Skin Lesions Classification Into Eight Classes for ISIC 2019 Using Deep Convolutional Neural Network and Transfer Learning
importance 7/10
ai assisted
Cutaneous Melanoma Detection and Management dermatology stage: research
Problem
Eight types of skin lesions look highly similar, causing inaccurate diagnosis and missed early melanoma detection.
Why it matters
Early accurate skin-lesion classification enables dermatologists to treat patients and save lives.
What the AI did
A deep convolutional network using transfer learning from pre-trained GoogleNet classified eight skin-lesion classes on the ISIC 2019 dataset.
Results
not stated in source
Impact
Evidence supports eight-class classification on a public challenge dataset; the accuracy figure is cut off in the abstract.
Record
| Venue | IEEE Access |
|---|---|
| Published | 2020 |
| Work type | article |
| Authors | Mohamed A. A. Kassem, Khalid Mohamed Hosny, Mohamed Mahmoud Fouad |
| Citations | 407 (OpenAlex, at import time) |
| References | 47 |
| Model or method | GoogleNet (transfer learning, deep CNN) |
| Organization | not stated in source |
| Primary source | https://doi.org/10.1109/access.2020.3003890 |
| Open access | gold |
| Keywords | skin lesion classification, melanoma, transfer learning, GoogleNet, ISIC 2019, deep CNN |
Importance rating
7
significant — thousands to millions affected
Abstract states melanoma has a high mortality rate and that early accurate classification enables dermatologists to treat patients and save lives.