Gastroenterology & Artificial Intelligence: 3rd Annual Global Artificial Intelligence Summit

8- AI and Radiology_Kahn

Pdf Summary

In the article "AI in Radiology: Lessons for GI Endoscopy" by Charles E. Kahn, Jr., MD, MS, the author discusses the lessons that radiology has learned regarding the use of artificial intelligence (AI) and applies them to the field of gastroenterology (GI) endoscopy.<br /><br />The author emphasizes the importance of testing in the development and implementation of AI systems. This includes evaluating the data used, defining variables, determining the quantity of data, and ensuring that the training data aligns with the intended clinical use.<br /><br />Metrics such as the Dice similarity coefficient, Jaccard index, and Hausdorff distance are discussed as ways to measure the performance of AI systems in radiology. The article also highlights the use of calibration curves to assess the accuracy of AI systems in predicting malignancy.<br /><br />The concept of "ground truth" is explored, emphasizing the need for clear and well-defined annotations of data, as well as the qualifications and training of those who annotate the data. The article also discusses inter-rater and intra-rater variability and the process for adjudicating discrepancies.<br /><br />The author includes examples of findings in liver imaging and provides references to resources such as the Radiology Lexicon and Common Data Elements. The importance of setting standards and continually challenging oneself is emphasized throughout the article.<br /><br />In conclusion, the author suggests that the lessons learned from AI in radiology can be applied to GI endoscopy. The article provides insights and recommendations for researchers and practitioners in the field to improve the development and implementation of AI systems in gastroenterology.

Keywords

AI in Radiology

GI Endoscopy

Lessons

Artificial Intelligence

Testing

Data Evaluation

Performance Metrics

Ground Truth

Liver Imaging

Standards