Technical - Emerging CMR Methods - Acquisition & Reconstruction-What Will It Take For Me To Adopt Your New AI and Machine Learning Method

What Will It Take For Me To Adopt Your New AI and Machine Learning Method

Video Transcription

Video Summary

Tim Leiner argues that AI will affect every stage of medical imaging, from scheduling to prognosis, but adoption depends on rigorous validation. He outlines a framework for machine learning studies: define a clinically relevant problem, justify the model and sample size, clearly separate training, validation, and test data, and use external validation when possible. He stresses proper standards of reference, appropriate performance metrics, explainability, reproducibility, generalizability, clinical applicability, and potential impact on patient outcomes. He also encourages open code and transparent reporting of failures to build trust and improve clinical uptake.

Keywords

medical imaging

artificial intelligence

machine learning validation

clinical applicability

external validation