ISIC skin lesion images are organized into separate training and testing folders for machine learning development. The dataset contains dermatoscopic images of skin lesions. The International Skin Imaging Collaboration (ISIC) is the authoritative source for this data.
Use Cases
- Classify skin lesion images into malignant or benign categories using image features.
- Train convolutional neural networks on labeled skin lesion images for automated diagnosis.
- Benchmark image classification algorithms on a standardized medical imaging dataset.
- Perform data augmentation on skin lesion images to improve model generalization.
Strengths
- Images are pre-split into training and testing sets for direct model development.
- Sourced from the ISIC archive, a recognized standard in dermatological imaging.
Limitations
- The specific number of images, class distribution, and image resolution are unknown.
- The temporal range and geographic origin of the images are unspecified.
Provenance
- Source
- International Skin Imaging Collaboration (ISIC)
- Collection Method
- Collection of dermatoscopic images from clinical settings.