A new technique for interpreting mammograms is a better way to predict future breast cancers, Melbourne researchers have found.

Melbourne researchers have found a new way to interpret mammograms that can more accurately predict which women will develop breast cancer.

Radiologists have long known that areas of breast density, which show up as white on mammograms, are a risk factor for breast cancer and can also hide existing cancers.

PhD student Tuong Linh Nguyen led a research team that focused on the bright areas, not the white areas, in 1300 mammograms of Australian women aged in their 20s to 70s.

The results, published in the International Journal of Epidemiology on Sunday, found that while white areas were important for hiding existing breast cancers, bright spots provided information about future breast cancers.

"These results also apply to young women and women with a family history of breast cancer," Mr Nguyen said.

Professor John Hopper, director of research at the Centre for Epidemiology and Biostatistics, said the technique was 30 per cent more accurate at predicting cancers.

"We've discovered that the best predictor of a woman developing breast cancer in the future is how much of her mammogram is covered by bright areas -- even more than all the known genetic factors discovered over the past 20 years," Prof Hopper said.

He said his colleagues were working on digital automation to make the new technique available for screening services.