Step 1: Image quality control
To ensure reliable assessments, the AI first verifies that the uploaded image meets quality standards. It checks if the lesion is centered and correctly captured without distracting elements.
Step 2: AI pattern recognition
The AI model was trained using a comprehensive image database, curated by medical professionals. By recognizing key patterns in skin lesions, the system can offer a well-founded pre-diagnosis. However, it can only classify lesions that match known patterns and cannot evaluate completely novel abnormalities.
Step 3: Heat maps for enhanced transparency
SkinDoc generates heat maps to highlight the most relevant areas in an image. This helps users understand which features influenced the AI’s decision, increasing trust in the results.
Step 4: Symmetry analysis
Symmetry plays a key role in dermatological evaluations. Many suspicious skin changes exhibit asymmetry, which the AI detects and assesses accordingly.
Step 5: Skin feature segmentation
By precisely segmenting the lesion from the surrounding skin, the AI improves diagnostic accuracy. This step ensures that the analysis focuses only on the relevant skin area.
Step 6: Presentation of results
After analysis, users receive a detailed report, including an AI-generated diagnostic assessment and a transparent explanation of the methods applied. Medical professionals can provide feedback to refine the system further.
