Artificial Intelligence in Abdominal Imaging (2024)-M7-CGI07-2024

Video Transcription

Video Summary

In a recent session on AI in gastrointestinal (GI) imaging, Dr. Tessa Cook highlighted the prevalence of bias in AI, emphasizing that it can begin at any phase in the AI model development pipeline and manifest as either statistical or social bias. Statistical bias can lead to outputs that do not accurately represent reality, while social bias results in inequities in healthcare delivery. Dr. Paul Yee discussed the importance of knowing how to implement AI in radiology, emphasizing understanding one's motivations (the "why"), testing AI in one's own practice settings, and monitoring AI performance over time. Dr. Kirti Magudia explored FDA-approved AI products and research in body imaging, particularly in opportunistic screening, which leverages incidental imaging data for identifying health risks. She detailed the potential of AI in analyzing body composition and coronary calcium scoring, demonstrating its implications for predicting health outcomes like cardiovascular risks. Dr. Matt Lee presented opportunistic imaging at the University of Wisconsin, focusing on using CT data for identifying conditions such as osteoporosis and cardiovascular risks, emphasizing AI's role in advancing this field for public and population health. Overall, the session underscored AI's promising role in improving healthcare outcomes through mitigation of bias and adoption in clinical practice.

Keywords

AI in GI imaging

bias in AI

statistical bias

social bias

AI in radiology

opportunistic screening

body composition analysis

coronary calcium scoring

healthcare outcomes