Computer Algorithms Show Potential for Improving Dermatologists’ Accuracy to Diagnose Cutaneous Melanoma; Results of ISIC 2017.

Computer Algorithms Show Potential for Improving Dermatologists’ Accuracy to Diagnose Cutaneous Melanoma; Results of ISIC 2017.

Publication date: Jul 12, 2019

Computer vision has promise in image-based cutaneous melanoma diagnosis but clinical utility is uncertain.

To determine if computer algorithms from an international melanoma detection challenge can improve dermatologist melanoma diagnostic accuracy.

Cross-sectional study using 150 dermoscopy images (50 melanomas, 50 nevi, 50 seborrheic keratoses) from the test dataset of a melanoma detection challenge, along with algorithm results from twenty-three teams. Eight dermatologists and nine dermatology residents classified dermoscopic lesion images in an online reader study and provided their confidence level.

The top-ranked computer algorithm had a ROC area of 0.87, which was higher than the dermatologists (0.74) and the residents (0.66) (p

Marchetti, M.A., Liopyris, K., Dusza, S.W., Codella, N.C.F., Gutman, D.A., Helba, B., Kalloo, A., Halpern, A.C., and (ISIC), International Skin. Imaging. Collaboration. Computer Algorithms Show Potential for Improving Dermatologists’ Accuracy to Diagnose Cutaneous Melanoma; Results of ISIC 2017. 23371. 2019 J Am Acad Dermatol.

Concepts Keywords
Algorithm Dermatologist melanoma
Computer Vision Cutaneous Melanoma
Confidence Level Melanomas
Dermatologist Cutaneous conditions
Dermatology Melanoma
Dermoscopy Organ systems
Lesion Dermatology
Melanoma Dermatoscopy
Melanomas Seborrheic keratosis
Nevi Nevus
ROC Dysplastic nevus
Fernando Alfageme
Computer algorithm
Computer algorithms

Semantics

Type Source Name
disease DOID skin cancer
disease MESH skin cancer
drug DRUGBANK Saquinavir
gene UNIPROT THOP1
disease MESH seborrheic keratoses
disease MESH nevi
pathway BSID Melanoma
disease DOID melanoma
disease MESH melanoma
disease MESH diagnosis
disease DOID Cutaneous Melanoma

Similar

Original Article

Leave a Comment

Your email address will not be published. Required fields are marked *