|
Briefing
|
- Skin Cancer Screening Algorithm – Stanford researchers trained algorithm to identify skin cancer by feeding it thousands of images, learning to determine if lesion is benign, malignant or non-neoplastic (not a tumor)
- More Accurate Than Humans – Scored with 71% accuracy compared to two dermatologists’ 65-66% scores, with separate test among 21 dermatologists showing algorithm outperforming humans
- More Training Required – Needs to be taught to distinguish lesions in people of darker skin by feeding it more images
- Smartphone App Prototype – Researchers plan on making smartphone app version of software
|
|
Accelerator
|
|
|
Sector
|
Healthcare/Health Sciences, Information Technology
|
|
Organization
|
Stanford University
|
|
Source
|
|
|
Original Publication Date
|
January 25, 2017
|
