Intelligent Back Body Scanner for Early Detection of Skin Diseases

Use Case:

Early detection of skin conditions—specifically melanoma—through robotic 3D body scanning and image analysis via deep learning.

Outcome:

AI module integrated with TUINA robotic arm to detect dermatological anomalies; Custom CNN developed and trained using ISIC 2019 dataset due to third-party tool unavailability;; Seamless combination of physiotherapeutic care and AI-based dermatological screening; Pilot testing initiated at a Madrid clinic in collaboration with medical professionals; Robust risk mitigation plan for clinical, regulatory, and technical challenges.

Ecosystem Support:

The StairwAI program provided technical and business mentorship, aiding in risk identification, feasibility modeling, and development of a tailored AI adoption strategy.

AI Relevance:

This project showcases how SMEs can harness AI to enhance diagnostic accuracy and patient engagement through: I) integration of AI in hybrid human-robot systems; II) real-time image processing with convolutional neural networks; III) adaptable product design for clinical and commercial scalability; IV) emphasis on ethical and user-centered AI deployment.

Summary:

Canonical Robots, a Spanish SME specializing in robotics, successfully developed an AI-powered extension for its TUINA robotic physiotherapy platform. The initiative focused on enabling early detection of skin conditions, particularly melanoma, during therapeutic sessions. The core innovation involved integrating a convolutional neural network (CNN) trained on the ISIC 2019 dataset into the TUINA system, enabling it to analyze high-resolution 3D back scans for dermatological irregularities without interrupting the patient’s physiotherapy experience. Initially, the project planned to incorporate an existing AI module from the TUINA ecosystem. However, when the selected CNN was found unavailable, the team adapted quickly by developing and training their own model. This pivot maintained technical progress without compromising diagnostic potential. The system captures images via TUINA and processes them externally, generating a diagnostic report for clinician review. The team delivered all planned milestones on time or ahead of schedule, including a comprehensive feasibility plan and a business model for AI adoption. Risks related to clinical effectiveness, regulatory approval, AI accuracy, and hardware constraints were thoroughly addressed with targeted mitigation strategies. Pilot trials were launched in a Madrid clinic, with medical staff evaluating system performance through real patient interactions. Looking ahead, Canonical Robots plans to build its proprietary image database using data acquired via TUINA to further refine the CNN’s performance. Market entry efforts will focus on Spain initially, leveraging direct sales agents, while international expansion will follow a distributor-based model. Active participation in medical trade fairs and the creation of a compelling pitch deck are among the next business development steps. This project exemplifies the practical application of AI in robotic health solutions and lays the foundation for commercial deployment in predictive, preventive, and personalized medicine contexts.