Academic develops offline skin cancer detector using Raspberry Pi 3B (85% accuracy)

Heriot-Watt University researcher Tess Watt has developed LesionIQ - a Raspberry Pi 3 Model B-based system that analyses skin lesions with 85% diagnostic accuracy without requiring internet connectivity. The device processes images through onboard machine learning models trained on existing dermatological datasets, aiming to serve remote populations where smartphones and consistent internet access are unavailable. Current UK legislation still requires physician validation of all diagnoses, with clinical trials and ethical approvals through NHS Scotland pending. The main technical limitation remains insufficient diverse training data, particularly for varying skin tones.