The Lancet Oncology journal recently published a study indicating that implementing artificial intelligence (AI) in mammogram cancer screening may effectively reduce radiologist workload by nearly 50%, without increasing the risk of false-positive results. The findings suggest that AI’s recommendations in such scenarios are equivalent to the collective suggestions of two radiologists.
The research showed that mammography screenings aided by AI had a similar cancer detection rate to traditional double readings, but with significantly less workload for the screen-reading. This suggests that AI utilization in mammogram screenings is a safe practice, according to the study.
The research, led by a team from Lund University in Sweden, involved tracking 80,033 Swedish women (with an average age of 54) over the course of slightly over a year in 2021-2022. Out of 39,996 patients who received AI-supported breast cancer screenings, 28% or 244 tests identified screen-detected cancers. In contrast, of the 40,024 patients who underwent traditional screenings, only 25% or 203 tests detected cancers.
Out of the additional 41 cancers detected through AI screenings, 19 were invasive. Both AI and conventional screenings showed a false positive rate of 1.5%. Remarkably, radiologists using AI had 36,886 fewer screen readings to examine than their traditional counterparts, marking a workload reduction of 44%.
Despite these encouraging preliminary safety results, lead author Dr. Kristina Lång emphasized in a press release that these findings alone do not indicate that AI is ready for full-scale deployment in mammogram screening. Lång stressed the need for further investigation of the technology’s impact on patient outcomes, particularly its potential to identify interval cancers often missed by conventional screenings, as well as its cost-effectiveness.
Cancer detection through AI has been a long-sought objective among computer vision researchers and AI firms. As machine vision systems continuously improve, they have shown in certain instances to be as dependable as human clinicians. Tech giants like IBM, Google, MIT, and NVIDIA have been actively investing in this type of cancer screening research over recent years. Their goal is to develop an infallible tool that can detect cancerous growths at their earliest stages.
Frequently Asked Questions (FAQs) about AI-assisted mammography
Can AI assist in mammogram cancer screening and reduce the workload for radiologists?
Yes, a study published in the Lancet Oncology journal found that AI can be used in mammogram cancer screening to safely reduce the workload of radiologists by nearly half, without increasing the rate of false-positive results.
How effective is AI in detecting cancer compared to traditional screening methods?
The study indicated that AI-supported mammography screenings had a similar cancer detection rate to traditional double readings. Of the patients who received AI-supported screenings, 28% of tests detected cancers, compared to 25% detected through traditional screenings.
Is the use of AI in mammography screening safe?
The study suggests that the use of AI in mammography screening is safe, as it does not increase the risk of false-positive results. However, the lead author emphasized that more research is needed to understand the implications on patients’ outcomes and the cost-effectiveness of the technology.
How does AI impact the workload of radiologists in cancer screenings?
The research showed a significant decrease in the workload of radiologists using AI in mammogram screenings. The study found that there was a 44% reduction in screen readings that radiologists had to examine when using AI.
What is the potential of AI in cancer detection?
The potential of AI in cancer detection is significant. Tech companies such as IBM, Google, MIT, and NVIDIA have been investing in similar cancer screening research. The goal is to develop reliable machine vision systems that can detect cancerous growths at their earliest stages, on par with human clinicians.