A new spin on mammograms has managed to take a badly flawed technology... and make it even worse.

Imagine that!

The technology is called computer-aided detection, or CAD, and it's supposed to help radiologists find potential cancers in breast tissue -- which would be great if it actually worked.

In reality, all it seems good for is giving women a good scare. That's because this "advanced technology" is great at detecting "abnormalities" -- but not so great at detecting actual cancer.

If you want proof, take a look at the data from a study published in Journal of the National Cancer Institute. Researchers looked at data on 1.6 million mammograms given over an eight-year span. They found that even though CAD revealed more "abnormalities," the rate of invasive breast cancers that were detected remained the same -- whether CAD was used or not.

Clinics with CAD also had a lower rate of accurate diagnoses of "abnormal" mammograms than clinics without the technology.

And, for the cherry on top, clinics with CAD had a higher rate of false positives -- which meant that more women had to undergo additional tests, when none of it would have been necessary to begin with.

The same team carried out a similar study in 2007 with similar results -- so if anything, CAD is consistent.

Consistently bad.

Despite that lousy track record, CAD is now widely used. If you've had a mammogram recently, chances are your own radiologist used CAD.

The obvious answer is for radiologists to give CAD the boot. But even if they did, there's a bigger picture to keep in mind.

CAD or no CAD, mammograms are dangerously overused and wildly inaccurate -- and they've led to the mass overtreatment of a disease that won't hurt most women.

The radiation used in mammograms alone can cause the very cancers they're supposed to detect. Even mainstream mammogram backers will be forced to admit it if you ask them directly (expect some hems and haws first, though).

Instead of tweaking this failed screening, it's time to scrap it altogether and find one that really works instead.