10,015 dermatoscopic images of pigmented skin lesions categorized into seven diagnostic classes. The dataset includes patient metadata such as age, sex, and anatomical localization, alongside the specific diagnostic confirmation method for each lesion.
Use Cases
- Train a multi-class image classification model to detect skin cancer using the 'dx' label and image data
- Perform demographic bias audits in medical imaging by evaluating model performance across 'age' and 'sex' columns
- Analyze the prevalence of specific lesion types across different body parts using the 'localization' attribute
Strengths
- 10,015 high-resolution dermatoscopic images of skin lesions
- Seven diagnostic categories including melanoma (mel), basal cell carcinoma (bcc), and melanocytic nevi (nv)
- Metadata columns for 'age', 'sex', 'localization', and 'dx_type' (diagnostic confirmation method)
- Includes 'lesion_id' to identify multiple images belonging to the same lesion