A new deep learning model combining dermoscopic images with patient metadata achieves 94.5% accuracy in melanoma detection.
Key Details
- 1AI model integrates dermoscopic images and patient data (age, gender, lesion site) for diagnosis.
- 2Achieved 94.5% accuracy and an F1-score of 0.94 using the SIIM-ISIC dataset of over 33,000 images.
- 3Outperformed popular image-only AI models like ResNet-50 and EfficientNet.
- 4Feature analysis showed lesion size, age, and site significantly impact diagnostic accuracy.
- 5Model developed by international team led by Incheon National University; to be published in Information Fusion (Dec 2025).
- 6Potential applications include smartphone-based diagnosis, telemedicine, and AI-assisted dermatology clinics.
Why It Matters
This study demonstrates the value of multimodal AI in diagnostic imaging, highlighting better performance and transparency in early skin cancer detection. Such advances can inform future imaging-AI integration strategies, with implications for clinical workflow and patient care in dermatology and radiology.
Source
EurekAlert
Related News
•EurekAlert
AI-Powered OCT Enables Rapid 'Optical Biopsy' for Early Endometrial Cancer Detection
A team at Washington University has developed a catheter-based 3D OCT system with AI to quickly and noninvasively detect early endometrial cancers.
•EurekAlert
AI Clinical Reasoning in Diagnostics and Digital Fatigue in Healthcare
Recent JMIR features explore large language models in clinical diagnostics and digital fatigue among healthcare professionals.
•EurekAlert
KAIST, MIT, Microsoft Develop Efficient AI Image Upsampling for Robotics
KAIST, MIT, and Microsoft have created 'Upsample Anything,' a training-free AI method to restore high-resolution visual data from compressed images with up to 16x improved GPU memory efficiency.