Related Experiment Video
Updated: Jun 13, 2026
05:33
Introduction of an Integrated Pathology Image Management, Artificial Intelligence, and Reporting System
Published on: July 11, 2025
MEL-IA: An Interoperable AI System for Multimodal Skin Lesion Classification in Hospital Settings
Pablo Candela Córcoles1, Alberto De Ramón Fernández2, Marcelo Saval Calvo3
1Bioinspired Engineering and Health Informatics research group (IBIS), University of Alicante, Alicante, Spain.
Journal of Medical Systems
|June 11, 2026
Summary
This study introduces MEL-IA, an AI system for skin cancer detection that integrates with hospital systems. It shows strong performance in classifying skin lesions, improving clinical workflows.
Area of Science:
- Artificial Intelligence in Medicine
- Dermatology
- Medical Informatics
Background:
- Skin cancer is a growing health concern, necessitating advanced diagnostic tools.
- Efficient integration of AI into clinical workflows is crucial for effective lesion assessment.
Purpose of the Study:
- To present MEL-IA (MobilE skin Lesion dIAgnosis), an AI system for automated skin lesion classification.
- To demonstrate the system's interoperability and seamless integration with hospital information systems.
Main Methods:
- A multimodal AI model (EfficientNet-B4) combining images and clinical data was developed.
- Internal validation used 5-fold cross-validation; external validation on HAM10000 dataset.
- Interoperability tested via real-world hospital deployment using HL7, DICOM, PACS, and HIS/RIS standards.
Main Results:
- Internal validation achieved 0.86 accuracy and 0.85 macro F1-score.
- External evaluation showed 0.84 accuracy, with sensitivities of 0.60 for melanoma and 0.72 for basal cell carcinoma.
- Deployment confirmed >99% study integration success and near real-time processing.
Conclusions:
- MEL-IA offers multimodal skin lesion classification and integrates effectively into hospital infrastructures.
- The system demonstrates technical feasibility, interoperability, and operational viability for dermatology workflows.
- Further clinical studies are needed to evaluate its impact on patient outcomes.
