Neural network cuts false-positive recalls

Published in MPW, 15 Aug 2014

Researchers in the US have developed an image-analysis technique that could cut the number of false-positive results in routine screening mammography. The technique, which uses a neural network to simultaneously analyse four different images, could save on costs associated with recalls, and relieve patients of undue worry (Phys. Med. Biol. 59 4357).

Routine mammography has been shown to significantly reduce mortality associated with breast cancer, which according to the World Health Organization is by far the most common cancer in women worldwide. Unfortunately, searching for true cases of malignant tumours among a majority of healthy women is a difficult task. Doctors naturally err on the side of caution, which means that of the 6 to 12% of women who are recalled, just one-tenth or fewer are truly in need of treatment. According to one study, in 10 years of screening more than half of women will experience a false-positive recall. […]

The rest of this article is available here.