The use of Artificial Intelligence (AI) to read mammograms could lead to worrying misinterpretations, a new study reveals. Conducted by the Radiological Society of North America (RSNA), the study suggests that AI-supported decision support systems could adversely affect the accuracy of breast cancer detection.
To evaluate the potential consequences of using AI for mammography, 27 radiologists were presented with 50 mammograms. Ten of these had correctly assigned Breast Imaging Reporting and Data System (BI-RADS) categories by AI, and 40 had incorrect categories.
The results displayed a significant decrease in the accuracy of the radiologists’ BI-RADS assessments following an incorrect assignment by the AI. Less experienced radiologists possessed a BI-RADS accuracy of 79.6%, but following an erroneous AI prediction, this number dropped to 19.4%. Similarly, the accuracy of highly experienced radiologists fell from 82.2% to 45.5% after the AI showed inaccurate BI-RADS categorizations.
Thomas Dratsch, M.D., Ph.D., lead author of the study and a radiologist at the University Hospital, Cologne, commented on the research, “We anticipated that inaccurate AI predictions would influence the decisions made by radiologists in our study, particularly those with less experience. Nonetheless, it was surprising to find that even highly experienced radiologists were adversely impacted by the AI system’s judgments, albeit to a lesser extent than their less seasoned counterparts.” He further added, “Our findings underscore the need for implementing appropriate safeguards when incorporating AI into the radiological process to mitigate the negative consequences of automation bias.”
These results emphasize the importance of radiologists being educated about the AI system and of them feeling accountable for their decisions. The team intends to use the likes of eye-tracking technology to better understand the human decision-making process when using AI.
Breast cancer is a type of cancer that starts in the breast tissue, usually in the cells lining the milk ducts (ductal carcinoma) or the lobules that produce milk (lobular carcinoma). It is the most common type of cancer among women, although it can occur in men too. Symptoms of breast cancer can vary and include lumps in the breast, changes in dimensions or size of the breast or changes in skin texture. Risk factors of the disease include age, gender, family history, genetic mutations, exposure to estrogen, personal history of breast conditions, obesity, alcohol consumption, and radiation exposure. Diagnosis is based on a combination of physical examination, mammography, ultrasound, and biopsy. Treatment options vary and depend on the stage and type of breast cancer, along with the overall health of the patient.
The RSNA is a professional membership organization committed to advancing medical and health care through a variety of initiatives, such as its flagship monthly publication, Radiology. It has more than 53,000 members from over 120 countries who are dedicated to improving patient care and providing quality education.
