Researchers Find AI Improves Breast Cancer Detection by 20%

Initial analyses also show AI screening lowers associated radiologist image reading workload by half

Both radiologists and pathologists analyze images to make cancer diagnoses, although one works with radiological images and the other works with tissue biopsies as the source of information. Now, advances in artificial intelligence (AI) for cancer screenings means both radiologists and pathologists may soon be able to detect cancer more accurately and in significantly less time.

Pathologists may find it instructive to learn more about how use of this technology shortened the time for the radiologist to sign out the case without compromising accuracy and quality.

Led by researchers at Lund University in Sweden, the Mammography Screening with Artificial Intelligence (MASAI) trial found that using AI during high-risk breast cancer screenings improved breast cancer detection by 20% without affecting the number of false positives, according to a Lund University news release.

Even better, AI screenings reduced doctors’ workload in interpreting mammography  images by nearly 50%, the news release states. Such an improvement would also be a boon to busy pathology practices were this technology to become available for tissue biopsy screenings as well.

The researchers published their findings in the journal The Lancet Oncology titled, “Artificial Intelligence-Supported Screen Reading versus Standard Double Reading in the Mammography Screening with Artificial Intelligence Trial (MASAI): a Clinical Safety Analysis of a Randomized, Controlled, Non-Inferiority, Single-Blinded Screening Accuracy Study.“

Can AI Save Time and Improve Diagnoses?

One motivation for conducting this study is that Sweden, like other nations, has a shortage of radiologists. Given ongoing advances in machine learning and AI, researchers launched the study to assess the accuracy of AI in diagnosing images, as well as its ability to make radiologists more productive.

The MASAI trial was the first to demonstrate the effectiveness of AI-supported screening, the Lund news release noted.

“We found that using AI results in the detection of 20% (41) more cancers compared with standard screening, without affecting false positives. A false positive in screening occurs when a woman is recalled but cleared of suspicion of cancer after workup,” said breast radiologist Kristina Lång, MD, PhD, clinical researcher and associate professor in diagnostic radiology at Lund University, and consultant at Skåne University Hospital, in the news release.

Not only did the researchers explore the accuracy of AI-supported mammography compared with radiologists’ standard screen reading, they also looked into AI’s effect on radiologists’ screen-reading workload, the Lancet paper states.

Impetus for the research was the shortage of radiologists in Sweden and other countries. A Lancet news release noted that “there is a shortage of breast radiologists in many countries, including a shortfall of around 41 (8%) in the UK in 2020 and about 50 in Sweden, and it takes over a decade to train a radiologist capable of interpreting mammograms.”

That makes it even more challenging for providers to meet European Commission Initiatives on Breast and Colorectal Cancer (ECIBC) recommendations that two radiologists screen a woman’s mammogram, the Lancet news release pointed out.

More Breast Cancer Identified with Lower Radiologist Workload When Using AI Screening

Here are study findings, according to the Lancet paper:

• AI-supported screening resulted in 244 cancers of 861 women recalled.
• Standard screening found 203 screen-detected cancers among 817 women who were recalled.
• The false positive rate of 1.5% was the same in both groups.
• 41 (20%) more cancers were detected in the AI-enabled screening group.
• Screen readings by radiologists in the AI-supported group totaled 46,345, as compared to 83,231 in the standard screening group.
• Workload dropped by 44% for physicians using screen-reading with AI.

“We need to see whether these promising results hold up under other conditions—with other radiologists or other algorithms,” Lang said in the Lund news release.

“The results from our first analysis show that AI-supported screening is safe since the cancer detection rate did not decline despite a reduction in the screen-reading workload,” she added.

Is AI a Threat to Radiologists?

The use of AI in the Swedish study is an early indication that the technology is advancing in ways that may contribute to increased diagnostic accuracy for radiologists. But could AI replace human radiologists’ readings. Not anytime soon.

“These promising interim safety results should be used to inform new trials and program-based evaluations to address the pronounced radiologist shortage in many countries. But they are not enough on their own to confirm that AI is ready to be implemented in mammography screening,” Lång cautioned. “We still need to understand the implications on patients’ outcomes, especially whether combining radiologists’ expertise with AI can help detect interval cancers that are often missed by traditional screening, as well as the cost-effectiveness of the technology.”

In an Advisory Board daily briefing, breast radiologist Laura Heacock, MD, of the NYU Langone Perlmutter Cancer Center said, “If you spend a day with a radiologist, you’ll see that how an AI looks at screening a mammogram is really just a fraction of how radiologists practice medicine, even in breast imaging.

“These tools work best when paired with highly trained radiologists who make the final call on your mammogram. Think of it as a tool like a stethoscope for a cardiologist,” she added.

Whether a simple tool or an industry-changing breakthrough, pathology groups and clinical laboratories that work with oncologists can safely assume that AI advances will lead to more cancer research and diagnostic tools that enable earlier and more accurate diagnoses from tissue biopsies and better guidance on therapies for patients.

—Donna Marie Pocius

SOURCE: Dark Daily