littlEye AI

littlEye AI grew from a simple idea: the microscopic world gives meaning to the macroscopic one. It is built for forensic pathologists and pathologists in practice, but above all for medical students as an educational companion to histology and pathology.

Teach your own AI model to recognise histology slides — right on your iPhone, iPad or Mac, offline, without a single pixel leaving your device. Overview guides you step by step: import images, assign diagnoses, review with colleagues, train, evaluate and share.

TRAIN YOUR OWN MODELS
• H&E slide dataset of 27 organs and tissues in a clear grid by organ system — from brain, lung and heart to thyroid, breast and colon
• Colleague slide review with an audit trail — the majority of reviews sets the final diagnosis
• On-device Core ML training: a test set drawn from other cases, confusion matrix, per-class sensitivity and precision
• Mark the finding with a crop; magnification and stain recorded for every image
• Learning from corrections: fix a wrong result and add the image to your dataset — the next training round learns from it
• Portable .le3AI models and .le3set slide datasets with metrics and audit trail — encrypted, shared via AirDrop or email
• Cases: slides attached to a case number, a slide mix-up check and a concise summary to the clipboard; complete backup and restore

RESEARCH TOOLS
• Image quantification: colour deconvolution; area, count and distribution of structures with an overlay; Otsu threshold, measurement across fields, stereological point counting, sensitivity analysis
• Desmin IHC: the DAB + H&E image pair, CPA and indices placed against the cohort — without a verdict
• Measurement log with blinding, staining batches, inter-rater agreement (ICC, Bland–Altman), CSV export and a Methods paragraph with a checksum

KNOWLEDGE BASE
• Atlas of over 100 diagnoses across 27 organs with microscopic descriptions, key features and suggested conclusions
• Image evaluation with a preliminary diagnosis and a report draft, confidence always as a percentage
• Education: a "what next after H&E" guide, special stains, IHC markers, cellular elements and pigments
• Every diagnosis, stain, marker and method has linked sources — textbooks, articles via DOI and the WHO classification
• Fully bilingual SK/EN, professional and lay terminology, dark mode and VoiceOver

FREE — AND WHY
littlEye AI is free and fully functional — no subscription, no advertising, no features behind a payment. It is meant above all for students and junior colleagues. The voluntary contribution in the app unlocks nothing.

ABOUT THE PROJECT
littlEye AI is developed in collaboration with specialists from three institutes of the Faculty of Medicine, Comenius University in Bratislava: Histology and Embryology, Pathological Anatomy, and Forensic Medicine and Forensic Toxicology. It builds on the apps 4n6 Tools and Colorime3ka.

WHAT THE APP CONTAINS
The app is a tool — it ships with a single sample model trained on synthetic patterns. It contains no real slides and no ready-made diagnostic models. You create .le3AI models yourself by training on your own images, or exchange them with colleagues as separate files.

PRIVACY
The app does not connect to the network on its own; the only exceptions are the voluntary contribution through the App Store and literature links, which open in the browser. Images, cases and models remain exclusively on your device. The transfer files .le3AI and .le3set are encrypted — only littlEye AI can open them.

littlEye AI is a research and decision-support tool for qualified professionals. It is not a medical device and does not provide diagnoses — the final assessment always belongs to the pathologist or forensic physician.