Artificial Intelligence Algorithm Matches Human Performance in Mammography Reading

Artificial Intelligence Algorithm Matches Human Performance in Mammography Reading

A recent study published in Radiology found that a commercially available artificial intelligence (AI) algorithm performed comparably to human readers when interpreting mammograms. Mammography is commonly used for breast cancer detection, but it has limitations such as false positives and missed cancers. To address these challenges, two readers often interpret each mammogram, but this is a labor-intensive process. AI has been proposed as a potential solution, but its safety and effectiveness need to be established.

The researchers used the Personal Performance in Mammographic Screening (PERFORMS) assessment, which evaluates the performance of human readers of mammograms, to test an existing AI algorithm. The study involved 552 readers, including radiologists, radiographers, and breast clinicians, as well as the AI, who were asked to read mammograms from two PERFORMS test sets. The test sets included normal, benign, and abnormal findings.

Both the AI and human readers achieved high levels of performance, with similar sensitivity and specificity scores. However, the researchers emphasized the need for further research before AI can be used as a second reader in clinical settings. AI performance may change over time and be influenced by changes in the operating environment, which could impact patient outcomes. Monitoring AI performance and ensuring its safe deployment in clinical practice is crucial.

This study highlights the potential of AI in medical imaging, with radiology being a field of particular interest. Researchers have explored how AI can transform medical imaging, and one such application is the detection of tuberculosis (TB) through chest radiographs. Google Health researchers demonstrated that their deep learning tool achieved comparable performance to human radiologists in detecting TB. This tool could facilitate TB screening in limited-resource areas where there is a lack of expertise in chest radiograph interpretation.

– Radiology (journal)
– United Kingdom’s National Health Service Breast Screening Program (NHSBSP)

All Rights Reserved 2021.
| .