Deep Learning Drops Error Rate for Breast Cancer Diagnoses by 85%

[HardOCP News]

If there’s one thing doctors and patients want from breast cancer diagnoses, it’s reliability. Keeping up with the massive flow of research data on breast cancer is a challenge for scientists. And the variety of methods used to analyze that data make reliable predictions difficult to come by. A team from Harvard Medical School’s Beth Israel Deaconess Medical Center (BIDMC) tackled this issue using deep learning, in the 2016 Camelyon Grand Challenge. Hosted by the International Symposium on Biomedical Imaging, the competition aims to determine how algorithms can help pathologists better identify cancer in lymph node images. The team’s results were dramatic, dropping the human error rate in diagnosis by 85% when aiding a pathologist’s efforts with GPU-powered deep learning analysis.

 

[B00nie]

As a hobbyist gynecologist I approve this.

 

[westrock2000]

As a hobbyist gynecologist I approve this.

This is great news worth celebrating. You bring the bread, I'll bring the milk.

 

[Shintai]

We need more of this

 

[jwcalla]

This is the sort of thing that I'd be interested in working on (medical software). Hopefully more software jobs open up in the future in this area.

 

[DukenukemX]

We need this to replace those terrible doctors.

 

[Uvaman2]

We need this to replace those terrible doctors.

I know, years and years studying, for nothing soon :/

 

[JDon]

Accurate, early diagnosis is good at improving odds for all sorts of cancers, but I'd greatly appreciate some focus on figuring out how to just get rid of it altogether.... Years of this shiiituff running my family has done a severe amount of damage to my calm about cancer research, and jackass groups who raise money for "awareness" just to pat themselves on the back instead of getting funds where they're really needed.

If we can somehow use computer learning to help speed up cancer research, I'm all for it!

 

[JDon]

Also, this:
xkcd: Lanes

 

[drescherjm]

I have been in this field (medical imaging specifically breast & lung cancer research) my entire career. We actually discussed this topic last week. If I would be involved it most likely would be on the study end to verify the results with radiologists. And possibly develop a training tool.