(Health Korea News / Yoo Ji-in) The mammography AI image analysis solution ‘Lunit Insight MMG’ developed by domestic company Lunit was found to significantly reduce the burden on both patients and doctors by predicting breast cancer up to 6 years in advance.
Lunit announced on the 23rd that these research results were published in the American Medical Association’s ‘JAMA Network Open’ and the Radiological Society of North America’s ‘Radiology: Artificial Intelligence’, respectively.
The study, published in JAMA Network Open and led by Solveig Hofvind, PhD, of the Norwegian Institute of Public Health, included 110,000 women aged 50 to 69 who participated in the Norwegian national breast cancer screening program (BreastScreen Norway). Data from 6,495 people was analyzed.
As a result of the research team analyzing mammography images conducted three times at two-year intervals using Lunit Insight MMG, it was found that AI can predict the occurrence of breast cancer up to six years in advance.
AI assigns a score from 0 to 100 to each breast, with a higher score indicating a higher likelihood of breast cancer. The study found that the difference in mean AI scores between the breast where cancer was later discovered and the contralateral breast where cancer was not detected increased over time. This difference increased significantly to an average of 21.3 points in the first examination, 30.7 points in the second examination, and 79.0 points in the third examination. On the other hand, for women who did not develop breast cancer, the difference in scores between both breasts remained low at less than 10 points in all examinations.
The research team classified the top 1% with an AI score of 91.3 or higher as a high-risk group. Among cancer patients discovered, 4.5% had already had a positive AI score (10 or more) 4 to 6 years ago, 8.6% had had a positive AI score 2 to 4 years ago, and 52.9% had had a positive AI score 2 to 4 years ago.
The results of this study show that by quantifying breast cancer risk through AI, it is possible to establish customized prevention and treatment strategies, such as identifying high-risk cancer groups at an early stage and providing them with intensive monitoring or additional tests.
The clinical trial, conducted by Dr. Mohammad Talal Elhakim, MD, PhD at Odense University Hospital in Denmark and published through the journal Radiology: Artificial Intelligence, analyzed 249,402 mammography images. It was done. This study verified the effectiveness of three scenarios of introducing Lunit Insight MMG into the ‘double reading’ system by two radiologists currently recommended in Europe.
As a result of the study, when AI performed readings on behalf of the first doctor, it reduced the amount of readings by doctors by 48.8% while maintaining cancer detection accuracy. When AI replaced the second doctor, the reading volume decreased by 48.7% and the recall rate decreased by 2.2%, but sensitivity decreased slightly (1.5%).
The most effective method is to use AI to pre-classify high- and low-risk patient groups (triage). In this case, the number of reads was reduced by 49.7%, while producing positive results in all sensitivity, positive predictive value (PPV), and negative predictive value (NPV). It seemed.
This study shows that clinical applications of AI can significantly improve the efficiency of the healthcare system. In particular, it was confirmed that the introduction of AI in breast cancer screening can significantly reduce the workload of medical staff while maintaining or improving diagnostic accuracy.
In a phone call with Health Korea News on the 23rd, a Lunit official said, “This study will be an important basis for proving the effectiveness of AI-based medical services in countries suffering from a shortage of medical personnel.”
“Research has once again proven that AI can play an important role in early diagnosis of breast cancer and efficiency of the medical system,” said Seo Beom-seok, CEO of Lunit. “Based on these research results, Lunit will accelerate the actual clinical application of AI technology. “We will strive to ensure that more patients benefit from accurate and rapid diagnosis,” he said.
Copyright © Health Korea News Unauthorized reproduction and redistribution prohibited
Source: www.hkn24.com
