's three-point checklist in 2004. Each approached the same problem from a slightly different angle, but all shared a common goal: transforming dermoscopy from an art that required years of pattern-recognition experience into a procedure that could be taught, standardised, and safely deployed beyond specialist centres.

A landmark meta-analysis by Kittler and colleagues, reviewing studies from 1987 to 2000, confirmed that dermoscopy significantly improved diagnostic accuracy for melanoma in trained hands, with an improvement of approximately 49% in diagnostic odds compared with naked-eye examination. The message was clear: the instrument alone was not enough. Education was everything.