Machine learning could help with early breast cancer detection: Study

About 70 percent of the lesions are benign, 20 percent are malignant, and 10 percent are high-risk lesions. The model built by the team resulted in fewer unnecessary surgeries. And, were also being able to diagnose more cancerous lesions than the strategy of only doing surgery on traditional high-risk lesions. 

A new study reveals that machine learning can help identify high-risk breast lesions that could become cancerous.

A study from Massachusetts General Hospital/Harvard Medical School reveals that such prediction could lead top early detection and avoid the need for unnecessary surgery. It should be noted that biopsy-diagnosed lesions that can be cancerous are removed through biopsy.

Explaining how it works, the report adds that when a mammogram detects a suspicious lesion, and a needle biopsy is performed to check if it is cancerous. Now, about 70 percent of the lesions are benign, 20 percent are malignant, and 10 percent are high-risk lesions, points out MIT News.

In case of high-risk ones, they are managed in a different way. Now, high-risk do not always have cancer. The approach for detecting requires patients to undergo surgery that is painful and time-consuming and of course expensive survey. “The vast majority of patients with high-risk lesions do not have cancer, and we’re trying to find the few that do,” reportedly said Bahl, a fellow doctor at MGH’s Department of Radiology.

The model built by the team resulted in fewer unnecessary surgeries. And, were also being able to diagnose more cancerous lesions than the strategy of only doing surgery on traditional high-risk lesions, the report explains further.

MGH radiologists will start including the model into their clinical practice in the next one year. However, they are also working to further hone the model.