The aim of screening mammography is to detect breast cancer in women as early as possible before signs of the disease become clinically obvious. In a study published in Nature McKinney et al found that AI bested radiologists in detecting breast cancer in screening mammograms.
Mammograms of 25,856 women in the United Kingdom and 3,097 women in the United States were used to train the AI system. AI was then used to identify the presence of breast cancer in mammograms of women who were known to have had either biopsy-proven breast cancer or normal follow-up imaging results at least 365 days later. The study included mammograms; by conventional digital (2D) mammography and tomosynthesis (also known as 3D mammography).
The authors report that the AI system outperformed diagnoses made by the radiologists who initially interpreted the mammograms, and the decisions of 6 expert radiologists who interpreted 500 randomly selected cases.
The study reports an absolute reduction of 5.7% and 1.2% (USA and UK) in false positives and 9.4% and 2.7% in false negatives. In an independent study of six radiologists, the AI system outperformed all of the human readers: the area under the receiver operating characteristic curve (AUC-ROC) for the AI system was greater than the AUC-ROC for the average radiologist by an absolute margin of 11.5%. The authors also performed a simulation in which the AI system participated in the double-reading process that is common in the UK, and found that the AI system maintained non-inferior performance and reduced the workload of the second reader by 88%.
The authors suggest that further assessment of the AI system with clinical trials may lead to improvements in the accuracy and efficiency of breast cancer screening by limiting the high rates of false positives and negatives which are known to take place in the interpretation of mammograms.