Model Performance Metrics for Melanoma Recognition via Dermoscopy

Performance metrics for a deep learning-ensemble model designed to differentiate malignant melanoma from other malignant skin lesions. The model was trained and tested using the ISIC-2024 and HAM10000 dermatoscopic datasets, achieving an AUC of 0.988 on the test set. The dataset was created by Jinyan Jiang and last updated on 2026-05-15.

Use Cases

• Benchmarking melanoma classification models based on reported accuracy, precision, recall, F1 Score, and AUC metrics.
• Training ensemble models using raw scores from multiple convolutional neural networks as feature vectors.
• Evaluating the clinical translation potential of AI models for dermatoscopic differential diagnosis.

Strengths

• Model performance is validated with specific metrics, including an AUC of 0.988 on the test dataset.
• The study employs nine established convolutional neural network architectures, suggesting a rigorous comparison.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• The dataset is very small at 11.7 KB, indicating limited scope, likely containing only summary results.

Provenance

Source

ISIC-2024 and HAM10000 dermatoscopic datasets.

Collection Method

Nine CNN models were used for initial classification; their raw scores were input into an XGBoost ensemble model.

Freshness

Last updated 2026-05-15 04:28:19; freshness should be verified.