Artificial intelligence (AI) in melanocytic lesions diagnosis: a systematic review of methods incorporating ABCD(E)-derived features.

What was studied

This review looked at AI systems that diagnose melanoma. It focused on systems that use the ABCD(E) rule, a screening pattern. Studies came from PubMed, LILACS, Latindex and the Cochrane register. The authors assessed 2,735 records. 21 studies were kept, with over 90,000 images or cases.

What they found

AI systems, mainly CNNs and DCNNs, showed good diagnostic performance. They often equaled or beat human experts. One result was an area under the curve of 90.3 for early melanoma differentiation. The authors saw patient selection bias. It limited how far the findings apply.

What it means for patients

The authors say AI shows significant and evolving ability in skin lesion diagnosis. They say it increasingly uses ABCD(E) criteria. They say it offers more consistent diagnoses.

Why it is not certain yet

Patient selection bias limited how well the findings apply. Challenges persist. These include limited data and how well models work in new settings. They also include how easy models are to interpret. A review can only be as good as the studies in it.

Source

Artificial intelligence (AI) in melanocytic lesions diagnosis: a systematic review of methods incorporating ABCD(E)-derived features.

We restate facts from the source record in our own words. The original wording belongs to its publisher.