AI Misses Nearly One-Third of Breast Cancers, Study Finds (www.emjreviews.com)

A recent study has raised concerns over the reliability of artificial intelligence (AI) in breast cancer detection, revealing that nearly 30% of cases were missed by AI-based diagnostic tools. Evaluating 414 women with a mean age of 55.3, researchers found that 127 cancers were overlooked, particularly due to dense breast tissue and small tumor sizes (2 cm or smaller). The study underscores a significant limitation of current AI systems in accurately identifying lesions, particularly in challenging imaging scenarios. To address this shortfall, the researchers tested a diffusion-weighted imaging (DWI) technique, which assesses water movement through tissue without the need for contrast agents. Their findings indicated that DWI successfully detected 83.5% of the cancers missed by AI, offering a promising supplemental approach, particularly in cases with dense breast tissue. While these results suggest a potential pathway to enhance AI-assisted breast cancer screening, the authors emphasize the need for multicenter trials to confirm the effectiveness of DWI in larger populations. This research highlights the evolving role of AI in healthcare and the importance of combining technological advancements with traditional methods to improve diagnostic accuracy.