Artificial intelligence is now a routine companion in many breast imaging workflows. In mammography, AI tools can help by pointing physicians to areas that require additional review or points of concern. When used well, these systems can support consistency and efficiency. However, if physicians become too reliant on these tools they can serve as a crutch and keep increasing the risk that the physician will spot potential cancerous growths.
A recent study highlights these concerns. Researchers found that radiologists can focus on spots flagged by AI and fail to review the entire image. This can translate to missing another area of the image that may have concerning results. The findings are less about whether AI is “good” or “bad” and more about how human perception and decision-making can change when physicians use AI.
Key takeaways from the study
The study points to three patterns that can inform training, quality assurance and day-to-day reading habits.
- Question “clean” reads. AI may claim an image is “clean” but this should not change the reviewing physician’s approach. The radiologist should still carefully review the image.
- False positives were often caught. Radiologists did increase scrutiny of images labeled with positive flags and were able to reduce the risk of false positives.
- Bias when using AI is a problem. The data shows that physicians are relying on AI in a way that could harm patients.
The researchers warn physicians not to let AI override their search pattern. This is particularly concerning when AI notes a read as “clean” as it can lead to missed cancers. When AI does not flag anything, radiologists can unconsciously reduce visual search effort, treating the absence of a prompt as reassurance.
AI an important tool but physician oversight remains the gold standard
The research discussed in this article reveals a critical insight: AI is most effective when it serves as a supportive tool rather than a replacement for thorough human review. Your radiologist should examine your entire mammogram carefully, regardless of whether AI flags specific areas or labels your scan as “clean.” A comprehensive visual search pattern remains the gold standard to mitigate the risk of a missed cancer diagnosis.