Study finds AI screening detects more breast cancers
A study published in The Lancet has found that AI-supported breast cancer screening in Sweden is more effective than standard methods. It detects more cancers, reduces workload for radiologists and lowers interval cancer rates, though experts say India needs more research
New Delhi: Use of artificial intelligence in Sweden’s national breast cancer screening programme has revealed that AI-supported mammography screening is more effective than standard mammography, according to full results from a 2023 trial published in The Lancet journal.
AI-supported breast cancer screening identified more women with clinically relevant cancers without a higher rate of false positives, the results show.
Researchers from Lund University and other institutes in Sweden, Norway, Denmark and the Netherlands also found that women who underwent AI-supported screening were less likely to be diagnosed with more aggressive and advanced breast cancer in the following two years.
In August 2023, interim results from the Mammography Screening with Artificial Intelligence (MASAI) showed that use of AI resulted in the detection of 20 per cent more cancers compared with standard screening.
The team also found that screen-reading workload for radiologists reduced by 44 per cent.
The full results now show that AI-supported mammography also reduces cancer diagnoses in the years following a breast cancer screening appointment by 12 per cent, a key test of screening programme effectiveness, the researchers said.
“Our study is the first randomised controlled trial investigating the use of AI in breast cancer screening and the largest to date looking at AI use in cancer screening in general,” lead author Dr Kristina Lang, a breast radiologist and clinical researcher from Lund University, Sweden, said.
She added, “AI-supported screening improves the early detection of clinically relevant breast cancers, which led to fewer aggressive or advanced cancers diagnosed in between screenings.”
Between April 2021 and December 2022, over 1,05,900 women were randomly assigned to either AI-supported mammography screening or standard double reading by radiologists without AI.
The AI system was trained, validated and tested with more than two lakh examinations from multiple institutions across more than ten countries.
During the follow-up period of two years, 1.55 interval cancers per 1,000 women (82/53,043) were detected in the AI-supported mammography group, compared to 1.76 interval cancers per 1,000 women (93/52,872) in the standard double reading group, a 12 per cent reduction in interval cancer diagnosis for the AI arm.
An interval cancer is a malignancy detected between scheduled screening tests, after a prior negative result and before the next routine check-up.
Further, 81 per cent of cancer cases (338/420) in the AI-supported mammography group were detected at screening, compared to 74 per cent of cancer cases (262/355) in the standard reading group, a nine per cent increase.
Rates of false positives were similar for both groups, 1.5 per cent in the AI-assisted mammography reading and 1.4 per cent in the standard reading group.
“Widely rolling out AI-supported mammography in breast cancer screening programmes could help reduce workload pressures among radiologists, as well as helping to detect more cancers at an early stage, including those with aggressive subtypes,” Lang said.
Kalyan Sivasailam, Founder and CEO, 5C Network, a Bengaluru-based AI product creator for radiology and interpreting scans, and not involved with the study, told PTI, “The MASAI study provides strong evidence that AI can improve mammography screening in a well-organised, well-resourced Swedish programme with standardised equipment and experienced radiologists.”
However, translating the study’s findings into the Indian setting would require acknowledging that almost every favourable condition in Sweden is different here, from non-standardised equipment to a shortage of radiologists and a lack of a mature cancer registry system.
“India’s mammography landscape is fundamentally different from Sweden’s. India has fragmented equipment, and there is no organised population-based screening at a national scale. Image quality varies dramatically between a tertiary centre in Mumbai and a district hospital in rural Bihar. An AI system would need validation across the equipment mix actually present in Indian facilities,” he said.
Sivasailam also pointed to the country’s shortage of radiologists, with many district hospitals having none.
Further, a mature cancer registry system such as that in Sweden would be required.
“Interval cancers in the study were identified through linkage with the Regional Cancer Registry using participants’ national personal identification number. India lacks this infrastructure. Without robust follow-up systems, we would not even know if an AI system was missing cancers that later presented as interval cancers. We would only see the detection rate, not the full picture,” Sivasailam said.
He added that deployment of AI might also need to be different.
“Rather than ‘AI supporting double reading’, as in the MASAI study, it might be ‘AI as primary reader with radiologist confirmation of positive cases’. This is a more aggressive deployment than what MASAI studied. Before deployment, we would need prospective studies of AI as a standalone reader in resource-limited settings, not just AI as decision support for experienced radiologists,” Sivasailam said.
The evidence base for AI mammography in Indian settings does not yet exist, he added.