AI outperforms doctors in detecting secondary breast cancer, SNUH study finds

gettyimagesbank

A new study by Seoul National University Hospital (SNUH) has found that artificial intelligence (AI) is more accurate than doctors at detecting secondary breast cancer, particularly identifying cases that physicians missed. The findings suggest that AI could significantly improve the diagnosis and prognosis of cancer survivors.

On Friday, SNUH’s Department of Radiology, led by Professors Chang Jung-min and Ha Su-min, announced the results of a comparative study involving 4,189 patients who had undergone a mastectomy. The researchers compared the diagnostic accuracy of AI software with that of experienced radiologists by analyzing mammography images of the patients’ remaining breasts — where the risk of secondary cancer is known to be elevated.

Secondary cancer refers to a new malignancy that appears in a different location after successful treatment of the initial cancer. Breast cancer, the most common cancer among women, often develops in the opposite breast following a mastectomy.

The AI software used in the study is commonly employed in routine breast cancer screening for the general population. However, this is the first time its effectiveness has been evaluated for patients with a history of breast cancer.

The results showed that the secondary cancer occurrence rate was 2.7 percent, while AI detected cancer in 1.74 percent of cases compared to 1.46 percent by doctors. Notably, AI identified 16 out of 50 cases (32 percent) that doctors failed to diagnose. These cases primarily involved early-stage cancers (Stage 1 or 2) or hormone receptor-positive tumors, which often present more subtle signs.

“This study demonstrates that AI software can effectively detect secondary cancer in the remaining breast following a mastectomy,” said Professor Chang. She added, however, that AI and mammography still have limitations. “Some cancers, particularly in dense breast tissue, may not be visible even with imaging. For accurate screening, magnetic resonance imaging (MRI) should also be considered,” she said.

This article from the Hankook Ilbo, the sister publication of The Korea Times, is translated by a generative AI system and edited by The Korea Times.