Enhancing AI-Driven Medical Imaging: Noise Tolerance and Diagnostic Accuracy

Recent advancements in the field of artificial intelligence (AI) have transformed medical image processing tasks, notably in the domains of segmentation and classification. However, the efficacy of AI models in these tasks is intrinsically tied to the availability of substantial amounts of high-quality labelled data. The traditional reliance on human experts for labelling is both costly and labor-intensive, while crowd-sourced annotations often introduce noise and inconsistencies due to variations in annotation policies. These discrepancies can lead to reduced accuracy in AI algorithms, necessitating effective strategies to manage label noise and policy heterogeneity. While noise tolerance remains an underexplored research area, AI diagnostic accuracy has received substantial attention, with studies indicating an accuracy rate generally ranging from 90% to 95%. Despite these positive figures, the question remains: is this level of accuracy sufficient to gain clinicians' trust?This Research Topic aims to explore two pivotal areas: noise tolerance and diagnostic accuracy in AI-driven medical imaging systems. We seek to provide a platform for generating solutions, reevaluating current methodologies, and proposing advancements that enhance data handling and model accuracy. Addressing these issues will further the development of robust AI models capable of operating in complex medical environments.We will accept Original Research, Review, Mini Review, and Perspective articles. To gather further insights in AI-driven medical imaging, we welcome articles addressing, but not limited to, the following themes:• Strategies to manage noise in labelled data sets• Evaluation metrics for AI model accuracy validation• Bias and misrepresentation in AI data sets• Adoption of noise-suppressing neural network technologies• Clinical implications of AI diagnostic systems' accuracy Topic editors: Maria Del Carmen Valdés Hernández, Wenting Duan, Jose Bernal, Roberto Duarte Coello Source: https://www.frontiersin.org/research-topics/73279/enhancing-ai-driven-medical-imaging-noise-tolerance-and-diagnostic-accuracy