Classification of Dermoscopy Skin Lesion Color-Images Using Fractal-Deep Learning Features
importance 7/10
ai assisted
Cutaneous Melanoma Detection and Management dermatology stage: research
Problem
Detecting skin diseases and skin cancer from dermoscopy color images to aid dermatologists in diagnosis.
Why it matters
Skin cancer detection is described as a priority task worldwide due to increasing skin cancer rates.
What the AI did
Combined 1D fractal texture-signature features with 100 transfer-learning features from DenseNet-201, classified by an ensemble of KNN and two SVM types via plurality vote.
Results
Average accuracy of 97.35%
Impact
The evidence supports a computer-aided diagnosis tool intended to help dermatologists detect skin lesions.
Record
| Venue | Applied Sciences |
|---|---|
| Published | 2020 |
| Work type | article |
| Authors | Edgar Omar Molina-Molina, Selene Solorza-Calderón, Josué Álvarez-Borrego |
| Citations | 36 (OpenAlex, at import time) |
| References | 67 |
| Model or method | DenseNet-201 (transfer learning) with KNN/SVM ensemble |
| Organization | not stated in source |
| Primary source | https://doi.org/10.3390/app10175954 |
| Open access | gold |
| Keywords | dermoscopy, skin cancer, deep learning, fractal features, DenseNet-201, computer-aided diagnosis |
Importance rating
7
significant — thousands to millions affected
Abstract states skin disease detection is a priority task worldwide due to increasing skin cancer; computer-aided diagnosis aids dermatologists globally.