Many Korean Women 'Delicate Breast', Improve Breast Cancer Diagnosis Accuracy with AI and Ultrasound
Aug 08, 2024
|
On the 8th, a research team led by Jang Jung-min and Ha Soo-min of the Department of Radiology at Seoul National University Hospital announced the results of a study comparing the performance of AI programs and breast ultrasound as an auxiliary diagnosis method for breast cancer in dense breast women who underwent mammography and ultrasound examinations between 2017 and 2018.
Mammography is a priority test method for early detection of breast cancer and is selectively recommended every two years for women over the age of 40. However, 'Dense Breast', which is low in fat and high in parenchyma, appears as a high-density white mass in the image, so the test sensitivity is low, making it easy to miss cancer diagnosis.
In particular, since most women in Korea are dense breasts, additional tests such as breast ultrasound are essential for accurate diagnosis. Recently, attempts have been made to increase diagnostic accuracy by introducing AI technology into the breast cancer screening field, but objective evaluation of its performance has been lacking.
The research team compared and analyzed the results of ▲single mammography ▲ breast imaging + AI ▲ breast imaging + ultrasound diagnosis of 5,707 dense breast women aged 40 or older who were screened for breast cancer between 2017 and 2018.
As a result, in the case of mammography + AI, the specificity was higher (95.3% vs. 94.3%) and the reexamination rate was lower (95% vs. 6%) than the single test. In other words, the specificity of diagnosing a normal person as negative was improved, and false-positive cases diagnosed as breast cancer were reduced when the AI-assisted diagnosis program was combined.
On the other hand, the sensitivity of diagnosing breast cancer as positive was higher in mammography + ultrasound than in mammography + AI (97% vs. 60.6%). In addition, the 12 breast cancers that were additionally diagnosed through mammography + AI are early breast cancer without lymph node metastasis, and the research team emphasized the importance of combining breast ultrasound in early diagnosis of breast cancer.
Professor Jang Jung-min said "AI could help radiologists read by improving the specificity of mammography, but could not completely compensate for the low diagnostic sensitivity of dense breast women."If breast ultrasound is applied together according to patient characteristics, we will be able to establish a customized breast cancer screening strategy for early detection of breast cancer."
Meanwhile, the results of this study were published in the latest issue of the international journal `Radiology (IF;12.1)' in the field of radiology.
|
bellho@sportschosun.com
