学术报告（吴晓锋 2024.4.27）

摘要：Clinically, doctors often make ambiguous diagnosis on the complex cases due to uncertain understanding of the diseases, which may lead to undesirable medical consequences. The existing machine vision algorithms designed by machine learning and deep learning of medical images are incapable of dealing with the ambiguous diagnosis of lumbar disc degeneration (LDD). In this talk, a label distribution is introduced to quantitatively express the diagnostic ambiguity of LDD, based on which, a new AI-diagnoser that can perceive and predict the diagnostic ambiguity of LDD is proposed and modeled by integrating the cell metabonomics, subjective probability quantification, and label distribution learning. Numerical experiments demonstrate the ability of the proposed diagnoser to automatically diagnose the ambiguous grading of LDD.