AI-powered Cancer Detection Could Reduce Radiologist Workload by 50%
According to a recent study published in the Lancet Oncology journal, the utilization of artificial intelligence (AI) in mammogram cancer screening can significantly reduce the workload of radiologists by almost 50% without compromising the accuracy of results. The study concluded that the AI’s suggestions were comparable to the combined recommendations of two radiologists working in tandem.
“AI-assisted mammography screening resulted in a similar cancer detection rate compared to standard double reading and with a significantly lower number of screens, demonstrating that the use of AI in mammography screening is safe,” the study concluded.
The study was conducted by a research team at Lund University in Sweden, and accordingly followed 80,033 Swedish women (average age 54 years) for just over a year in 2021-2022. Of the 39,996 patients who underwent random breast cancer screenings performed with the help of artificial intelligence, 28 percent, or 244 tests, returned cancers detected in the screening. Of the remaining 40,024 patients who received conventional cancer screenings, only 25 percent, or 203 tests, returned cancers detected in the screening.
Of the 41 additional cancers detected by the AI side, 19 turned out to be invasive. There were 1.5 percent false positives in both AI-equipped and conventional screenings. Most impressively, radiologists on the AI side had to see 36,886 fewer screens than their counterparts, reducing their workload by 44 percent.
“These promising interim safety results should be used to build on new trials and program-based evaluations to address the significant radiology shortage in many countries, but alone are not sufficient to confirm that AI is ready for use in mammography screening.” Lead author Dr. Kristina Lång warned in the release: “We still need to understand the impact on patient outcomes, especially whether combining radiologists’ expertise with artificial intelligence can help detect interval cancers that are often missed in traditional screening, and the cost-effectiveness of the technology.”
Cancer detection has been a goal of computer vision researchers and artificial intelligence companies for years. I mean, who wouldn’t want to be a company that builds a tricorder that unmistakably detects cancerous tumors in their early stages? Machine vision systems designed for these screenings have steadily advanced in recent years, and in some cases have proven to be as reliable as clinicians such as IBM, Google, MIT, and NVIDIA have invested in similar cancer screening research in recent years.
