Medical Imaging AI — Diagnostic Support for Clinical Images

I will build a medical imaging AI system for clinical decision support — covering classification or detection on your imaging modality (X-ray, CT, MRI, dermatology, or pathology), model development following clinical AI best practices (prospective validation design, subgroup performance analysis, uncertainty quantification), and a results interface designed for clinical reviewer workflow integration. Medical imaging AI requires higher methodological rigour than general CV applications — subgroup performance analysis ensuring the model doesn't perform significantly worse on specific demographic groups, uncertainty quantification communicating when the model is operating outside its training distribution, and a prospective validation design that correctly estimates real-world performance rather than in-sample metrics.

The model covers imaging modality-specific preprocessing, transfer learning from a medical imaging pre-trained backbone, subgroup performance analysis, uncertainty estimation, clinical workflow integration design, and regulatory consideration documentation.