• The first prospective study to conduct a large-scale clinical trial of 55,579 patient mammograms using Lunit INSIGHT MMG

• Team of Lunit’s AI and one radiologist showed both an increased cancer detection rate and a lower recall rate compared to team of two radiologists

• Dr. Fredrik Strand, Karolinska Institutet: “This prospective study lays the groundwork for the widespread adoption of AI in breast cancer screening”

[SEOUL, South Korea, November 28, 2022] Lunit (KRX:328130.KQ), a global provider of AI-powered cancer solutions, today announced that it unveiled real-world data (RWD) validating the clinical efficacy of AI in actual medical settings for the first time in the field of breast cancer research at this year’s RSNA meeting. The company will be presenting a total of 12 abstracts—8 oral presentations and 4 poster presentations—at the 108th RSNA Annual Meeting to be held from Nov. 27 – Dec. 1

The highlight of Lunit’s program is an oral presentation on a large-scale population-based prospective study of Lunit’s AI solution for mammography as a potential replacement for radiologists. While past retrospective studies have shown promising performance by AI in mammography screening, this clinical trial conducted in a prospective screening setting truly reflects the real-world impact of AI in a breast screening workflow.

A Swedish team of researchers, led by radiologist Dr. Fredrik Strand at the Karolinska Institutet, conducted a prospective interventional study using Lunit INSIGHT MMG; the CE- and FDA-cleared AI solution for mammography analysis is currently in clinical use across numerous medical institutions around the world. Study results were collected over approximately 14 months (April 1, 2021 – June 9, 2022) analyzing 55,579 patient cases of breast cancer screening in actual clinical settings.

Interpreting mammograms is an inherently difficult task. Once suspicious lesions are found through mammograms, women are recalled to the clinic for further tests and biopsies. While recalls are aimed to increase the detection of cancer, only fewer than 1 in 10 women called back are diagnosed with breast cancer. These false-positive mammograms can cause patient anxiety and physical discomfort, as well as cost significant time and money due to extra tests, potentially doing more harm than good. To tackle this problem, most European countries have adopted a system of double reading in which two radiologists decide whether the patient should be recalled or not.

In keeping with this practice, the study included three independent readers: Radiologist 1, Radiologist 2, and Lunit INSIGHT MMG. This study design allows for results to be reviewed according to different reader combinations: two radiologists, Lunit AI + one radiologist, and each reader separately.

▶︎ Comparison of the Cancer Detection Rate (CDR) Between Lunit INSIGHT MMG and Radiologists

Results from the study showed that Lunit INSIGHT MMG, when combined with a single radiologist, can detect more cancers compared to two radiologists. Furthermore, using AI alone showed non-inferior cancer detection rate (CDR) performance compared to the traditional double reading setting: the CDR per 1,000 exams of two radiologists, one radiologist + AI, and AI alone was 4.1, 4.3, and 4.1, respectively.

▶︎ Comparison of the Recall Rate (RR) Between Lunit INSIGHT MMG and Radiologists

Lunit INSIGHT MMG also showed a lower recall rate (RR) in combination with one radiologist as well as alone, compared to the traditional double reading setting. The RR per 1,000 exams of two radiologists, one radiologist + AI, and AI alone was 29.3, 28.0, and 15.5, respectively. This result indicates that AI use for breast screening analysis, in combination with a radiologist as well as alone, leads to better performance for RR.

“While double readings have been established as the common practice across Europe and Australia, many countries are experiencing great difficulties due to the shortage of radiologists,” explained Dr. Strand, breast radiologist and associate professor at Karolinska Institutet. “This prospective study lays the groundwork for the widespread adoption of AI in breast cancer screening by filling the role of one radiologist, which in turn can reduce medical costs and lead to healthcare reimbursement.”

“AI will eventually become the standard of care for breast cancer screening. While developing a high-performing product has always been critical, this study is key in that Lunit is the first to prove the effectiveness of AI through a large-scale prospective clinical trial in actual medical settings,” remarked Lunit CEO Brandon Suh. “With our recent contract to supply our AI for the BreastScreen NSW State program in Australia, Lunit’s AI is continuing to prove its positive impact in the medical field.”