Aug 28 2020
At Karolinska Institutet and Karolinska University Hospital in Sweden, scientists have compared the potential of three different artificial intelligence (AI) algorithms to determine breast cancer depending on mammograms taken earlier.
Fredrik Strand, MD, PhD, researcher and radiologist, Department of Oncology-Pathology, Karolinska Institutet. Image Credit: Martin Stenmark.
The best algorithm was found to be as precise as the average radiologist. The study results were reported in the JAMA Oncology journal and may guide the way in revolutionizing breast cancer screening in the years to come.
This is the first independent comparison conducted to assess the accuracy of several different AI algorithms. We can demonstrate that one of the three algorithms is significantly better than the others and that it equals the accuracy of the average radiologist.
Fredrik Strand, Study Author and Researcher, Department of Oncology-Pathology, Karolinska Institutet
Strand is also a radiologist at Karolinska University Hospital. A huge number of suppliers have designed several AI-based medical imaging solutions. This study performs a comparison of the potential of three such solutions to detect breast cancer in mammograms taken earlier.
As part of the study, mammograms of 8,805 women aged between 40 and 74 years were analyzed, all of whom were screened for breast cancer between 2008 and 2015. Out of these, 739 members had been diagnosed with breast cancer, either during screening or by the next 12 months. Of the three AI algorithms, the most effective one diagnosed the same percentage of women with cancer as that of the average radiologist.
One AI Algorithm Better than the Others
Moreover, the study findings illustrated that one of the AI algorithms was considerably better compared to the others.
We conducted the study in order to find out how far the algorithms have developed and whether there is any difference between the available systems. The results show that, in principle, the best algorithm is ready for use and that there is a significant difference between the various algorithms on the market.
Fredrik Strand, Study Author and Researcher, Department of Oncology-Pathology, Karolinska Institutet
One more study by the same research team was recently published in The Lancet Digital Health and demonstrates that an AI algorithm has the capability to classify mammography images to denote which would need more attention and which were easily assessed by AI alone without missing any cancer that would have been detected by a radiologist otherwise.
Prospective Clinical Study in the Works
The team is currently keen on analyzing how AI can play a vital role in enhancing breast cancer screening. In the existing practice, two radiologists examine every mammogram. If one of them finds a presumed tumor, the examination advances to a discussion between at least two radiologists who collectively decide whether the woman should be called back for additional examination.
We are planning a prospective clinical study to see how AI works as an independent reviewer of mammograms in a day-to-day clinical environment, both by acting as a third reviewer and by helping to select women who can be offered complementary MRI scans in order to detect cancers at an earlier stage.
Fredrik Strand, Study Author and Researcher, Department of Oncology-Pathology, Karolinska Institutet
The study was carried out at Karolinska Institutet and financially supported by Region Stockholm.
Journal Reference:
Salim, M., et al. (2020) External Evaluation of 3 Commercial Artificial Intelligence Algorithms for Independent Assessment of Screening Mammograms. JAMA Oncology. doi.org/10.1001/jamaoncol.2020.3321.