NOMA AI
Welcome
NOMA AI: Where Machine Learning Meets Clinical Medicine Our system doesn't just classify skin conditions—it thinks like a clinician. By fusing computer vision with the ABCDE clinical assessment framework, NOMA AI provides a comprehensive screening experience that mirrors how dermatologists evaluate lesions.
Meet the Developer
A programmer driven by innovation, equity, and the mission to make early disease detection accessible to all.
Brief About📋
Functionalities⚙️
Image Input & Capture:
Users capture high-resolution images of skin lesions using the integrated Arducam camera, with real-time preview displayed on the 5-inch touchscreen. The device automatically preprocesses images to ensure uniformity in size and format for optimal AI analysis.
Feature Extraction & Analysis:
The MobileNetV3-based CNN automatically extracts critical features including asymmetry, border irregularity, color variation, and diameter from lesion images. Clinical feature extraction provides objective measurements that complement the AI's visual analysis.
Multi-Modal Classification:
The system fuses AI visual classification (24 skin conditions) with clinical ABCDE risk assessment and patient risk factors to generate a comprehensive risk score. Top-3 alternative diagnoses with confidence levels are displayed, and an uncertainty estimate flags when results may be unreliable.
Interpretability & Education:
Grad-CAM heatmaps visualize which image regions influenced the AI's decision. Educational tips and a health passport (saving assessment history) promote skin health literacy and longitudinal tracking. LED indicators (red/yellow/green) provide immediate physical risk feedback.
Diagram of various skin cancers inside the different layers of skin
NOMA AI is a portable, AI-powered skin cancer screening system that fuses computer vision with clinical risk assessment to provide accessible, preliminary skin lesion evaluation. Built on a Raspberry Pi 4, it democratizes dermatological screening through an integrated hardware–software solution designed for point-of-care use.
How accurate is the NOMA AI device in detecting melanoma and other skin conditions?
The device achieves high accuracy through multi-modal analysis, combining AI visual classification with clinical ABCDE assessment and patient risk factors. Results include confidence scores (0-100%) and uncertainty estimates to indicate reliability. For optimal performance, ensure lesions are well-lit and clearly centered in the camera frame.
Is the model suitable for all skin types?
The model is trained on a diverse dataset, but its performance may vary across different skin types. It's essential to consider this when interpreting results.
Can this model replace a dermatologist?
No! The model is a diagnostic aid and should not replace professional medical advice. Always consult a healthcare professional for diagnosis and treatment.
Contact Us
Have a question or need more information? Reach out and we'd be happy to assist you.
🩺 24 Skin Conditions | 📊 Clinical Risk Integration | 🔥 Explainable AI with Grad-CAM | 💡 Educational Focus